For founders

Stop being the support team for your own product

You are answering the same eight questions every week, at night, between the things only you can do.

What you will hear back

“Is this really the thing to spend time on right now?”

And the honest answer

The honest answer depends on whether you are answering the same questions repeatedly. If you are, this is an afternoon that returns a few hours a week indefinitely. If your inbound is genuinely varied, it will not help much and you should not bother yet.

The case, in the order it lands

01
An afternoon, not a project
Crawl the site, paste two lines, set a cap. There is no integration, no data model and nothing to maintain afterwards.
02
It answers at the hours you do not
Self-serve signups happen in the evening from timezones you do not staff. Those are the questions that currently wait for you.
03
It cannot promise what you have not built
It only knows your published content, so it cannot invent a roadmap in front of a prospect. Every request it could not answer is recorded instead.

Send this to whoever has to approve it

It reads our site, answers the repetitive questions with a link to the page it came from, and refuses anything it cannot source. Setup is a crawl and two lines of script, with a hard spend cap set in advance.

Support is the work that expands to fill everything

Early on, answering customers personally is an advantage. Past a certain volume it becomes the thing that stops you building, and there is no obvious moment where hiring for it is justified.

The same questions arrive weekly and you answer them personally
Support interrupts deep work at unpredictable intervals
Hiring for support cannot be justified at current volume
Enquiries at nights and weekends either wait or eat your evening

Get the repetitive share back

Point it at your site and docs and the questions you have answered a hundred times answer themselves, with a citation, at any hour. What reaches you is the genuinely new question — which is the one worth your attention anyway.

Live the same afternoon: A crawl and two lines of script. There is no project behind it.
The repetitive share, gone: The eight questions you always answer stop reaching you.
Captured with context: What needs you arrives described, so you reply once rather than three times.
A cost ceiling you set: Hard daily caps, so a spike cannot produce a bill you did not plan for.

It will not promise things you have not built

The failure mode that would actually hurt you is a bot telling an evaluator you support something you do not. Because it answers only from your published material, it cannot invent a feature — and it records the request, so you find out what people expected instead of losing them silently.

Read what people actually ask

The ranked list of questions your content does not answer is unusually good product feedback at this stage. It is your market telling you, in their own words, what is unclear about what you have built.

Before you take this to anyone

Do not install this before you have anything to point it at

If your site is three pages and a waitlist form, there is nothing to retrieve from and the assistant will decline nearly everything. That is the correct behaviour and a bad experience. Write the FAQ first; the tool is worth adding the week after.

Coverage decides quality far more than any setting does
A hard daily cap means a traffic spike cannot produce a surprise bill
Nothing enters your stack — removing it is deleting two lines
Refusals become a ranked list of what to write next

A founder at a big studio window with a coffee, looking out and thinking in bright morning light.

What you will want to know first

Short answers. If yours is not here, the assistant on this page will try it — and tell you honestly if it cannot.

Is this worth it before we have much traffic?

Honestly, it depends on whether you have enough published content for it to answer from. With a thin site it will refuse a lot — though every one of those refusals is logged, which is then a fairly direct instruction on what to write. The clearest signal you are ready is answering the same question for the third time.

How much time does setup actually take?

A crawl and a snippet — most people are live the same afternoon. The ongoing work is reading the gap list occasionally and writing the missing page, which is work worth doing regardless.

What if it says something wrong?

It answers from your content and links the source, so a wrong answer means your page is wrong — which is fixable and worth knowing. It cannot invent a policy, because it has nothing to invent one from.

All solutions