Autonomous Finance Is Starting to Feel Real

I’ve written quite a lot about the announcements from Oracle AI World already. About the move from AI helping people work to AI increasingly getting the work done. About the importance of data and context. And about some of the foundations Oracle is putting underneath all of this. So I’m not going to cover all of that again. Instead, I want to focus on something that became much clearer when looking specifically at the ERP announcements.

Autonomous finance is starting to feel real. We’ve talked about autonomous finance for years, but I’m not sure we’ve always had a particularly clear idea of what it would actually look like in practice. We’ve had automation, machine learning, intelligent document recognition and, more recently, AI agents. All useful, but generally focused on making an individual activity quicker or easier.

What Oracle showed at AI World feels like the next step. It’s less about giving finance teams another tool to complete a task and more about applications continuously looking at what is happening, identifying what needs attention and working towards an outcome. That might sound like a subtle difference. I don’t think it is.

Oracle talked about the progression from embedded AI through AI agents, teams of agents, workflow agents and now Agentic Applications. I’ve covered that wider evolution before in a previous blog, so I won’t labour the point here.  What interests me now is what happens when you apply that thinking to an entire area of finance.

Take period close. A traditional system helps people complete close activities. Automation might remove some of the repetitive steps. An AI agent might investigate something, explain a variance or recommend an action. But what happens when the application starts continuously looking at the close itself? Identifying issues, analysing exceptions, understanding why something has happened and determining what needs to happen next? That starts to look rather different.

The Ledger Agentic Application gives us quite a good idea of where this could be heading. Rather than somebody finding a problem and then asking AI to investigate it, the application can bring together information around the close and identify areas that need attention.

Variance analysis is a good example. Today, someone may need to identify the variance, gather the information behind it, investigate the transactions and work out what has happened. Increasingly, AI can do much more of that investigation before the finance professional becomes involved. And I think that is where the real value starts to emerge. It isn’t necessarily about taking people out of the process. It’s about deciding at which point they need to come into it.

Do I really need someone spending time finding the exception, gathering the information and working out the likely cause? Or do I want them looking at the exception when judgement or a decision is actually required? That feels like a much more useful conversation about AI in finance.

The same thinking is appearing across Payables, Payments, Billing, Cash Application and Collections, with a number of these Agentic Applications on Oracle’s roadmap for 27A. Invoice Operations, for example, is about much more than reading an invoice. The direction is towards continuously ingesting invoices, enriching them, applying policy, routing where necessary and prioritising the exceptions that genuinely need attention. Cash Application follows a similar pattern, with straightforward receipts being handled automatically while more complex cases can be investigated and escalated. Collections can increasingly use payment activity and customer context to work out where attention is needed and what action should happen next.

None of those things sounds particularly revolutionary when you look at them individually. Put them together, though, and you start to see the shape of something different. Instead of finance teams working through queues of activity and asking the system for help when required, the application increasingly monitors what is happening and brings people in when they are needed. That, to me, is a much more useful definition of autonomous finance.

Of course, autonomy needs boundaries. This is also where I think things get more complicated. Because the obvious response to autonomous finance is: how autonomous do we actually want it to be? I don’t think many CFOs are asking for AI to have complete freedom to make financial decisions. What organisations really need is the ability to decide what AI can do, under what conditions, within which limits, and when a person needs to approve something. That makes policies, security, access, approval thresholds, auditability and data quality much more important.

Oracle confirmed that access to the Agentic Applications is controlled through the relevant duty roles, so organisations can restrict which users see and use those capabilities. It sounds like a small point, but I think it illustrates something much bigger. AI governance can’t really remain a separate conversation from application governance. If agents are increasingly becoming part of how Fusion processes operate, then governing AI becomes part of how we design roles, security, controls and processes in Fusion.

There is also the question of cost. This is another area I think customers need to start understanding. Agentic Applications bring an additional consideration because Oracle confirmed that a separate licence is required before they are published into production. They can be built and tested in development and staging using existing AI Units, with the platform fee triggered when an Agentic Application is published to production.

That means the business case for AI is going to need to mature too. The question can’t just be, “How much time does this save?” It also needs to be: what work is the application now doing, what does that consumption cost, what capacity does it release and, ultimately, what is the value of the outcome? That feels like a far more meaningful measure than simply counting the number of agents an organisation has enabled.

Fusion Companion adds another interesting layer to all of this. Where Agentic Applications potentially do more of the work in the background, Companion could change how users interact with Fusion in the foreground. Companion uses plugins and skills to access Fusion Applications data rather than requiring users to separately run something such as a BI Publisher report first. Oracle also confirmed that Companion does not have a separate subscription, although its AI usage is counted.

I think this could prove really important. Most users don’t want to know which agent they need, which report to run or which technical service sits underneath something. They simply want to get something done. The clever bit increasingly happens underneath.

For me, this is where the ERP announcements from Oracle AI World become much more interesting than another list of new AI features. Finance leaders need to start thinking about where people genuinely add value today. Where are people spending time finding exceptions rather than resolving them? Where are they gathering information rather than making decisions? Which activities could safely happen without somebody being involved every time? And where is human judgement absolutely necessary? Those aren’t really technology questions. They’re operating model questions. And perhaps that’s why autonomous finance suddenly feels a little less theoretical.

We’re beginning to see what it could actually look like: a close that identifies where attention is needed, invoices that move through processing unless there is a reason for someone to intervene, cash being applied automatically where the answer is clear, and collections activity being prioritised according to what is actually happening.

We aren’t at the point where finance suddenly becomes autonomous overnight. Some of Oracle’s capabilities are available now, others remain on the roadmap, and organisations will still have plenty to work through around security, governance, process design, adoption and commercial models. But I think the direction is becoming clearer.

Autonomous finance isn’t really about removing people from finance. It’s about being much more deliberate about where people add value. And if AI can take care of more of the finding, gathering, checking and investigating, while people spend more time deciding, challenging and improving, that feels like a rather more interesting future for finance than simply doing today’s processes a little bit faster.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Oracle AI Success Navigator 26.2: The Shift from Guidance to Action

I’ve written about Oracle AI Success Navigator a few times now, and I’ve always liked the principle behind it. But with 26.2, I think we’re starting to see something more significant happening. Success Navigator is moving beyond simply helping customers understand their Fusion estate and what Oracle recommends they consider next. Increasingly, it is starting to help connect those insights to decisions, roadmaps, implementation and adoption.

And that matters because most Oracle Fusion customers don’t have a shortage of information. In fact, I’d argue that the problem is often the opposite. There are quarterly releases, new AI capabilities, Redwood changes, business process improvements, configuration recommendations and an ever-growing list of things that you could be doing.

The harder questions are much more practical. Which of those things actually matter to us? What should we prioritise? And how do we turn that into something we can realistically deliver? That, for me, is where Oracle AI Success Navigator 26.2 starts to get really interesting.

One of the changes that immediately caught my attention is Innovation Opportunities. Rather than expecting customers to work through everything Oracle has released and decide what might be relevant, Success Navigator will use information from the customer’s Account Profile to surface relevant Fusion capabilities and suggested actions. Customers can review those opportunities and, where appropriate, add them directly into their Adoption Roadmap. It sounds like a relatively simple change, but I think it addresses a very real problem.

I speak to customers regularly who know there is functionality within Fusion they aren’t using. Sometimes they aren’t sure what they are entitled to. Sometimes they know about a capability but haven’t had the time to investigate whether it is relevant. And increasingly, particularly with AI, the sheer pace of new functionality makes keeping up difficult.

