We Ran an AI Audit on Our Own Portfolio — Here's What It Found
AIQ Industries builds fast and ships often, which means we also accumulate ventures faster than any one person can properly evaluate. This week we tried something most companies never do out loud: we pointed a fleet of AI research agents at our own portfolio, gave them full access to the actual code (not the pitch decks), asked them to research the real competitive landscape for each product, and told them explicitly not to be diplomatic.
Here’s what came back.
The headline
Across six active ventures, only one thing came out unambiguously well-timed and worth continued investment: an internal tool built to solve our own problem, never intended as a product in the first place. Everything else needed either a real pivot, a pause, or an honest retirement.
That’s a strange thing to publish. It’s also the most useful thing we could tell you, because it’s true, and because “we build fast” only means something if we’re also willing to look hard at what we’ve built.
What we actually found
Two of our Web3 products had marketing ahead of their code. Our own site described one as “live on mainnet” when the real repository showed a promising hackathon prototype still running on a test network. Another was listed as “in active development” with no code, repository, or spec behind it anywhere. Both are fixed now — you’ll see honest status labels instead, because a claim a technical visitor can disprove in five minutes is worse than no claim at all.
A beautifully-built real estate demo is sitting on top of a business plan that requires becoming a securities issuer. The engineering is genuinely good. The business model, as scoped, requires legal infrastructure a small independent team doesn’t have yet. So we’re separating the two: shipping the parts that don’t require a securities offering now, and treating real tokenization as a distinct, later phase gated on the right regulated partner.
Three separate attempts at the same “AI-powered project management” idea, built independently over time, had never been used by anyone outside our own team. That’s not a signal to try a fourth time with more polish — it’s a signal the idea wasn’t the bottleneck. It stays an internal tool.
A blockchain dashboard and a warehouse UI were both genuinely well-crafted front-ends with no realistic path to standing alone as products in categories already served by free or entrenched incumbents. The craft in both is worth keeping — as reusable components and as sales material for our build-services division — just not as standalone bets.
What changes because of this
Near-term, we’re doing less venture-building and more of three things: writing honestly about what we’re building and why (this post is the first of that), doing the client-facing work we’re actually good at, and turning a few of the things we already validated — including some of the tools mentioned above — into things we can give away or sell directly, instead of letting them sit half-finished in a repository.
If that sounds like the kind of company you’d want building something for you, or the kind of process you want to watch happen in public, we’re just getting started publishing it. More on each of these individually soon.