R&D operations

What does a fractional AI operator actually deliver in the first 30/60/90 days?

Concrete deliverables at day 30, 60, and 90 of a fractional AI operator engagement, what done looks like at each milestone, and how progress gets reported.

The first question buyers usually don’t ask is: what will we actually have in three months? Here is what a well-run fractional AI operator engagement looks like, milestone by milestone.

The exact shape depends on the engagement model and the starting point, but the underlying progression is consistent: the first month establishes truth, the second translates it into a working structure, and the third proves it in motion.

Day 1 to 30: establishing what is actually true

The first month is diagnostic. Not in the consulting sense of “we are producing a report,” but in the operational sense of needing to know what is actually going on before touching anything.

This means getting into the real state of the project: the actual code or model performance if there is a technical workstream, the real vendor relationship and what the friction is, the genuine priorities of the people closest to the work, and the gap between what leadership believes is happening and what is actually happening. That gap is usually larger than anyone expected.

Typical day-30 deliverables:

A written project state assessment: what exists, what the performance baselines are, what the real blockers are, and where the risk is. This is not a slide deck for a board presentation. It is a working document that the team can disagree with and annotate.

A scoped engagement plan for the next 60 days: what gets fixed first and why, what decisions need to be made and by whom, and what the operator’s focus will be week to week.

First conversations with key vendors or external partners, with an honest read on the state of those relationships.

What “done” looks like: leadership has a clear-eyed view of where the project actually is, and there is an agreed operating plan for what comes next.

Day 31 to 60: building the structure

The second month is structural. The diagnosis from month one is translated into a running delivery model: governance, reporting, and the working habits that determine whether the project runs well or poorly.

For a Business Finland R&D project, this typically means getting the work package structure and reporting cadence into a form that is audit-ready. For a vendor-led AI project, it means getting the vendor relationship onto an accountability footing with clear milestones and a functioning escalation path.

Typical day-60 deliverables:

A risk register: not an elaborate document, but a live list of the things that could kill the project and what is being done about each one.

A vendor accountability framework if vendors are involved: specific milestones, acceptance criteria, and a clear process for what happens when milestones are not met.

A reporting rhythm agreed with the client: how the operator communicates progress, what the format is, and how often.

For funded projects: at least one reporting-period milestone documented in a form that will satisfy audit.

What “done” looks like: the project has operating structure. There is a register of risks. Vendors are on milestones. The operator is not doing everything, but they are accountable for the parts they own.

Day 61 to 90: proving it in motion

The third month is operational. The structure built in month two is running, and the question is whether it is working. The operator’s role shifts from building to governing: holding the structure, catching the drift, escalating what needs to be escalated, and making decisions the team cannot make without them.

Typical day-90 deliverables:

A first milestone review: what was committed at the 60-day mark, what was delivered, what the gap is, and what it means for the next period.

An updated project plan with revised timelines if the first 90 days revealed surprises.

A frank assessment of what needs to change in month four: this might be vendor performance, internal resource allocation, or scope.

What “done” looks like: the project is moving, the risks are visible, the client knows what is going on without having to chase, and the operator has earned enough context and trust to make the decisions the project needs made.

How progress is reported

Operators who disappear for three weeks and then surface with a status update are doing it wrong. The reporting model that works is lightweight and frequent: a brief written update at the end of each week that covers what moved, what is blocked, and what needs a decision. Monthly reviews add depth. The goal is that the client always knows what the operator is working on and never has to wonder whether the engagement is generating value.

This is also the mechanism for surfacing scope drift and renegotiating if the engagement needs to expand or change direction. The weekly update is where that conversation happens before it becomes an invoice dispute.