Success Navigator starts to change the question from: “What has Oracle released?” to: “What could actually make a difference to our organisation?” For me, that is a much more useful conversation. There is an important caveat though. A recommendation isn’t a business case. Just because something is technically available doesn’t mean it should automatically go onto the implementation backlog. You still need to consider process maturity, data, security, integrations, change capacity and, most importantly, the business outcome you are trying to achieve. I see Innovation Opportunities as providing a much better starting point for those conversations, not replacing them.

Then we get to Agentic Implementation. This is probably the area that gives us the clearest indication of where Oracle is heading, although it is important to separate what is available in 26.2 from what has been announced for the future. At AI World in Las Vegas, Oracle introduced its plans for specialised implementation agents covering Discovery, Design, Configuration, Testing, Integration and Innovation. Central to that vision is the Oracle Fusion Setup and Configuration Registry, or FSCR, which Oracle describes as an authoritative, machine-readable source for Fusion setup, configuration and product-defined best practice. FSCR isn’t really a feature of this AI Success Navigator release yet. It is better thought of as part of the emerging architecture behind where Oracle wants Agentic Implementation to go. But for me, it is still one of the more interesting parts of the overall direction.

For years, implementations have relied on consultants taking product knowledge, documentation, previous experience and customer requirements and translating all of that into designs, configuration workbooks, setup activities and test scenarios. The promise of FSCR is that more of that underlying Oracle product knowledge could become machine-readable. Configuration options, dependencies, sequencing, execution, testing and validation could increasingly become reusable AI skills that implementation agents can work with.

Importantly, Oracle’s model still includes expert review, with AI-generated configurations able to be reviewed, adjusted and approved before deployment. So I don’t see this as “AI replaces the implementation team”. I think that is far too simplistic.

Instead, I see the potential for implementation teams to spend less time on repeatable mechanics and more time on the things that genuinely require experience and judgement: understanding the organisation, challenging processes, dealing with exceptions, designing controls, considering security, managing integrations, assuring what AI has produced and making sure the solution actually delivers the intended business outcome. That is potentially quite a fundamental shift.

Oracle has talked about some significant improvements in design and configuration effort from its early work in this area. I’d still be cautious about turning those into generic productivity claims, particularly while FSCR and the associated implementation capabilities are still emerging. For now, I see this less as a headline feature of 26.2 and more as a signpost towards what AI Success Navigator could become.

Another change that may receive less attention than the agents, but could be extremely useful in practice, is the evolution of Adoption Roadmaps, Release Readiness and the Update Planning Center. Success Navigator already supports quarterly Fusion release planning. With 26.2, Oracle is also introducing monthly maintenance guidance alongside improvements intended to make it easier to explore features, prioritise innovations and track progress. For me, this brings us closer to continuous innovation.

A quarterly release really shouldn’t be a frantic exercise every three months where teams download a spreadsheet, identify the mandatory features, send dozens of emails, decide what they can realistically test and then repeat everything the following quarter. There should be a continuous process for understanding change, assessing impact, deciding what to adopt, planning testing and change activities, recording decisions and then coming back afterwards to see whether the expected value was actually achieved. Bringing more of that into the same environment could help move release management away from being a periodic technical exercise and towards becoming part of the organisation’s ongoing Fusion operating model.

The Configuration Assessment updates also deserve some attention. The existing capability already helps customers look at the relationship between Fusion entitlements, configuration and usage, including identifying processes that are configured and used, configured but not used, or neither configured nor used. Oracle has confirmed Configuration Assessment updates for 26.2, although the detailed enhancements weren’t fully documented in the material I reviewed.

I think the opportunity here is bigger than simply finding unused configuration. If something has been configured but isn’t being used, why isn’t it being used? Was the functionality never properly embedded? Is the process wrong? Has training been missed? Is security preventing people from accessing it? Is an integration getting in the way? Or was it configured years ago and simply isn’t relevant anymore? That creates a very different optimisation conversation.

Instead of starting with a fairly generic question like “How can we get more from Fusion?”, you can begin with evidence about the customer’s existing environment and work out where there is genuinely an opportunity to improve. The technology provides the starting point. The important part is still interpreting what that evidence means for the business.

Oracle is also redesigning the Starter Configuration experience and refreshing its Business Processes content in 26.2. The detailed changes haven’t yet been fully published, but I think the principle behind Starter Configuration remains really useful. Let people see something tangible earlier.

Business users don’t necessarily want to spend weeks discussing an abstract future-state process captured in documents and diagrams. Being able to look at a working Fusion environment, walk through personas and processes and then have a fit-to-standard conversation can make design much more meaningful.

The interesting part for me is what happens when you connect that experience to Agentic Implementation. Discovery, process design, configuration and testing potentially start becoming much more connected rather than being separate implementation phases with documents passed between them. That connection between guidance and execution is where the direction of Success Navigator becomes much clearer.

Finally, Oracle is introducing new Adoption Centers with resources focused on product adoption, achieving business outcomes and working effectively with Support. And this brings me back to something I say quite often about AI and transformation. Technology availability does not equal adoption. You can recommend the right feature, configure it perfectly and release it successfully. That doesn’t mean anybody is going to use it properly.

People still need to understand why something is changing. Processes need to change around the technology. Managers need to reinforce new behaviours. Training and communications need to be relevant to the people actually doing the job. And somebody needs to come back afterwards and ask: Did this actually deliver what we expected? I actually think AI makes that part of transformation more important, not less. As more of the technical implementation activity becomes automated or agent-assisted, the proportion of effort needed around judgement, assurance, people and organisational change potentially becomes greater.

When I step back from the individual 26.2 enhancements, this is the real story for me. Success Navigator is beginning to form a much more connected lifecycle: Understand where you are. Identify opportunities. Decide what matters. Plan the change. Implement it. Adopt it. Measure it. Then optimise again. That is much closer to how customers actually need to manage a continuously evolving SaaS platform.

It is also why I’m particularly interested in how Innovation Opportunities, FSCR and Agentic Implementation develop. If Oracle can provide more product intelligence and repeatable execution through the platform, it changes where customers and implementation partners should spend their time. Less effort interpreting documentation and recreating standard configuration knowledge. More effort understanding the business, designing the right transformation, managing risk, integrating the wider technology landscape and making sure new capability translates into measurable value.

For me, that is where AI in enterprise applications becomes much more interesting. Not simply: “What can AI do?” But: “What can we now spend our time doing because AI can take care of more of the repeatable work?” Oracle AI Success Navigator 26.2 feels like another step towards answering that question. And I suspect the shift from guidance to action is going to become much more important than any individual feature in this release.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Fusion Claw Meets SCM: When Agentic AI Starts Doing the Work

A lot has been said about AI agents in enterprise applications. They can answer questions, summarise information, recommend an action and increasingly help us navigate complicated business processes.

But I think the much more interesting question is: what happens when we want AI to go beyond helping someone do the work and actually coordinate the work itself? That is where Oracle’s new Fusion Claw starts to get interesting, particularly when you look at what Oracle is planning for Supply Chain Management.

I recently attended an Oracle session on Fusion Claw for Product Readiness and Risk Management, and for me it helped put some practical substance behind the announcements from AI World. This isn’t simply another AI agent. It is an attempt to tackle something much harder: how do you allow agentic AI to execute complex enterprise processes without losing the controls, policies, permissions and accountability that organisations already rely upon?

Oracle describes the change as moving from systems that predominantly record information and automate tasks towards systems that can monitor, plan, analyse, coordinate, decide and act. I actually think a better way of looking at it is this: Fusion Applications remain the system of record. Agentic Applications increasingly become the system of outcomes.

Fusion still contains the authoritative enterprise data, business objects, policies, approvals, permissions, processes, context and history. Agentic Applications sit above that foundation and coordinate work towards an outcome. That distinction matters.

