• Playbook
  • Operations
  • 5 min read

Earning the First Autonomy Promotion

One class, a written standard for the clean run, bounds you can state, and a step-back you are ready to make.

Author
ftes.ai Editorial
ftes.ai Team

The short answer

Earning a first autonomy promotion means choosing the one class of action with the cheapest, most reversible errors, writing down the clean run the organization requires, setting the bounds within which the class is auto-approved, promoting it from the record at a monthly review with the decision logged, and watching its first week with the step-back ready.

Put a Digital FTE on thisEvery role proposes with evidence and reports to you.

Key takeaways

  • Promote one class, never a role: exact matches, scheduling, cold first-touch, a routine reply category.
  • Write the clean-run standard before you need it; it becomes the organization's rule.
  • Bounds make the promotion narrow: amounts, lists, categories, thresholds.
  • The decision is made at the monthly review from the record and is logged.
  • The first send-back in the promoted class steps it back the same day.

Step 1: choose the candidate class

Choose the one class of action whose errors are cheapest and most reversible and whose pattern is least ambiguous: exact-match reconciliations for a Finance Controller FTE, interview scheduling into connected calendars for a Recruiter FTE, first-touch outreach to cold prospects outside the protected list for an SDR FTE, a single routine reply category for a Customer Support FTE, public-source field fills for CRM hygiene. Never a class on the always-escalate list, and never anything on the never-automated list, which the ladder does not offer anyway.

One class. A role with five classes promoted at once has no record for any of them.

Step 2: write the clean-run standard

Before the record exists, write down what the organization will require before promoting any class: a run of a stated number of proposals in that class approved without edits or send-backs, over a stated number of weeks, reviewed at a monthly review. Choose it by the stakes of the class; a class whose errors are cheap needs a shorter run than one whose errors reach a customer. Write it in the governance note and apply it to every later promotion.

This standard is what a governance reviewer will ask for, and it is what stops promotion from being a mood.

Step 3: define the bounds

Define the bounds within which the class will be auto-approved, and state them in Setup: exact matches only, under a stated amount; scheduling only into the connected calendars, only for candidates already shortlisted; cold prospects only, never a reply thread, never a named or protected account; the one reply category only, only when a source is cited, never when the ticket mentions anything on the escalate list.

Anything outside the bounds stays a proposal that waits for a person. Bounds are what make a promotion narrow enough to be safe and specific enough to be reviewed.

  • Amount or size limits
  • Lists excluded: protected accounts, named customers
  • Categories included, and only those
  • Conditions required: a cited source, a prior approval

Step 4: promote at the monthly review, from the record

At the monthly review, open the class's approval record in Setup: the run length, the approvals, any edits, any send-backs and their reasons. If the run meets the written standard and the send-back reasons show nothing unresolved, promote the class to auto-approve within the stated bounds. The grant is logged with your name, the class, the level, the bounds, the date, and the record it rests on.

If the run falls short, do not promote, note why, and review again next month. Declining is a normal outcome.

LEVEL 0Proposes only
every action waits for a named approver
LEVEL 1Tight bounds
one class auto-approved within small limits
LEVEL 2Broader bounds
same class, wider limits, still logged
LEVEL 3Delegated class
runs unattended; every action in the trail
PROMOTED PER CLASS · FROM A CLEAN APPROVAL RECORDSTEPPED BACK IN ONE LOGGED CHANGE
Figure 1The promotion in the playbook moves one class one step, within bounds, from a clean record, and the dashed return arrow is the step-back the manager is ready to make in the first week.

Step 5: watch the first week

In the first week after promotion, read the auto-approved actions in the morning brief as if they were proposals, because they are the first actions in the class nobody reviewed before they happened. Auto-approved actions are counted separately in the approval rate, and any send-back on one, meaning an action a person on seeing it would have rejected, is flagged.

Expect nothing to go wrong, and be ready for it to. The point of the first week is that the step-back is one click away and you know where it is.

Step 6: know what steps it back, and do it without hesitation

Step the class back to gated the same day on any of four signals: a send-back in the class; a change in the inputs the class relies on, such as a new fee schedule, a rewritten knowledge base, or a migrated tool; an incident or detection that touches the class; or a change of manager. The step-back is one logged change in Setup with immediate effect. Then find the cause, fix it, and let the record rebuild.

The class earns promotion again the same way, and many managers require a longer run the second time. Both directions are deliberate, logged, and from evidence.

What does done look like?

Done is one class of action auto-approved within stated bounds, a written standard for how it got there, a logged grant a reviewer can read, a first week watched, and a manager who knows the four step-back signals and where the button is. The role's approval time has fallen for that class and nothing else has changed: every other class is gated, the always-escalate list is where it was, and the never-automated list was never on the table.

The second promotion is easier because the standard exists; the discipline of one class at a time does not change.

Frequently asked questions

Which class should be promoted first?

The one with the cheapest, most reversible errors and the clearest pattern: exact matches, interview scheduling, cold first-touch outreach, or one routine reply category.

How long a clean run is needed?

The length the organization writes down in advance, chosen by the stakes of the class, and applied to every promotion.

What are bounds?

The limits within which the class is auto-approved: amounts, excluded lists, included categories, required conditions. Outside them, proposals still wait.

What steps a class back?

A send-back in the class, a change in its inputs, an incident touching it, or a new manager. One logged change, same day.

Is declining to promote a failure?

No. It is a normal review outcome. Note why and review again next month.

  • Playbook
  • Operations
  • Autonomy ladder
  • Governance
  • Digital FTE
Further reading

Keep going.

Related guides on hiring and managing Digital FTEs.

Ready to hire your first Digital FTE?

Pick a role, connect the tools it needs, and review its first proposals this week.