A startup reaches a point where support is consuming a founder's week. The obvious answer is to hire somebody.
They do. For four months that person asks the founders roughly fifteen questions a day, because nothing is documented and every answer lives in two people's heads.
The founders are more interrupted than before, the new person is frustrated, and by the time it works properly it has cost five months and a salary. The volume was real and the diagnosis was wrong.
What a person is genuinely better at
Judgement, difficult conversations, edge cases, relationships with important customers and noticing that something is systematically wrong.
What automation is genuinely better at
Volume, consistency, availability at all hours, and the questions where the answer is identical every time.
What early support volume actually contains
Mostly the second category. Access problems, how to do something, what a feature does, billing questions and whether something is possible.
The step that must happen either way
Writing the answers down, since neither a person nor anything else can be handed knowledge that exists only in conversation.
Test 1: how much requires a decision
Count it for a fortnight. Most founders discover the judgement calls are a small fraction of the volume and the rest is repetition.
Test 2: when does the volume arrive
If a significant share is outside working hours, a single hire does not solve it and never will.
Test 3: is the product causing it
A large share of early support volume is a design problem, and hiring somebody to absorb it permanently funds the symptom.
Test 4: is there a whole job
Many first support roles are a few hours a day, which is hard to recruit for and produces somebody underused.
Test 5: what would the hire actually do
If the honest answer is answer the same questions, that is the case for documenting and automating rather than employing.
The sequence that works
Document, automate the repetition, see what remains, and hire against that, which is the same order as choosing between hiring and automating the first admin role.
When to hire anyway
When customers are large enough to need a named person, when the judgement volume is real, or when the founders' time is worth considerably more elsewhere.
What the hire becomes afterwards
Frequently a better role than the one originally planned, since somebody handling the difficult cases rather than the repetitive ones is more useful and easier to keep.
What the volume actually says
A rising support load is evidence of growth and not evidence that a hire is due. The question is what the load consists of, and founders who list it honestly usually find the answer is not a person.
What to do about the founder who likes it
Many founders answer support because it is the most direct contact they have with customers and they are reluctant to lose it. Keeping the difficult conversations and giving away the repetition preserves exactly what is valuable about that.
Document, automate, then hire
An assistant handling the repeating questions and the access problems at any hour shows exactly what is left for a person, and the first support hire is then made against a real job rather than against a feeling of being overwhelmed. The early question set sits in early stage startups and their first customer questions, the candidate side in hiring enquiries and candidate questions at early stage, and the comparison in ai employee vs part time admin. See it working on the demo, compare on pricing, and the sector view sits under startups.