One of my concerns whenever we talk about autonomous AI in enterprise applications is that we can make it sound as though we are simply handing an LLM access to ERP or SCM and telling it to get on with things. That clearly isn’t going to work.

Enterprise processes have approvals, segregation of duties, business rules, tolerances and exceptions for a reason. An AI model being confident about an answer doesn’t give it the authority to perform a transaction. And that is really the problem Fusion Claw appears designed to solve.

Fusion Claw is a governed agentic execution runtime and an execution engine for complex, high-value work. The principle is that AI handles the parts of a problem where reasoning is valuable, while the execution layer handles the things that need to be precise and governed.

In practice, there are three important elements working together. Planning uses AI reasoning to understand an objective, analyse the situation, evaluate alternatives and create a plan. Computation handles things such as reconciliation, simulation, optimisation, calculations and business rules deterministically and at scale. Execution then performs authorised actions and transactions within the controls of the enterprise application.

And there is one line in Oracle’s material that I think is probably more important than anything else: “The model can plan and recommend but never grants itself authority to act.” That, for me, is the difference between an interesting AI demonstration and something organisations could realistically contemplate putting into production.

Supply Chain is almost the perfect environment in which to demonstrate this. Think about a relatively straightforward question such as: Is this product actually ready to go into production?

The answer might depend upon information spread across Product Lifecycle Management and the BOM, suppliers and sourcing, regulatory declarations, quality records and financial information including pricing, tariffs and currency exposure.

Traditionally, people have to assemble that evidence, reconcile what it means and then coordinate whatever happens next. This isn’t necessarily because the underlying systems cannot store the information. The challenge is bringing the information together in context, understanding its significance and deciding what should happen next. That is where the Product Readiness use case starts to show the potential of an Agentic Application.

To resolve this, Oracle have introduced the BOM Design Risk and Grading Analyser, part of the Product Readiness Workspace. Its objective is to identify BOM risk earlier and establish a mitigation path before design freeze. The Analyzer can grade design risk, explain blockers, compare mitigation scenarios and prepare a governed readiness decision, with the intended outcome of getting to launch decisions faster while reducing late blockers. What I like about this example is that it isn’t really an AI chatbot use case.

What I like about this example is that it isn’t really an AI chatbot use case. The process connects the business and product records sitting within Fusion Applications with agent capabilities and tools from AI Agent Studio, the Product Readiness Agentic Application and, importantly, the people, policies, objectives and approval authorities that govern the business. Fusion Claw then provides the governed execution layer that brings those elements together.

That combination matters because an agent isn’t simply being asked“what do you think?” It is being asked to investigate a business outcome against actual enterprise evidence and policy.

Oracle breaks this down into three broad stages, but I think they are better understood as a continuous process rather than three separate capabilities. It starts by investigating. AI can understand the required BOM outcome, trace the available evidence, compare the available options and help plan recovery activity. The process then moves into assessment, where defined policy becomes critical. Evidence scope can be validated, scores and grades calculated and mandatory gates applied. Finally, it can follow through, comparing possible recovery options, preparing permitted work and tracking the evidence and resolution. Throughout that process, authority still governs what the system is actually allowed to do.

I think that’s an important pattern for enterprise AI because it acknowledges something we occasionally lose sight of when talking about generative AI: not everything should be probabilistic.

We might be perfectly comfortable asking an AI model to reason about possible ways of solving a problem. We are probably much less comfortable asking that same model to arbitrarily decide whether a mandatory compliance condition has been met. Fusion Claw appears to separate those responsibilities.

The Product Readiness component assesses risk across five dimensions, but what makes this interesting is how those dimensions come together rather than the fact that there are five of them. Lifecycle considers whether an item remains viable for the required horizon, taking into account things such as market status, end-of-life information, support and successor qualification. Sourcing looks at whether supply can actually meet the product’s requirements through areas such as qualified sources, supplier concentration, coverage and lead time. Compliance examines whether regulatory evidence genuinely covers the relevant part and revision, while Quality looks at unresolved conditions, qualification tests and gates. Cost then adds another perspective by considering exposure from price variation, tariffs and currency movement.

None of those questions is particularly unusual in its own right. Organisations already answer them today. What’s different is the possibility of evaluating them together as part of one continuously orchestrated business outcome, rather than somebody manually moving between different records, spreadsheets, reports, emails and application screens.

And this is where the value starts to become much clearer to me. The really clever bit: policy still belongs to you. Another element I particularly like is that Oracle isn’t suggesting that AI should decide what constitutes an acceptable risk. The customer defines that.

An organisation can configure its SOPs, evidence requirements, qualification rules, business rules, risk priorities and tolerances. Policy assessment can then combine evidence and scope with thresholds, weights, profiles, grade bands and mandatory gates to apply those criteria consistently. Crucially, Oracle also shows the policy version being retained so that the result remains traceable. That’s a subtle but really important distinction.

AI can help investigate the situation and develop a plan. The enterprise remains responsible for determining the rules of the game. And for organisations considering how they govern agentic AI, I think that model is more compelling than the idea of simply giving increasingly capable models more access.

That’s a completely different level of AI. And this is where it becomes agentic. Identifying the problem is useful. Resolving it is where the business value lives. We’re talking about a system that can help coordinate an outcome.

And importantly, that does not mean uncontrolled autonomy Depending on the authority provided, Fusion Claw can execute within that authority, recognise when approval is required or escalate when necessary, while maintaining audit and monitoring around the process.

The SCM roadmap shown during the webinar covered Agentic Apps and workspaces across a significant proportion of supply chain. Alongside Product Readiness and Design-to-Source there are planned use cases around Maintenance Operations, Batch Process Manufacturing, Logistics Execution, Sourcing, Shipment Load Planning, Supply Planning Allocation, Procurement Spend, Warehouse Operations, Sales Orders, Cost Accounting Close, Inventory Planning, Supplier Qualification, Production Readiness, Kanban Administration, Business Growth Planning, Product Genealogy, Quality Assurance, Cost Management, Warehouse Execution and Pricing.

That’s quite a list. And it starts to reveal the bigger strategy. Fusion Claw isn’t the Agentic Application. It is the common governed execution runtime underneath them. That distinction is going to be important as we start to see these capabilities appearing across Fusion.

For the last couple of years, enterprise AI has largely been about making existing work easier: summarise this, generate that, answer my question or recommend what I should do next.

Useful? Absolutely. Transformational? Sometimes. But the Fusion Claw and Agentic Applications story points towards something different. The shift is from AI helping a user perform a process towards AI helping orchestrate the process around an outcome, while humans retain control of the objectives, policies and authority. For SCM in particular, that’s potentially significant.

Supply chains involve huge datasets, constantly changing conditions and numerous interconnected policies and controls. A decision in one area can quickly create consequences somewhere else. Product readiness is a great example because lifecycle, sourcing, compliance, quality and cost risk don’t exist independently.

If Agentic Applications can genuinely reason across those dimensions, apply deterministic policies where required, coordinate the resulting work and maintain the underlying governance and audit trail, we’re getting much closer to what enterprise agentic AI needs to look like.

Not autonomous AI doing whatever it thinks is best. Governed autonomy working towards an outcome the business has defined. For me, that distinction is what makes Fusion Claw one of the more interesting announcements to come out of Oracle AI World this year. And if Product Readiness is an indication of what Oracle intends to do across the rest of SCM, I’ll definitely be watching what comes next.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Fusion Claw: Stop Looking for AI Use Cases. Start Looking for Work.

