Engineered intelligence
Outthink Human Intelligence.
Now with VEOX you can invent algorithms that otherwise humans can't — continuously engineering better intelligence from your data.
VEOX is a deep-tech company. We build machines that search for algorithms instead of running the ones people wrote — and jAIn is the first one we have pointed at a problem.
What is VEOX
One engine. Pointed at one problem.
VEOX builds engineered intelligence: systems that search for algorithms rather than execute the ones a person writes. RecurEvolv is the synthesis engine underneath them, and it is general — what changes from one application to the next is the space it searches and what it hands back.
Today we have pointed it at one thing, on purpose. jAIn searches for predictive models over your data: you give it a table and the column you want predicted, and it proposes, scores and discards candidates until it finds one that beats what you already shipped. It hands back runnable code you own — usually more accurate than your incumbent, and cheaper to serve on every request.
How it works · jAIn
One table in. One model out.
Say you run churn for a telco. Your team built a gradient-boosted tree, tuned it over three weeks, and shipped it the day the quarter closed. It has sat at that score ever since.
jAIn takes the same table. It proposes candidate pipelines across preprocessing, model family, hyperparameters and ensembling as one joint search — and scores each candidate twice: held-out accuracy under your metric, and what it costs to serve. Wide ensembles that win on accuracy alone get distilled down or dropped.
What comes back is a model, not an API. Plain code and weights, with the full log of what it tried and rejected.
The category
Engineered intelligence.
There's a name for what happens when you stop hand-building models and start searching for them. We've been calling it engineered intelligence: a system that proposes candidate algorithms, scores them against your data, keeps what wins, and doesn't get tired.
The distinction that matters is where the intelligence sits. In every previous generation of tooling, a person decided what to try and the machine executed it faster. Here, the machine decides what to try. A person sets the objective and the budget.
Engineered intelligence is a young category and we're not going to pretend otherwise. What we can do is publish the record, including the runs we lose, and let you check it. The full definition states the mechanism, and the methodology states how the record will be measured — published before the results exist.
One engine, several outputs
jAIn is the first thing we've shipped. It won't be the last.
Everything below runs on the same search. What changes is the space it searches and what it hands back.
The record
There isn't one yet, and we won't invent one.
jAIn doesn't yet score itself against the model you already have, and doesn't write a result you could audit, so there is no run log to publish and nothing here we would stand behind. The research brief under Resources is a study and reads as one; it is not a record of what the product does on your data. That record arrives with the Research Preview, as receipts rather than as four big numbers.
VEOX Research
Read the argument before you take the call.
Written for someone who'll check the method. Start with the summary — it reads on the site, and links to the full paper.
Unnatural Selection
The technical brief in nine minutes: why the ceiling is structural, how the results were sealed before the search started, and what turned up on seven hard datasets.
The 2026 AI Mandate
The shorter one, for the meeting you have to run. What engineered intelligence changes about where modelling effort belongs and who does it.
Who's building this
Small on purpose.
VEOX was founded by Jepson Taylor — ex-quant, ex-HireVue, holder of 14 AI patents. He sold his deep-learning company Zeff.ai to DataRobot, then held C-level roles at DataRobot and Dataiku. He has watched two generations of AutoML get sold and knows exactly what this isn't.
The engineers behind him came from Intel, Micron, Cisco, VMware and Adobe. We work with a deliberately small number of design partners at a time, and we're selective about fit — if your problem is one the search won't help with, we'd rather say so on the first call.
See what the search finds.Start with one table — your features, and the outcome you want predicted.
Free while it is a preview: no SLA, fair-use quotas, and results and the service can change. Preview terms →
Evaluating VEOX for a team? There’s a page for your security review →