I name MindWave, Super AI, and Vibe AI publicly. I write the postmortems with dollar amounts, named decisions, and dates. I link them from the homepage. I link them from every relevant blog post. The default founder advice is to keep failures vague, optimised for the next conversation. I do the opposite, and the bet has paid off in concrete ways. This post is the case for that bet.

The hub at three startups, three shutdowns walks through what each company was and why it failed. The individual postmortems live at mental-health-platform-postmortem, super-ai-postmortem, and vibe-ai-postmortem. This post is the meta-argument: why I write them under the company names instead of laundering them into "two prior startups in the AI space".

Why specificity beats vagueness

"I had two prior startups that did not pan out" is a sentence that gets nodded at and forgotten. "MindWave shut down in November 2022 because the unit economics on a 1.3x improvement to a generic mental health chatbot did not clear the 10x test in a category dominated by free incumbents" is a sentence that either survives scrutiny or does not. The first sentence is decoration; the second is testable.

Specificity does three things vagueness cannot. It signals that the founder has done a postmortem with real diagnostic depth rather than vibe-based regret. It dates the failure, which lets the audience calibrate how recent the lesson is. And it creates an artifact that can be linked, shared, and quoted, which matters for SEO, for investor due diligence, and for the next founder who is trying to avoid the same mistake.

The most useful response I have ever gotten to a postmortem was a four-line email from another founder saying "I am about to make this same mistake; can we talk?" That email does not happen if my failures are described as "two prior ventures". It happens because the named version is searchable.

The hiring and fundraising counter

The objection most founders raise to naming failures is that it will hurt hiring or fundraising. The empirical answer in my own data is: less than founders fear, and the cases where it does hurt are the cases that filter out poorly-aligned counterparties anyway.

On hiring, of the senior IC and engineering manager conversations I have had in the last eighteen months, exactly one filtered me out for a named past failure. That conversation also filtered me out for not having a Stanford-or-equivalent pedigree, which suggests the failure was not the binding constraint. Every other conversation either did not raise the failures or raised them constructively. CB Insights' founder-survey data on second-time founders backs this up: the penalty for a known failure is much smaller than the asymmetry founders assume.

On fundraising, the case is similar. Investors who want a polished narrative are investors who would have been a bad fit anyway. The investors who responded to the Vibe AI shutdown were the ones who specifically said they wanted the unedited version. The selection effect is in the right direction: naming filters out the people who would not have backed the founder I am trying to be.

Outcomes after naming failures, last 18 months (n=43) Conversations that opened doors ~52% (22) Neutral, no impact either way ~42% (18) Filtered me out for the failure ~6% (3) Sample: investor pitches, hiring conversations, customer discovery calls. Self-reported. Source: personal conversation log, Sept 2024 to Apr 2026.
Outcomes from 43 named-failure conversations across pitches, hiring, and customer discovery. The penalty is small; the upside is asymmetric.

Owning the search result

Future investors will Google. Future hires will Google. Future customers will Google. The question is not whether the failures will surface; it is whether the version they surface is the one you wrote or the one a forum, competitor, or disgruntled ex-employee wrote. The former is a postmortem with context. The latter is whatever Google's algorithm surfaces from the public sites that talk about your name.

I write the named postmortems specifically so they outrank anything else for "MindWave shutdown" or "Vibe AI postmortem". As of the latest analytics check, the gravity.fast pages rank in position one for those queries on Google in India and the US, ahead of any forum thread or news mention. That is the SEO outcome of naming. The narrative outcome is more important: when someone reads the postmortem first, they read it with the dollar amounts and the framework intact, not with the framing of an outsider.

Vs build-in-public

Naming failures is not the same as build-in-public, even though they share aesthetics. Build-in-public is a marketing tactic optimised for momentum: ship updates, post graphs, celebrate wins. Most founder accounts on X or LinkedIn doing build-in-public skip the hard half. Revenue lines that did not move are not posted. Pivots that look like indecision are reframed as iteration. The narrative is curated.

The transparency thesis is the unglamorous half. Postmortems with dollar amounts. Decisions that did not work. The mental health post at founder-mental-health-after-three-shutdowns and the financial postmortem at the-honest-cost-of-three-shutdowns exist because build-in-public alone would let me skip both. The honest version includes the half that does not look good in a celebratory social post.

The limits of the thesis

The transparency thesis applies to the founder's own decisions and the company name. It does not apply to co-founders, investors, customers, or employees who have not consented to being named. The line is: facts you owned, decisions you made, dollars you spent. Everything else stays out unless the other party explicitly agrees.

This matters because vague references in a postmortem usually mean someone else's information that is not yours to publish. "We had a co-founder dispute" is a sentence that names you and protects them; "X did Y in March" is a sentence that names them. Use the first form. The transparency thesis is about your own accountability, not about settling scores or breaking other people's privacy.

The thesis also does not apply to live competitive information that benefits competitors more than it benefits the audience. Pricing tests in flight, security incidents under investigation, customer-specific information covered by NDA. Those wait until they are no longer live.

What Gravity ships as proof

Gravity itself is the live test of the thesis. The founder section of /about names all three prior companies. The blog hub at three startups, three shutdowns sits at the top of the cluster. The economics post at economics of bootstrapped AI agents shows the actual unit-economics math. None of that is hedged. If the thesis is wrong, the cost is paid by Gravity directly.

The bet is that named honesty compounds. The first three startups bought the right to write this post. The next three thousand blog readers are the test. If you have a failure you have been laundering into "a prior venture" because of the fear that surfacing the name will cost you, the answer in my own data is that the cost is much smaller than the upside. Email me at the top of /contact if you want to talk specifics.

Frequently asked questions

Why name failed startups in public?

Naming the failure compounds credibility because it forces a specific accountability that vague references do not. When I write MindWave, Super AI, and Vibe AI by name, the postmortems are testable, the lessons are dated, and a future founder or customer can verify that the framework is grounded in real attempts. Vague failure language is a credibility tax.

Does naming failures hurt your hiring or fundraising?

Less than founders fear and more than founders admit. The hiring conversations I lose are with employers who would penalise the failure regardless; the ones I win are with employers who specifically wanted the founder mode. Fundraising is similar: investors who want a polished narrative do not back honest founders. The selection effect is in the right direction.

What about future investors who Google the names?

They will Google. They should Google. The named postmortem at gravity.fast slash blog ranks above any anonymous founder forum chatter, which means the version they find first is the version I wrote. Owning the failure narrative on the open web is a smaller risk than letting it be told by someone else with less context and worse motivations.

Is build-in-public the same as naming failures?

No. Build-in-public is a marketing tactic that often skips the hard parts. Naming failures is the unglamorous half: revenue lines that did not move, decisions that did not work, postmortems with dollar amounts. The honest version of build-in-public includes both halves. Most founder accounts on social media skip the second half completely.

When should a founder not name a failure?

When the failure involves co-founders, investors, or customers who have not consented to being named, name only your own role. The transparency thesis applies to the founder's own decisions and the company name, not to other people's private information. The line is: facts you owned, decisions you made, dollars you spent. Everything else stays out.

Three takeaways before you close this tab

Sources