One of the questions I hear most when talking about AI in Oracle Fusion is: “What are the use cases?” It’s a perfectly reasonable question. But I’m increasingly convinced it’s also slightly the wrong one. We have spent the last couple of years looking for places where we can add AI to a business process. Where can we summarise something? Where can we generate something? Where can we make an interaction more conversational?

With Fusion Claw, I think we need to turn that thinking around. Don’t start by looking for an AI use case. Start by looking for work. Specifically, look for the work that people are still stitching together manually across Oracle Fusion. The things that involve checking multiple sources, applying judgement, producing an output, chasing exceptions and deciding what needs to happen next. That distinction matters because Fusion Claw isn’t designed simply to answer another question or perform another transaction.

Oracle describes a Fusion Claw use case as“a business result that someone owns and wants handed over.”  I think that is probably one of the most useful ways of thinking about agentic AI that I’ve seen.

Imagine someone in Finance asks: “What is the balance for account 3210?” That isn’t really a Fusion Claw use case. There is a question. Oracle already knows the answer. We just need the quickest and most appropriate way of retrieving it. Now compare that with: “Get us ready for month-end close and show me anything that could prevent us closing.” That is very different.

To achieve that outcome, somebody might need to check the ledger and subledgers, compare information against close policy, identify variances, investigate exceptions, work out which ones actually matter, escalate anything unresolved and produce evidence explaining what they found. There isn’t one answer. There is work to be done. And that, to me, is where Fusion Claw starts to get interesting.

Oracle’s guidance suggests a useful rule of thumb: if a knowledgeable person would normally spend 30 minutes to several hours researching, comparing information, producing something or chasing follow-ups across systems, it may be worth considering Fusion Claw. That immediately gives us a much more practical way of finding opportunities.

Forget brainstorming futuristic AI scenarios for a moment. Go and talk to the people actually using Fusion and listen for sentences like: “Every Friday I pull together…”; “I download this into Excel and then…”; “It takes me half a day to…” etc. Oracle highlights very similar phrases in its guidance because they often reveal processes involving research, judgement, multiple steps and exceptions. Those are potential Claw conversations. And they are much more interesting than asking a room full of people,“What could we do with AI?”

So what might those use cases actually look like? Once you start looking at Fusion through that lens, there are opportunities everywhere.

Finance: “Tell me if we’re actually ready to close.” Don’t just show the close tasks. Review the position across the ledger and relevant subledgers, check it against policy, identify unusual balances or incomplete activity, investigate exceptions and give the Finance team an evidence-backed view of what’s stopping them being close-ready. The outcome isn’t another dashboard. It is confidence about whether you can close.

Procurement: “Find the things we should be worried about.” Imagine regularly reviewing purchase orders, suppliers, invoices and relevant policies to identify unusual activity. Which transactions deserve attention? Which exceptions can be explained by the available information? Which need a human to investigate? And importantly, can you provide the evidence alongside the recommendation? That is considerably more useful than simply producing a list of transactions which meet a predefined query.

HR: “What needs my attention today?” A manager doesn’t necessarily want to navigate five different parts of HCM to understand what is happening across their team. They want to know: What do I actually need to deal with? That could mean looking across outstanding approvals, absence, recruitment activity, upcoming talent actions and other relevant signals, then identifying what matters and why. The deliverable isn’t the individual pieces of data. It’s the prioritised brief.

Projects: “Tell me where this project is going off track.” Imagine analysing project costs, commitments, milestones and relevant operational information, then investigating significant exceptions rather than simply flagging every variance. The value isn’t telling a project manager something is red. It’s doing some of the work required to understand why it is red.

Oracle identifies seven characteristics that indicate a job may suit Fusion Claw:

  1. Multi-step: the work needs to be planned and carried out in stages.
  2. Long-running: it can take minutes or hours rather than requiring an instant response.
  3. Research-heavy: somebody needs to examine multiple records, documents or policies.
  4. Cross-functional: the outcome spans several objects, agents or business areas.
  5. Exception-heavy: investigating what went wrong is part of the job.
  6. Produces something: the outcome might be a report, spreadsheet, plan or evidence pack.
  7. Governed action: the process may ultimately make changes, but only through controlled and approved capabilities.

Not every use case needs all seven. But if you’re ticking several of them, it’s probably worth investigating. I would add an eighth question of my own: Would somebody be genuinely pleased not to have to do this anymore? Because that’s often where the real value is hiding.

Not everything needs Claw. This may actually be the most important part. One of the risks whenever a new technology arrives is that we immediately try to apply it to everything. That would be a mistake.

If somebody simply needs to ask “How many open requisitions do I have?”, a conversational experience such as Ask Oracle may be more appropriate. If the requirement is “Route every new requisition through the same approval steps”, that’s a workflow. If someone has already made a decision and simply needs to execute a transaction, the existing Fusion action may already be the right answer.

Oracle explicitly makes this distinction: lookups, quick status questions, fixed workflows and straightforward transactions don’t suddenly become better because we put a more autonomous agent in front of them. Use the simplest technology that solves the problem. Agentic doesn’t automatically mean better.

There’s another idea in Oracle’s guidance that I particularly like. Good Fusion Claw opportunities may actually be two use cases rather than one. The first researches. The second acts. For example: Research: Review the available information, investigate the exceptions, apply the appropriate policy and recommend what should happen. Then Action: Once that recommendation has been reviewed, carry out the approved next step.

Oracle describes this as starting in Research mode, reviewing the evidence and recommendations, and gradually handing over actions as confidence develops. For enterprise AI, that feels like a much more sensible adoption model than jumping straight to full autonomy.

Trust is earned. Let the agent demonstrate that it can research the problem properly. Let people inspect the evidence. Let them challenge the recommendation. Then decide what you’re comfortable allowing it to do.

My advice: don’t start with your biggest process. I’d be tempted to start somewhere much more boring. Find something that happens every week. Something with a clear owner. Something that currently takes an experienced person an hour or two. Something involving enough research and judgement to be annoying, but not something so business-critical that everyone becomes terrified of changing it.

Run it in Research. Compare what Fusion Claw produces with what the person would have produced. Understand where it gets stuck. Understand where your data, policies or processes aren’t quite as clear as you thought they were. Then improve it.

A weekly process may teach you far more about agentic working than attempting to automate an enormous quarterly process on day one. Oracle makes a similar recommendation: start with a recurring use case that can be proven in Research mode.

For a long time, enterprise AI conversations have started with: “Where can we use AI?” With Fusion Claw, I’d change the question. Ask your Finance team: “What takes you half a day every week?” Ask HR: “What information do you have to pull together before you can make a decision?” Ask Procurement: “What exceptions do your experienced people spend their time investigating?” Ask managers: “What do you repeatedly have to check before you’re comfortable taking an action?” And perhaps most importantly: “What work are our people doing today where the value is in the outcome, not in them personally clicking through every step required to get there?”

That’s where I’d start looking for Fusion Claw. Because the interesting thing about agentic AI isn’t another clever way of interacting with Oracle Fusion. It’s the possibility that, increasingly, we can give Fusion an objective and let it help get the work done. And that feels like a much bigger change.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Oracle AI World Day 1: From Helping People Work to Getting Work Done

I wasn’t in Las Vegas for Oracle AI World this year. Like many Oracle customers and partners around the world, I followed the event remotely using Oracle’s free On Air pass, catching up with the keynotes from the UK rather than from the conference floor. While watching remotely means missing some of the atmosphere and conversations that happen between sessions, it does give you the opportunity to focus entirely on the messages being delivered.

