The businesses that deploy their first Digital FTE successfully all have one thing in common: they started with a role that was obviously wrong for a human to be doing.

Not a strategic role. Not a creative role. A role where the person spends most of their time doing the same predictable tasks, in the same predictable sequence, generating the same predictable outputs. A role where a bad week means errors. A role where a person calling in sick creates a visible problem for the business.

That role is where you start. Here is the seven-day process.

Day 1: Role Selection and Audit

List every role in your business where someone spends more than 10 hours per week on tasks that follow consistent patterns. Prioritise by:

  • Volume: More repetitions per week = higher priority
  • Cost: Higher-cost roles = faster ROI
  • Downstream impact: Errors or delays in this role cause problems elsewhere = higher urgency
  • Definability: The clearer you can define "correct output," the better the Digital FTE performs

Pick the highest-scoring role. This is your first Digital FTE candidate.

Day 2: Task Decomposition

Shadow the person in the role for one day, or interview them in detail. Map every task they perform: what triggers it, what they do, what decision they make, what they produce, where it goes.

Be specific. Do not write "processes customer requests." Write: "receives email with customer inquiry → opens CRM → checks account → determines if standard or non-standard case → if standard, selects template → customises with account details → sends → logs in CRM as resolved."

The more granular the map, the better the Digital FTE specification.

Day 3: Exception Mapping

Every role has exceptions — cases that do not fit the standard pattern. Document every exception the person in the role currently handles, how often it occurs, and how they resolve it.

Exceptions fall into three categories:

  • Automatable exceptions: Different inputs, same rule-based resolution
  • Human-required exceptions: Require judgment, empathy, or authority the AI should not have
  • Rare outliers: Occur less than once per month — document and design a clean escalation path

Day 4: Success Definition

Define what "completed correctly" looks like for every task type. This is the specification the Digital FTE is built to. Without it, there is no way to evaluate whether the AI is performing correctly — or to know when to intervene.

Success definitions should be measurable: "Invoice processed, matched to PO within tolerance, posted to accounting system within 2 minutes of receipt" — not "invoices are handled well."

Day 5: Integration Mapping

List every system the role touches: CRM, email, accounting software, calendar, ticketing system, communication tools. The Digital FTE needs access to these systems. Map the specific actions it needs to perform in each and the permissions required.

Day 6: Build and Configuration

With the full specification from Days 1–5, the Digital FTE is built: AI logic, integrations, exception handling, escalation paths, and logging. This is the technical build phase — typically completed in 1–3 days for a well-defined role.

Day 7: Supervised Launch

The Digital FTE goes live with monitoring. The first day is not unsupervised — every output is reviewed against the success definition from Day 4. Exceptions are catalogued. Fine-tuning happens in real time.

The first Digital FTE is never the last. Once you have one running and can see what it produces, the next three candidates become obvious immediately.

What Happens After Day 7

Days 8–30 are the optimisation phase. The exception rate — tasks requiring human intervention — should drop steadily. Most well-specified Digital FTEs reach autonomous operation (less than 10% exception rate) within the first month.

After Month 1, you have a fully operational AI employee running a business role permanently, at consistent quality, around the clock. That is the foundation. Most businesses then deploy two to three more within the following quarter.