AI in the systems you already use
The platforms a company already runs are adding AI features and offering them to the business. Operient works through one workflow at a time: whether to accept the AI the platform already includes, reuse the APIs, data platform, access control and integration layer the company owns, build something specific, or buy something new, and who owns that decision.
Illustrative example built on simulated data. Not a client engagement.
The situation
A logistics company with 900 employees runs an ERP platform, a service-management system, an office suite and a data platform. In the same quarter, each of those platforms announced AI features and offered them to the business: assisted coding of supplier invoices, AI categorisation and routing in service management, assistant licences in the office suite and AI functions in the data platform.
Each announcement arrived with its own owner in the business, its own budget line and its own assumption that the feature should be switched on. The leadership team wanted one decision per workflow, taken the same way every time, instead of a separate answer for each platform.
What Operient assessed
For each workflow the platform AI would touch, the assessment asked the same questions: does the workflow justify intervention at all, and does the AI the platform already includes cover it, under what conditions? Can the need be met by what the company already owns, the APIs, the data platform, the access control and the integration layer, without new licences? Is there a case to build something specific, and is there a case to buy something new?
Each answer was recorded with the evidence behind it, the conditions attached to it, and the person accountable for taking it and for running the result. The assessment also recorded what would make the decision wrong, so it can be reopened deliberately rather than by drift.
The decisions
| Workflow | What was on the table | Decision | Why | Decision owner |
|---|---|---|---|---|
| Categorise and route internal service requests | AI categorisation already included in the service-management platform | Accept and switch on | 4,200 requests a month, clear rules and low risk. Errors are easy to correct | IT service owner |
| Match supplier invoices to purchase orders | AI matching already included in the ERP platform | Accept with conditions | 8,000 invoices a month, but 15% are exceptions and financial controls apply. A person approves every exception | Finance operations, with the ERP platform owner |
| Report on delivery performance | A new AI reporting add-on sold on top of the data platform | Reuse what we own | The data and the permission model already exist. A new licence adds cost without new evidence | Head of analytics, as data owner |
| Draft replies from product documentation | Platform AI covers part of it, but the content and access rules are company-specific | Build a thin layer | Smaller than a new platform, and it runs on APIs the company already governs | CIO, with the platform owners |
| Extract terms from scanned supplier contracts | Nothing already in use covers it | Buy new, narrow scope | The need is bounded and proven, and no platform in use covers it today | Head of procurement, with legal |
What the company decided
Operient doesn't rank vendors or recommend products. It makes the decision for each workflow explicit and records it: which route was taken, on what evidence, who is accountable for it and when it is reviewed. Here the leadership team accepted platform AI for two workflows, one of them with additional conditions, reused what it already owned for one, commissioned a thin build for one and bought a narrow tool for one.
Each decision is recorded with its rationale and the version of the method used, so it can be reopened when the platforms change what they include, or when the rules change.
← All examplesAre your platforms already offering you AI features?
Start with a first assessment of one process area.