Having watched both Mike Sicilia’s opening keynote and Chris Leone’s product keynote, one theme stood out above everything else. Oracle is no longer talking about helping people work faster. Oracle is talking about getting the work done. At first glance, that might sound like a subtle distinction. I don’t think it is. In fact, I think it represents one of the biggest shifts we’ve seen in enterprise applications for years.

For the last couple of years, most AI conversations have been centred around productivity. How can AI draft content faster? How can AI summarise meetings? How can AI help employees find information more quickly? All of those things matter. They deliver value. But fundamentally, the human remains responsible for completing the task.

What struck me during the Day 1 session was how little time Oracle spent talking about AI as an assistant and how much time it spent talking about AI as an executor. Across both keynotes, the discussion repeatedly returned to outcomes. Not generating insights. Not producing recommendations. Delivering results. That feels like an important evolution in enterprise AI.

One of the things I liked most about Mike Sicilia’s keynote was that the focus wasn’t really technology. It was people. The message wasn’t that AI should replace people. It was that organisations should be able to remove the repetitive, burdensome activities that consume time and energy, allowing people to focus on higher-value work.

For years we’ve accepted certain activities as simply being part of work. Reconciling data. Chasing approvals. Managing exceptions. Navigating complex processes. Collecting information from multiple systems. What if those activities become the responsibility of software instead? That was the question I kept coming back to throughout the keynote. The vision wasn’t more intelligent software. It was more capable organisations. That’s a subtle but important difference.

If Mike Sicilia explained the vision, Chris Leone demonstrated how Oracle intends to deliver it. The progression shown during the keynote was fascinating. We have already seen Oracle introduce embedded AI capabilities across Fusion. We have seen AI agents emerge. Now Oracle is talking about agent teams, workflow agents, agentic applications and autonomous execution.

The key message was that AI is evolving from supporting processes to actively participating in them. That shift matters. Historically, enterprise software has recorded work. Increasingly, Oracle appears to be building software that performs work. Not without governance. Not without controls. But certainly with a higher degree of autonomy than most organisations are accustomed to today.

There were plenty of announcements during the keynote, but Fusion Claw was the one that immediately stood out to me. Not because it was the most futuristic. Actually, quite the opposite. What interested me was the emphasis on governance.

Most organisations are no longer questioning whether AI can generate content, identify patterns or make recommendations. The real question is whether AI can be trusted to execute business processes. That is where many current AI discussions begin to stall. The answer is rarely a technology problem. It is a governance problem.

Throughout the demonstrations, Oracle repeatedly emphasised operating boundaries, permissions, policies, approvals and organisational controls. In other words, autonomy with accountability. That feels much more relevant to enterprise customers than unrestricted AI experimentation. The organisations I speak to rarely say, “We need AI to make decisions independently.” What they actually say is, “We need AI to help us achieve outcomes without introducing risk.” Fusion Claw feels like Oracle’s response to that challenge.

Another concept that caught my attention was Oracle’s discussion around systems of outcomes. For decades, enterprise applications have been systems of record. They store information. Track transactions. Enforce processes. Maintain business truth. Those capabilities remain essential.

Oracle’s argument is that organisations increasingly need something more. A layer focused on achieving business outcomes rather than simply recording business activity. The more I think about it, the more significant that feels. After all, organisations don’t exist to process invoices. They don’t exist to approve purchase orders. They don’t exist to complete workflow transactions. Those activities support the real objectives. Better customer experiences. Stronger financial performance. Improved workforce productivity. Increased organisational agility. More resilient supply chains. Those are the outcomes that matter. If AI can continuously coordinate activity towards those outcomes, then we are moving into genuinely new territory.

Fusion Claw will probably attract most of the headlines. However, I suspect Fusion Companion could prove just as important over time. As enterprise landscapes become increasingly complex, users don’t think in terms of applications anymore. They think in terms of work. They move between systems, channels, data sources and processes. The value increasingly comes from connecting those experiences together.

The vision Oracle presented for Fusion Companion suggests a future where intelligence follows the user rather than remaining confined within individual applications. If delivered successfully, that could become one of the most practical and impactful developments announced during the keynote.

Anyone who has read my recent blogs will know I’ve already posted about Fusion Setup and Configuration Registry (FSCR). Seeing it formally announced during the keynote only reinforced my belief that this could become one of the most important long-term developments for the Oracle ecosystem.

For years, implementation partners have built methodologies, accelerators, templates and frameworks to help customers configure Fusion applications. FSCR has the potential to fundamentally change that model. If Oracle can successfully connect requirements, process design, configuration and AI-driven implementation, then the way organisations deploy Fusion could look very different in the future. It is still early days. But of all the announcements made on Day 1, FSCR may ultimately have one of the biggest impacts on how Oracle projects are delivered.

When I finished watching the keynotes, I kept coming back to one simple thought. For the last two years, the industry has largely been asking: “How can AI help people do their jobs faster?” Oracle now seems to be asking: “How can AI help organisations achieve outcomes with less human effort?” That is a much bigger question. And potentially a much more transformative one.

Whether discussing agentic applications, Fusion Claw, Fusion Companion or FSCR, the underlying message felt remarkably consistent. The goal is no longer simply helping people work. The goal is getting the work done. That doesn’t mean removing people from the process. It means allowing people to spend less time managing the mechanics of work and more time focusing on the decisions, creativity, relationships and innovation that humans are uniquely positioned to provide.

Watching remotely, that was my overwhelming impression of Oracle AI World Day 1. The conversation has moved beyond AI assistance. Oracle is now focusing on AI execution. And that feels like the start of a very different chapter for enterprise applications.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Why Oracle FSCR Could Change How Fusion Is Implemented

At Oracle AI World, there has been and will be plenty of attention on AI agents, autonomous workflows and the future of enterprise applications. Those announcements will undoubtedly generate the headlines. But one announcement caught my attention for a different reason.

Oracle introduced the Fusion Setup and Configuration Registry (FSCR), and while it may not be as immediately eye-catching as some of the AI announcements, it could prove to be one of the most important foundations Oracle has put in place for the future of Fusion applications. Because before AI can intelligently configure Oracle Fusion, it first needs to understand Oracle Fusion. FSCR appears to be Oracle’s answer to that challenge.

When organisations implement Oracle Fusion, a huge amount of knowledge sits outside the application itself. Configuration decisions are made during workshops. Dependencies are captured in design documents. Best practice is spread across implementation guides, project artefacts, customer experience and consultant expertise. Some of that knowledge is documented. Some lives in spreadsheets. Some exists purely in the heads of the people delivering the project. Over time, this creates an interesting situation.

The technology becomes more intelligent, but the knowledge needed to implement and configure it often remains fragmented. As Oracle continues its journey towards agentic applications and autonomous business processes, that fragmentation becomes increasingly important. After all, how can an AI agent configure a system if it doesn’t have access to the knowledge required to make those decisions?

At its core, FSCR is Oracle’s central repository of Fusion setup and configuration knowledge. It brings together setup tasks, dependencies, validation rules, best practices and implementation knowledge into a structured, machine-readable framework. The phrase “machine-readable” is important here. This isn’t simply another set of implementation documents.

Oracle is creating a knowledge layer that can be consumed not only by people, but also by AI-powered services, agents and automation tools. In effect, Oracle is moving implementation knowledge closer to the product itself. That feels like a significant shift.

Most customers don’t wake up in the morning thinking about configuration registries. They care about delivering projects faster, reducing risk and getting value from their Oracle investment. That’s where FSCR becomes interesting.

By capturing implementation knowledge in a structured way, Oracle creates the potential for a much more consistent approach to configuration and deployment. Instead of repeatedly recreating the same implementation knowledge across multiple projects, that knowledge can become reusable, governed and accessible. The obvious benefit is efficiency. The bigger benefit may be consistency.

