AI’s value comes down to integration and articulation within systems

The organisation only feels the value of AI when it is deeply connected to the business it serves, writes Jacqui Bond, practice head of AI transformation services at Braintree.

In 2026, companies are simultaneously pressured to invest in AI and showcase the value of their AI investments. It is not, in many use cases, delivering its promised economic value – PwC’s 2026 AI Performance study showed that 74% of AI’s economic value is felt by only 20% of companies. Gartner’s report has a similarly dour message with only 28% of AI use cases in infrastructure and operations fully succeeding and delivering the value companies expected. In many instances the reasons for this failure to launch value come down to handling errors around ownership, misaligned design, a lack of measurement infrastructure, and an accountability vacuum.

And the reason for these errors often comes down to unclear expectations. Many companies expect the AI to simply start working within the business, thinking for them and reading the business intelligently all while delivering conclusions and data points that will drive the business forward. Unfortunately, AI will only thrive within guardrails that have been established by the business and will only flourish when there is a human in the loop.

A case in point is how AI can flourish when running within the organisation’s database. It can run compliance and policy and retrieve information from within your systems flawlessly and faster than anything else you have, but it falls short on the soft issues. It doesn’t understand how a border closing in one country will impact the cost of fuel in another or recognise that a person about to retire is taking with them a wealth of information. If the soft issues are excluded, AI fails, if companies take the factual output and overlay it with their own interpretations, then the AI succeeds.

The value was never in the AI but in how deeply it is connected to the business it serves. An agent bolted onto the side of a system knows nothing about that system, but when it is embedded it has the context, the customer history, the journal history, the vendor history, and the tables it needs to answer a question properly. An embedded agent inside your ERP is capable of providing you with an overall financial view because it is sitting inside the business.

It’s an approach that does take time because value is incremental as the quality of documents rises, as policies are more effectively enforced because AI is checking against them, and as reworking drops. The teams close to policy and approval notice first because the quality improves from the ground upwards, but when it comes to month end? AI embedded in your ERP is going to catch those mistakes and flag them early so your teams aren’t working through the night to resolve issues. Instead, your teams are ditching the reactive life for a proactive one that is supported by an intelligent system that understands the business.

Microsoft has built its strategy around this approach. In Dynamics 365 Finance & Operations, the intelligence is embedded with Copilot already optimising and refining systems. Open a product listing, and Copilot is offering insight. Ask it to raise a purchase order, and it already knows how. The AI sits within the wider ecosystem so it can assess finance-level requests, check queries in Outlook, assess files in SharePoint and so much more. It is entirely possible to verify a journal entry with the AI and it will confirm the fields are filled, notice any figure discrepancies, query stray numbers, and flag any descriptions or issues that don’t make sense. This is real intelligence applied out the box but integrated within the business rather than from without.

None of this pays off, however, if the business doesn’t do the work. And the work means a well-defined AI strategy that answers questions like: Is the AI embedded in the systems where the work already happens or bolted on? Is the data feeding it secured to the standard the business is held to? Is there a strategy above agents and an orchestration layer between them, or are departments buying in isolation? And is anyone measuring the things that are hard to measure like hours saved, errors avoided and delighted customers?

This is where the value lives, and is also why Microsoft’s strategy has prioritised AI that sits within its systems, delivering an ERP that works coherently with the business and helps it to answer all the right questions.

The post AI’s value comes down to integration and articulation within systems appeared first on IT-Online.