If your AI tool needs onboarding, do not launch until the first run is guided

A lot of AI launches fail in the first five minutes, not the first five days. User signs up. Empty state. Chat box. No example. No sample file. No "try this prompt." They bounce and you blame the board.

I run EarlyHunt (https://earlyhunt.com) for weekly AI attention, so I see this a lot. Hunters will try a product once. If the first run feels like homework, they leave a polite silence. Silence looks like "distribution does not work." Usually it is onboarding.

What I force myself to ship before any EarlyHunt week now:

- One sample input already loaded

- One suggested first action that produces a visible result in under a minute

- One sentence on the empty state that says what good looks like

- A short loom if the UI still needs a tour

EarlyHunt starts Monday 8:00am UTC and stays a week. Free and Premium (about $19) exist. The week only helps if strangers can succeed without a call with you. Catalogs like TAAFT (BLP 77) will list you either way. A hunter board will expose the empty state in public. I would rather fix that on a quieter weekly room than on a spike day with a warm audience watching.

Best Launch Platforms (https://bestlaunchplatforms.com) is where I keep the board vs directory split written down. Use the weekly AI room for conversation after the first run works. Use the catalog layer for coverage later. Do not book either while the empty state is still a blank prompt box.

Be honest. Is your first-run guided, or are you hoping smart users will invent the workflow?