Anyone who has worked on multiple Oracle Fusion programmes knows how much time is typically spent checking dependencies, validating setup choices and understanding the impact of changes across different areas of the application. A central knowledge framework has the potential to reduce much of that effort.

I’ve written quite a lot recently about Agentic Applications and the broader move from automation to autonomy. Most conversations focus on the visible layer. The agent. The user experience. The conversational interface. The outcome. What receives far less attention is the knowledge framework that sits underneath.

Enterprise AI becomes significantly more powerful when it can move beyond making recommendations and start executing trusted actions. But trusted execution only happens when there is confidence in the underlying knowledge. That’s why FSCR feels important.

It provides the structured understanding of Fusion configuration that future AI capabilities will need if they are going to do more than simply answer questions. In many ways, it feels like Oracle is putting the foundations in place before building the next floor of the house.

The interesting thing about FSCR is that its potential value isn’t limited to new implementations. Most organisations spend far longer running Oracle Fusion than they do implementing it. Over time they introduce new functionality, adopt new releases, optimise processes, expand into new business areas and respond to changing organisational priorities.

Every one of those activities requires knowledge of how the system has been configured and how different setup decisions interact with each other. If Oracle can maintain that knowledge as a living framework, the opportunities extend well beyond implementation projects. It could support ongoing optimisation, change management, governance and continuous improvement throughout the lifecycle of a Fusion environment. For many organisations, that’s where the real value may emerge. This functionality isn’t there yet though, the initial focus is on new implementations, but Oracle have been plans on expanding it’s functionality.

You might be saying, so what? It’s still early days, and like any significant innovation, the true impact will depend on how Oracle develops the capability over the coming releases. But the direction of travel is clear. Oracle is looking to move implementation knowledge from documents and individuals into a trusted, structured framework that can support both people and AI. For customers, that could ultimately mean:

  • Faster implementations
  • Greater consistency across projects
  • Reduced dependency on individual expertise
  • Improved validation and governance
  • Better support for future AI-driven configuration and optimisation

None of these benefits happen overnight. Organisations will still need experienced teams, strong governance and informed decision-making. But if Oracle’s vision becomes reality, FSCR could help remove much of the repetitive effort that currently surrounds system configuration and implementation.

I’ve often said that the most important developments in enterprise AI are usually the ones people notice the least. Everyone notices the chatbot. Everyone notices the agent. Everyone notices the flashy demo. Far fewer people pay attention to the infrastructure that makes those capabilities possible.

FSCR feels like one of those foundational innovations. On its own, a setup and configuration registry isn’t particularly exciting. But viewed in the context of Oracle’s wider AI strategy, it starts to look much more significant. Because before Oracle can deliver truly autonomous implementation and optimisation capabilities, it first needs a way for AI to understand how Oracle Fusion works. FSCR might not be the announcement that generates the most headlines from AI World. But in a few months’ time, it may prove to be one of the most important.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Is RODS the Future of Oracle Fusion Data Extraction?

If you’ve been working with Oracle Fusion for any length of time, you’ll know that getting data out of Fusion is often just as important as getting data into it. Whether it’s feeding a data warehouse, supporting reporting requirements, populating a data lake, or increasingly providing the data needed to power AI initiatives, organisations are extracting more Fusion data than ever before.

Traditionally we’ve had a number of options available to us, including HCM Extracts, BI Publisher and BICC. They all have their place and continue to solve different business challenges. But Oracle has recently introduced something new that could become a significant part of its long-term data extraction strategy: the Read Optimised Data Store (RODS).

So, what is RODS? At a high level, RODS is exactly what the name suggests: a read-optimised copy of Fusion data designed specifically for extraction and reporting workloads. Instead of running large extraction jobs against the live transactional database, Oracle replicates selected Fusion data into a separate environment using GoldenGate technology. Extraction processes then run against this replicated data store rather than the core application database. The goal is simple. Make high-volume data extraction faster while reducing the impact on day-to-day Fusion operations.

For many organisations, extraction volumes continue to grow. Data is no longer only being extracted overnight for a monthly report. Today’s integrations, analytics platforms, AI services and operational reporting solutions often require much larger volumes of data and much greater frequency.

Historically, there has always been a balance between extracting the data you need and consuming resources that could otherwise be used by the application itself. RODS changes that dynamic. Because extraction workloads are separated from the transactional environment, organisations can run large-scale extracts without placing the same demand on the Fusion application database. That could be particularly valuable for customers with significant reporting requirements or complex integration landscapes.

RODS is supported by Oracle’s new Data Extraction Tool (DET), which provides the framework for defining, scheduling and managing extracts. The Data Extraction Tool supports both full and incremental extracts and can output data in either CSV or JSON format. For organisations building modern data platforms, that’s likely to be exactly what they need. Clean, structured data delivered quickly and efficiently.

Does this replace HCM Extracts? No. And I think that’s an important point to understand. Whenever Oracle introduces a new capability, one of the first questions customers ask is whether it replaces what’s already there. The answer, at least today, is no.

HCM Extracts remain incredibly powerful, particularly where organisations require complex formatting, business logic, Fast Formula processing or highly tailored output files. RODS isn’t trying to solve those problems. Instead, it is focused on extracting large volumes of data efficiently and making that data available for downstream consumption. In reality, I suspect most organisations will continue to use a combination of technologies depending on the requirement.

The most interesting aspect for me isn’t necessarily the technology itself. It’s Oracle’s direction of travel. Oracle has openly discussed RODS as the long-term strategic direction for high-volume extraction and replication workloads, with many seeing it as the natural successor to BICC. That doesn’t mean existing extraction methods are going away anytime soon, but it does suggest where Oracle’s investment and innovation are likely to be focused moving forward. For customers making long-term architectural decisions, that’s worth paying attention to.

One area where I think RODS could become increasingly important is AI. As organisations start deploying more AI services, AI agents and autonomous processes, access to timely, high-quality operational data becomes critical. Oracle has confirmed that data replication into RODS occurs in near real-time, creating opportunities for more responsive analytics, automation and AI-driven decision making. Whilst we’re still at the early stages of seeing how customers will use this in practice, the link between modern data architectures and enterprise AI is becoming increasingly obvious.

RODS probably isn’t a feature that will get end users excited. Most employees will never know it exists. But for technical teams, data teams, integration specialists and architects, it could become a very important capability. The ability to extract large volumes of Fusion data quickly, without impacting transactional workloads, addresses a challenge many organisations have wrestled with for years.

I don’t see RODS replacing every extraction method. HCM Extracts, BI Publisher and other tools still have an important role to play. What I do see is Oracle laying the foundations for the next generation of Fusion data extraction, one that is built around scale, performance and increasingly the data demands of analytics and AI. If you’re currently using BICC extensively or are designing your future Fusion integration architecture, RODS is definitely something I’d be taking a closer look at.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Fusion Claw: Is This Oracle’s Most Important AI Announcement Yet?

Over the past couple of years, we’ve seen Oracle steadily evolve its AI story. It began with AI capabilities embedded within Fusion Applications. Then came AI Agents, followed by AI Agent Studio, giving customers the ability to build, extend and govern their own AI-powered processes. Oracle has unveiled the next piece of the puzzle: Fusion Claw.

I’ll admit that when I first heard the name, I wasn’t entirely sure what to make of it. Another AI agent? A new orchestration framework? Perhaps just another acronym to add to the ever-growing AI dictionary? Having sat through the recent webinar, I came away thinking this may actually be one of the most significant Oracle AI announcements we’ve seen so far.

