Which function should a founder hand over first?
Hand over the function that takes your afternoons and has a checkable right answer. For most founders that is reconciliation, which a Finance Controller FTE does daily with every match evidenced; first-line support in one category, which a Customer Support FTE answers from your knowledge base with a citation; or search visibility, which a Search Visibility FTE runs as a measured loop. Outbound with an SDR FTE follows once the protected account list is written.
Choose by hours recovered and evidence clarity, not by what sounds most impressive. The function you resent doing on Sunday is usually the right one.
What does an owned function cost a founder in time?
In the first week, an afternoon: connecting the tools with scoped credentials, writing the remit in the first-run wizard, and reading the Brand Brain it pre-filled. In the first month, minutes a morning on the brief and approvals, because every proposal arrives with its evidence, and half an hour a week reading send-back reasons. From the second month, less, as routine classes earn autonomy and the remit settles.
The trade is doing for approving. A founder who spent four hours matching transactions spends ten minutes approving evidenced matches, and the difference is the afternoon back.
What should a founder ask before starting?
Ask what it costs and how it stops: a plan buys a block of work metered in tokens, the smallest plan is the way to try it, and the wallet pauses everything at zero with nothing on credit. Ask what it can never do: decide about a person, quote a price, post an entry unattended, publish without you. Ask what happens when you are away: the FTE keeps proposing and nothing happens until you or a delegated approver decides, which is the safe failure mode. Ask what you keep if you stop: the audit trail and the confirmed memory, exported, and every credential revoked in one action.
On ftes.ai those are the answers, and they are visible in the workspace rather than in a sales conversation.
- Cost and stop: block of tokens, hard stop at zero
- Never: decisions about people, prices, unattended posting or publishing
- Away: proposals wait; nothing happens without a decision
- Leaving: trail and memory exported, credentials revoked
What should a founder refuse?
Refuse to hand over three functions at once; one settled role teaches you more than three muddled ones and costs less. Refuse to promote anything in the first month; the record is not there. Refuse to skip the baseline because the pilot is small; two weeks of your own hours in a sheet is what tells you whether it paid. Refuse to let the FTE reach a customer with money or a complaint; that is on the always-escalate list for a reason.
And refuse a tool that asks you to move your work into its system. A Digital FTE works inside the tools you already run.
What about the people you will hire later?
The Digital FTE does not replace the first hires; it changes what they do. The first finance hire inherits a ledger reconciled daily with a short exception list and spends their time on judgement. The first support hire inherits a knowledge base improved from tickets and handles the escalations with summaries. The first marketer inherits a measured prompt set and a cluster started from evidence. Each inherits the remit, the memory, and the record, and starts as the FTE's manager with a direct report on day one.
A function that had an owner before the hire is a function the hire can run rather than rescue.
What do the first 30 days look like from the founder's seat?
Days one to three: log your hours on the function for the baseline, honestly, even for a fortnight if you can. Days three to five: connect the tools with scoped credentials, write the remit with the three lists, and read the Brand Brain. Weeks two to four: approve with evidence in minutes each morning, send back with reasons, and read the reasons weekly. Day 30: read the FTE's own report, decide targets, remit, and whether the one routine class has earned promotion, and write five lines.
By day 30 one function has an owner and you have an afternoon back. The second function is faster because the platform re-asks nothing it already knows.
What does the platform not do for a founder?
It does not run your company; it runs one function at a time inside your tools with your approval. It does not decide anything about a customer's money, a candidate, or a price. It does not work faster by being trusted sooner; trust is earned per class from the record, and a founder who promotes early gets a step-back, not a shortcut. And it does not measure what it cannot see; the performance view says unmeasured rather than guessing.
A founder who wants a magic employee should look elsewhere. One who wants a function owned, evidenced, and approvable in minutes a morning has the mechanisms here.