Seven markets. Two in production, one in the market, four parked.
I don't build products — I test a thesis. One thesis, seven markets, two in production. AI collapses the cost of expertise, enabling solo operators to compete at organizational scale.
In production
AgenTeam
AI Platform · AI agent orchestration platform.
43 AI specialists covering GTM, RevOps, legal risk, investor storytelling, product roadmapping and UX/UI through to implementation — enough depth for a solo founder to operate like a ten-person team, running pitch decks, competitive analysis, pricing models and growth loops simultaneously. Production Claude Routines (Nightly Test Guardian, GDrive Stale Lock Sweeper) ship with a deliberate no-auto-fix policy, preserving human oversight.
Signatura
HR Tech · AI career companion.
The Waze for career management. Grew to 650 organic users within three months of launch at a 65% conversion rate, on product-led growth and word of mouth with zero paid acquisition. User research found 57% cited loneliness and uncertainty as their primary pain, which is what drove Sage's emotional-intelligence architecture and background burnout monitoring rather than a feature list. Two distinct AI personas run inside the product: Sage, an emotional and mental-support companion built specifically for the psychological demands of job searching, and Siggy, the technical support specialist available to every user type. Runs a production AI evaluation loop — a golden CV dataset for scoring consistency, behavioural feedback from 600+ users across 300+ tailored CVs, full input/output logging — and multi-provider AI routing with server-side key isolation, so no provider credentials ever reach the browser.
In the market, undecided
Numerati
EdTech · Adaptive learning platform.
Published to an App Store in mid-2026. Live, in the market, and too early to call. The next data point decides whether it joins the first group or the third.
Validated, not built
Each of these four reached the point where the market question was answered. The answer did not justify the build, so I stopped. That decision is the same one I get hired to make for other people, and it is considerably easier to make on someone else’s roadmap than on your own.
Open any of them for the signal, the verdict and what it taught me.
Nav2SeedTask management for funding rounds.Parked
- Signal
- Tested only with people I already knew. It never reached an audience outside that circle.
- Verdict
- I stopped because I could not get a read from anyone who was not already inclined to be kind about it.
- Lesson
- Friendly testers can tell you a thing works. They cannot tell you whether anyone would pay for it.
PriceWeaverB2B pricing AI and strategy.Parked
- Signal
- The build kept growing past what one person could carry alongside two products already in production.
- Verdict
- I stopped because it was more than I could chew on my own.
- Lesson
- Solo scale is a standing constraint, not a temporary one. A product that needs a team should wait for a team.
ReguLensCompliance AI automation.Parked
- Signal
- Compliance automation needs continuous attention, and the regulations it would automate are still shifting.
- Verdict
- I stopped because it needed dedicated focus aimed at a target that has not settled.
- Lesson
- Building on moving regulation means rebuilding, repeatedly. Either commit fully or wait for the ground to hold still.
JarzSmart savings concept.Parked
- Signal
- The market pushed back consistently: it read as a feature, not a standalone product.
- Verdict
- I stopped because buyers kept describing it as something that belongs inside a product rather than being one.
- Lesson
- A feature can be genuinely good and still not be a business. The test is not whether people want it — it is whether they would buy it on its own.
Why this page exists
Not to sell you any of them. AgenTeam and Signatura sell from their own sites, and the other five are not for sale at all. It is here because an advisor’s credibility rests on whether they still make the calls they advise on — and seven markets entered, two committed to, five declined is the clearest record of that I can offer.