Not because it introduces more AI. But because Oracle appears to be addressing one of the biggest challenges organisations face today: how do you move from AI recommendations to AI-driven outcomes whilst maintaining governance, accountability and human oversight?

Most enterprise AI today is still fundamentally assistive. Whether it’s generating content, summarising information, analysing data or helping users complete transactions, AI is primarily helping people do their jobs more efficiently. Even many AI agents stop at recommendations. They identify an issue, gather information, analyse options and suggest a course of action. A human then reviews the output and decides what happens next. Fusion Claw appears to be Oracle’s answer to that gap.

Oracle positions Fusion Claw as a governed execution runtime that sits between AI planning and enterprise execution. People define objectives, policies and desired outcomes. Agentic Applications orchestrate business processes. Fusion Claw plans and executes activities. Fusion Applications remain the system of record. That distinction is important.

Oracle is not proposing a future where AI operates outside enterprise controls or invents its own ways of working. Instead, Fusion Claw operates against the same business objects, security model and governance framework that organisations already trust. In simple terms, Oracle is creating a mechanism for translating business outcomes into governed action.

Fusion Claw is designed for complex, high-value work rather than simple transactional activities. Its core capabilities are Analyse; Reconcile; Simulate; Optimise; Predict; and Execute. Those first five capabilities are becoming increasingly common across enterprise AI platforms. The sixth is what makes Fusion Claw different. Execution has traditionally been the point where AI recommendations hand off to humans. Oracle is now seeking to close that gap.

Imagine an organisation identifying a workforce issue, supply chain disruption, financial variance or operational risk. A traditional AI capability might identify the problem. An AI agent might recommend a solution. Fusion Claw is designed to take things further by evaluating multiple options, simulating outcomes, optimising the approach and, where authorised, executing approved actions. That feels like a notable evolution in Oracle’s AI strategy.

The focus is no longer: “How can AI help me complete this task?” Instead, the question becomes: “How can AI help achieve this business outcome?”

One of the most interesting concepts introduced during the session was the distinction between two the operational models Deep Research vs Full Auto. In Deep Research mode, Fusion Claw investigates, analyses and simulates. It evaluates scenarios, compares potential outcomes, surfaces recommendations and presents options for human review.

Importantly, the human remains responsible for making the decision. For many organisations, particularly those operating in highly regulated environments, this will likely be the most comfortable starting point. The value comes from faster analysis and better decision support, whilst maintaining a clear human approval process.

Full Auto mode is where things become significantly more transformative. In this model, Fusion Claw can plan and execute authorised activities automatically in pursuit of a defined business outcome. That naturally raises questions. Who controls it? Who is accountable? What prevents unintended actions?

Anyone who has read my blogs over the past year will know that governance is often the first question that I and customers ask. The technology itself is rarely the challenge. Trust is. The encouraging aspect of Fusion Claw is that governance appears to have been designed into the platform from the outset rather than added afterwards.

Fusion Claw operates inside a private runtime container within the customer’s Fusion environment and works directly with authoritative Fusion business data. Perhaps most importantly, Fusion Claw does not grant itself authority. Organisations decide what actions can be taken, under which circumstances and with what level of oversight. Human accountability remains in place. For customers concerned about explainability, risk and compliance, that message will be every bit as important as the AI capabilities themselves.

One feature that particularly caught my attention was the concept of an Outcome Receipt. Every execution performed by Fusion Claw generates a detailed record that explains: What happened? Why it happened? What information was used? Which controls were applied? Which actions were executed?

At first glance, this sounds like an audit feature. In reality, it could become one of the most important capabilities in the platform. As AI becomes responsible for increasingly valuable business activities, organisations will need evidence. They’ll need to demonstrate how decisions were reached, what data was considered and who authorised the action. For public sector organisations, financial institutions and heavily regulated industries, this level of transparency could prove essential. Many organisations are no longer asking whether AI can make decisions. They’re asking whether they can prove how those decisions were made.

Most enterprise systems operate on a transaction basis. Users initiate activity, the system responds and the process ends. Fusion Claw appears designed to work differently. It is runtime that continuously monitors conditions, reassesses plans and adjusts activities in pursuit of an outcome. Execution can be triggered by user requests; scheduled activities or configured triggers. Rather than waiting for users to notice and react, Fusion Claw is designed to continuously work towards a defined objective. That’s a subtle but potentially significant shift in how we think about enterprise applications.

One other key point of note is Fusion Claw’s ability to work across business functions. Rather than being constrained to a single process or module, it is intended to support end-to-end activities spanning multiple Fusion pillars. This cross-functional capability is important because real business outcomes rarely exist within the boundaries of a single application module.

Most organisational challenges span departments, processes and data sources. Fusion Claw appears designed with that reality in mind. It should also be noted that whilst Fusion Claw is firmly rooted within Fusion Applications, it can interact with external services and other agents where appropriate access has been configured.

A practical concern for many customers will be whether existing investments become obsolete. Fortunately, Oracle’s answer appears to be no. They have confirmed that existing agents can be reused and enhanced rather than rebuilt from scratch. Given that many organisations are only just beginning their AI Agent Studio journey, that will be welcome news.

For those already thinking ahead to roadmap planning, there are a few key details you should know. Fusion Claw is currently targeted for the 26D timeframe. If you have already purchased the licence for the Agentic Applications, you will automatically be able to use Fusion Claw.

By the time you receive 26D, Oracle will have provided 75 Agentic Applications out of the box, of which 25 will be Fusion Claw-powered applications focused on more complex enterprise activities. You won’t be limited to just these 25 Fusion Claw applications, as with all AI Agents and Agentic Applications, you can build your own. This can be via the No Code methods in AI Agent Studio or the Low Code options using CLI.

So, Is This Oracle’s Most Important AI Announcement Yet? Maybe. It’s certainly one of the most ambitious. AI Assist features help users work faster. AI Agents help users make better decisions. Fusion Claw appears to be aiming for something larger: helping organisations achieve business outcomes through governed AI execution. Whether customers are ready to embrace that level of autonomy remains to be seen.

Some organisations will begin with Deep Research, keeping humans firmly in the decision-making loop. Others will be attracted by the possibility of allowing approved activities to execute automatically within carefully defined guardrails. Either way, the conversation is changing. We’re moving beyond asking how AI can assist users and starting to explore how AI can help organisations achieve measurable outcomes. For me, that’s the real significance of Fusion Claw.

The technology itself is impressive, but the bigger story is what it represents: Oracle’s vision for moving from AI that advises, to AI that acts, whilst still maintaining the governance, security and accountability that enterprises demand. And that may ultimately make Fusion Claw one of the most important steps yet in Oracle’s AI journey.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Oracle Learning Paths, Offerings and Redwood: What’s Actually Changing?

Before we get started, a quick thank you to one of my blog readers. This article came about following a direct request from a customer who asked if I could explain what is happening with Oracle Learning Offerings and the new Learning Paths functionality. It’s the first time I’ve been specifically asked to write a blog on a particular Oracle Fusion topic, and after looking into it I realised they’re probably not alone in being confused.

If you’ve recently seen references to Learning Paths, heard that Offerings are going away, or wondered what Redwood means for Oracle Learning, you’re not the only one. Oracle is changing terminology, introducing new Redwood-native learning objects and modernising Learning administration at the same time. That makes it difficult to understand exactly what is changing and, more importantly, what it means for customers who have spent years building and managing learning content in Oracle Learning. So let’s unpack it.

One of the biggest sources of confusion is that two separate changes are happening at the same time. In Oracle Fusion Learning 26B, Oracle rebranded Specialisations as Learning Paths. Oracle states that the self-service learning experience now consistently presents Specialisations as Learning Paths, although administrators may still see the old terminology in some classic administration pages for the time being.

