Make experienced colleagues easier to find with evidence from completed work

Your project lead needs someone who has already handled an intermittent drive fault on this equipment, and spends the morning asking around to find them. An expertise directory built on completed work records helps that lead reach the right colleague faster, and shows which job makes that colleague worth calling.

An ontology is a shared business model of your company: it links people, projects, equipment, service jobs and the role each person held, so software can answer “who handled this?” using the same business meaning your teams use. Functional AI Solutions builds that model over your existing service, project and HR systems and automates retrieval from the sources each user is allowed to see.

What does an evidence-backed expertise directory give you?

• Less asking around. Search by equipment and fault, and get colleagues linked to completed jobs on that equipment.
• Better first conversations. Each suggestion shows the job, date and recorded role, so the requester arrives with a specific question.
• More value from systems you already run. Your maintenance and project records already hold this history; the ontology makes it searchable by person.
• A foundation you reuse. Once equipment, employees and responsibilities share one identity across systems, the next workflow, such as dispatch or onboarding, starts from that model.

The mechanism is the relationship: a service record states that a technician diagnosed a fault on a specific machine on a specific date. The ontology ties that technician to one employee identity, that machine to one equipment family and that action to one defined role. The directory searches those relationships rather than loose text.

Why completed work records belong in the search

Microsoft’s 12 October 2026 account of its own HR transformation describes how its employees use the People agent (Frontier) to find colleagues, mentors and subject-matter experts, and to prepare for those conversations with relevant context.

The source of a suggestion shapes what it can tell you. Microsoft’s People Skills documentation, last updated on 5 June 2025, says its skill inference currently draws only on Microsoft 365 signals such as documents, emails, chats and meetings, and does not use activity in other line-of-business applications. The same page gives a typical 30-day refresh cycle and excludes activity older than 90 days. It notes performance may be limited for frontline and field workers who spend little time in Microsoft 365.

For equipment expertise, add the records that describe the work itself. A drive repair completed last year sits in the service system with a date, an asset and a named technician, alongside whatever skills profiles you already use.

Keep each suggestion clear about what it shows

The ontology keeps four kinds of evidence apart:

• Project participation: a documented role on a project, with dates and a supporting record.
• Inferred skill: a software suggestion from available signals, labelled as an inference.
• Employee-confirmed skill: a skill the employee has confirmed, stored separately from an inference.
• Professional qualification: a credential whose scope and validity have been verified.

Those distinctions show whether a colleague diagnosed the fault, coordinated the visit or approved the budget. Recorded experience points to a useful conversation; verified qualifications and current authorization govern assignment, which stays with the responsible manager.

What changes for a project lead looking for experience?

Consider a Taiwan manufacturer preparing a packaging-line retrofit, as an illustrative scenario. The lead needs someone who has handled intermittent drive faults on that equipment family.

Before: the lead messages familiar colleagues, searches project folders and checks the maintenance system separately.

After: the lead searches for the equipment family and the fault. A connector retrieves the completed service records the lead is authorized to see. The ontology links each diagnostic responsibility to an employee and each asset to its equipment family. The directory returns colleagues with the job date, asset reference and recorded role, and marks results that show only participation. The lead calls the right person; the line manager decides any assignment.

The question shifts from “Who might know?” to “Who handled this work, and what did they do?”

FAIS models those relationships and connects the retrieval automation to them while your service and project systems stay in place. Each new source needs a connector, reliable identity matching and permission checks.

How do you test the directory on one team?

An expertise-directory pilot works best on real requests in one engineering or service team, compared with asking around. Measure:

• time from request to a useful contact;
• how often requesters rate the suggested colleague as relevant;
• whether each cited record supports the stated responsibility;
• how many corrections employees submit, and whether they change later results.

Treat a missing record as a coverage gap, not a judgement on ability, and let employees correct mistaken associations. Microsoft’s Skills transparency documentation follows similar logic: users can confirm, reject and add skills, and Microsoft acknowledges that AI may misinterpret a person’s role or activity.

Where to start with your own records

Your first decision is which completed-work records make a suggestion trustworthy: service jobs, commissioning reports or project assignments. Functional AI Solutions builds the ontology and automation that turn those records into a directory your team can inspect and extend to the next workflow. If your expertise lives in service history and a few people’s memories, that is the conversation worth having.