Skip to the work

Independent AI product studio · Much more in the pipeline

We build AI products that refuse to make things up.

Golf operations, market research, food heritage — three unrelated worlds, one discipline. Every number on the screen is a real number, and where the evidence is thin, the app says so.

The work

Built to be
believed.

Three products in build, each for a different market, and each defined by the thing it refuses to do. None has opened to the public yet — that is deliberate, and what is below is what they already do.

TourneyHQ

From Registration to Recognition.

All-in-one golf tournament and league management — from the draw to the payout. Organisers set up on a desktop; players carry the app onto the course. Scoring runs on a tested engine kept separate from the screens, so a leaderboard and a bracket can never disagree about the same cut.

Plan it · Pair it · Play it · Crown it

Flight 1 · Round 3 Sample
PairThruNetPts
Rourke / Vasil18−47.5
Okafor / Lund18−16.0
Haines / Cho16+24.5
Bianchi / Reyes16+34.0
Novak / Adeyemi3.5

Illustration. A dash means no card has been returned — not a zero.

Runs on
Web and phone. One codebase, gated by role.
Money
Recorded, never moved. Clubs leave the cash with the shop; societies can run a float or a split.
Won't do
Call a handicap it worked out from returned cards a WHS Index. Store a forfeit that counts for nothing.

Ajent Signals

Buy the dip. Short the rip. Judge it by the real record.

Fades overstretched markets across indices, crypto and commodities — an exact entry, stop and target, exiting as price snaps back to the mean. Every signal paper-trades itself to a virtual account, win or lose, so the record is real.

Trend-following was built, tested and thrown away at a 30% win rate. Markets that lost money are named and excluded. With no feed, the app says no trade rather than inventing a confident one.

Signal card Structure
ES · 15m Long
RSI(2) 8.4
Stop 2 × ATR
Exit RSI 50
HSI · 15m No trade

No live feed. The app shows nothing rather than a guess.

Shape of the card, not a live quote. Paper trading and research only — not financial advice.

Runs on
Phone and web. Free, no card, entirely virtual money.
Covers
ES · NQ · RTY · YM · SX5E · DAX · TSX · N225 · XJO · HSI · BTC · ETH · SI · CL
Won't do
Fabricate a win rate, a P&L or a testimonial. Ship an overfit tweak because it backtests well.

WikiFoodia

The community decides what's authentic.

A free atlas of the world's dishes, built from open and citable sources and readable in a dozen languages. Its job is to hold traditions still long enough to be recognised, in a world that rewrites recipes for engagement.

So a dish's name, its traditional ingredients and its equipment are never translated — a gloss sits beside the original word. Modern substitutes get their own labelled box. Fusion dishes carry no authenticity score at all.

Record · reading in French fr

Mole negro oaxaqueño

Nom conservé · Oaxaca, Mexique

  • Chilhuacle negro piment séché d'Oaxaca
  • Hoja de aguacate feuille d'avocatier, grillée
  • Metate pierre à moudre

Substitution moderne
Un mixeur remplace souvent le metate. Noté ici — jamais dans la liste authentique.

Illustration. The reader's language gets a gloss, not a replacement.

Runs on
Phone-first web. The same build serves web and the stores.
Costs
Free to read, near-free to run. Donations would go to an Open Collective, which publishes its ledger.
Won't do
Rank a cook's video by view count. Substitute an ingredient to suit a filter.

Standing rules

What every
app refuses.

One constraint each, written into the data layer rather than the interface — so it holds whoever opens the screen next.

No fabricated numbers.

Every figure shown is a real backtest result or the live paper record. Nothing else is displayed.

Ajent Signals

Popularity is not authority.

How often a dish is viewed and how close it sits to its tradition are separate fields, and never mix.

WikiFoodia

Record it, never move it.

The app settles who won and what they are owed. The shop still takes the money and pays it out.

TourneyHQ

The pipeline

What comes
next.

What is being finished before each product opens up, stated the way the apps state everything else. No roadmap theatre, and no dates the studio cannot keep.

Building
Shared costs for a golf society Trip expenses, sweeps and payouts settled on a phone at the bar.
TourneyHQ
Building
Community contributions, opening carefully Readers proposing and confirming records, one confirmation per person. Read-only until the identity check is provably sound.
WikiFoodia
Testing
A forward-tested record, not a backtest The faster book runs live on paper until the real record stands on its own. Whatever it says is what gets published.
Ajent Signals
Next
More apps, in new markets Several are in design. Nothing is announced here until it is real — but the studio is only getting started.
AjAi Labs

The studio

How we
actually work.

Four rules, each written down after something went wrong or nearly did. Each has cost a feature at least once.

01

One rule, read in one place.

Almost every serious defect has been a second implementation quietly disagreeing with the first. So a rule gets one home, and everything asks it.

02

Look at the screen.

Tests protect the engine. Only opening the app on a phone protects the person using it — that is where every real problem has been found.

03

Say when the evidence is thin.

A short sample is labelled provisional. An empty screen names what is missing. Confidence is a claim like any other, and needs the same backing.

04

AI where it earns its place.

Drafting a write-up, reading a scorecard photo, translating the page somebody actually opened. Never a model's guess wearing the clothes of a fact.

Get in touch

Bring a hard, small problem.

The studio builds AI-native products end to end — design, engine, app, deployment. It works best where getting the answer right actually matters.

Ajent Signals is research and paper-trading software. Nothing on this site or in that app is financial advice, and past results do not predict future results.