Learning Paths are therefore the new name for what many Learning administrators have known as Specialisations. They are designed to provide a structured learning journey containing multiple learning items, helping organisations create role-based, compliance-based or career-development learning programmes. They are not a direct replacement for a single Offering.

For many years, Oracle Learning followed a fairly structured hierarchy. A typical eLearning item might look something like this: Course → Offering → Activity. This provided flexibility but often resulted in multiple objects needing to be created and maintained for even relatively simple learning content. If you wanted to publish a straightforward self-paced training course, you typically found yourself creating several related objects before learners could access the content. It worked, but it wasn’t always intuitive.

Oracle has been redesigning Learning as part of its broader Redwood strategy. One of the most significant developments arrived in 24D with the introduction of the new Redwood Self-Paced Learning experience. Oracle describes this as a redesigned process that combines what were previously separate activities into a single workflow, bringing together content, activities and learning publication within one experience.

The goal is simple: reduce complexity; simplify administration; improve usability and create a more modern learner experience. Rather than forcing administrators through multiple setup steps, Oracle is moving towards a more streamlined content creation model.

So are Offerings being removed? This is where things get interesting. There is a lot of discussion within the Oracle Learning community about Offerings becoming less central to the Learning architecture. Industry commentators have suggested Oracle’s long-term direction is towards Redwood-native learning objects rather than the traditional Offering-centred model. However, what is important is separating community interpretation from Oracle’s documented position.

What Oracle has clearly documented is:

  • Specialisations are becoming Learning Paths.
  • Redwood Self-Paced Learning introduces a simplified approach to managing self-paced learning content.
  • Oracle continues to invest heavily in Redwood Learning experiences.

What I have not yet found in Oracle readiness documentation is a statement saying: “Learning Paths replace Offerings.” In fact, these are different concepts serving different purposes. Think of a Learning Path as a curriculum rather than an individual training item. A Learning Path might include: multiple self-paced learning items; instructor-led training; assessments; recommended learning sequences or mandatory prerequisites. In other words, Learning Paths help define what learners should complete and in what order. They are ideal for: new starter onboarding; leadership programmes; compliance journeys; certification pathways and role-based development programmes. This is broadly the same purpose that Specialisations have always served, just under a new name.

For most organisations, I don’t think there is a need to panic. My advice would be: first, start becoming familiar with Redwood Learning administration if you haven’t already. Second, review how much you rely on traditional Course and Offering structures for new content creation. Third, start thinking about Learning Paths from a learner journey perspective rather than viewing them as a replacement for individual training content. Most importantly, don’t assume that every mention of Learning Paths is related to Offerings. In many cases Oracle is simply referring to the rebranding of Specialisations rather than introducing a direct replacement for existing Offerings.

Having worked with Oracle Learning for many years, I can understand why customers are asking questions. The terminology changes are significant, and because they are arriving alongside the wider Redwood transformation it’s easy to assume everything is changing at once. What I find encouraging is Oracle’s clear focus on simplifying Learning administration. The historic Course, Offering and Activity model could be powerful, but it could also feel overly complex for straightforward learning scenarios. Redwood Self-Paced Learning is a good example of Oracle reducing that complexity and making content easier to create and maintain.

For now, my advice is simple: treat Learning Paths as the evolution of Specialisations, keep an eye on Oracle’s Redwood Learning roadmap, and avoid assuming that Learning Paths are a direct one-for-one replacement for Offerings. The reality is a little more nuanced than that.

And finally, thank you again to the customer who suggested this topic. If there are other areas of Oracle Fusion you’d like me to explore in a future blog, feel free to get in touch. It seems I’ve accidentally opened the suggestions box.

Please note all screenshots are the property of Oracle and are used according to their Copyright Guidelines.

Could AI Be the Missing Piece in the Redwood Migration Puzzle?

I’ll admit it. This is probably the first time I’ve written a blog focused on Oracle CX. Most of my content tends to centre around HCM, ERP, SCM, AI Agent Studio, or the wider Oracle Fusion platform. However, over the last few months I’ve been spending a lot more time working with Oracle Helpdesk. As Helpdesk sits within the Oracle CX product suite, that has naturally led me to take a closer look at some of the innovation happening within CX.

When I discovered that Oracle had developed an AI-powered assistant specifically designed to help organisations accelerate their move from the Classic user experience to Redwood, I was intrigued. Redwood migrations are a challenge that many Oracle customers are facing, and the idea of using AI not just to answer questions, but to actively guide and support that journey, felt like an interesting glimpse into where enterprise AI is heading.

It’s worth calling out an important caveat before I go any further. This particular capability is currently designed for Oracle CX Service and its associated extensibility framework. It is not available for HCM, ERP or SCM. However, what interested me as much as the solution itself was the fact that the CX product team is actively exploring and proving patterns that could potentially influence future AI-assisted implementation and configuration experiences across other Oracle product areas.

One of the biggest challenges with Redwood adoption is not usually understanding why you should move. Oracle has been clear about the benefits for some time. Redwood provides a more modern user experience, improved efficiency, greater flexibility through Visual Builder Studio, intelligent recommendations, and access to functionality that is only available within the Redwood user interface.

The real challenge is often understanding how to get there. Customers frequently ask questions such as: How do I recreate an existing extension in Redwood? Is a particular extensibility option available in my target release? What is the supported way to add custom functionality to a specific page? How do I avoid building something that won’t survive the next quarterly update? These are exactly the types of questions implementation teams spend significant amounts of time researching.

What caught my attention was that Oracle’s approach isn’t simply to place a large language model in front of documentation and hope for the best. Instead, the assistant is designed to provide grounded guidance using curated, release-aware sources that are specific to Fusion Service extensibility. The aim is to help users understand supported approaches, identify limitations, and surface any missing information required to answer the question accurately. That distinction is important.

One of the biggest misconceptions about enterprise AI is that the value comes from the model itself. In reality, the value often comes from the quality of the information that sits behind it. A generic AI model may be incredibly capable, but if it doesn’t understand the specific product, release, extensibility framework, or support boundaries involved, its answers can quickly become unreliable.

The CX team appears to have recognised this. The assistant routes questions to curated repositories of release-specific information and clearly highlights uncertainty or gaps where additional context is needed. Rather than positioning AI as an infallible expert, it is positioned as an accelerator for research and decision making. I also liked the emphasis on human judgement.

The guidance repeatedly reinforces that users should provide context, review the recommendations, and validate the proposed approach before implementation. In other words, AI helps you get to an answer faster, but it doesn’t remove the responsibility to apply experience and critical thinking. That’s a message I’ve been repeating in many of my recent AI-focused blogs, regardless of whether we’re talking about Agentic Apps, generative AI, or implementation accelerators.

Perhaps the most interesting aspect is where this could lead. Today, the capability focuses on Fusion Service extensibility and Redwood migration activities. It can help users understand new release functionality, identify supported implementation patterns, and even assist with Visual Builder Studio-related work.

Tomorrow, could we see similar experiences helping HCM administrators understand Redwood customisations? Could ERP teams receive guided support when adapting role-based pages? Could SCM implementation consultants get release-aware recommendations about supported configuration options?

Those are questions only Oracle can answer. But what seems clear is that the CX team is exploring a practical application of AI that goes beyond simply generating content. Instead, it aims to reduce implementation effort, improve confidence, and help customers reach the right outcome more quickly. For me, that’s where enterprise AI becomes genuinely interesting. Not when it replaces expertise, but when it helps organisations make better use of it.

Please note that all screenshots are the property of Oracle and are used in accordance with Oracle’s Copyright Guidelines.