attimet

attimet's pitch is unusually direct about being a thesis rather than a product. The homepage frames the whole company around a single idea — Isaac Asimov's Psychohistory, the fictional science of predicting large-scale human behaviour through mathematics — and reduces it to three numbered ambitions: build the Prime Radiant, apply it to financial markets, manage the world's assets. That is the entire published offering. There is no product screen, no model documentation, no dataset description, no benchmark, and no customer story on the page; the two interactive elements are 'Join us' and 'Contact us', plus a LinkedIn link. What the site does establish is team credibility. Listing experience from Optiver and DRW — both recognised quantitative trading firms — alongside Y Combinator and Yale signals people who have worked with market microstructure, execution, and risk at firms where being wrong is expensive. For a company whose stated plan ends in managing assets, that background is the most concrete asset attimet has to show a prospective partner. Where this fits: if you are an institutional allocator, a hedge fund research lead, or a quantitative investor who takes meetings on concept-stage ideas, attimet is a legitimate conversation. The roadmap framing also tells you what they are not doing — there is no claim of a shipped forecasting product, and no published methodology to critique. Evaluate them on the team, the modelling approach as described in conversation, and what access they offer early partners. Where it does not fit: anyone who needs a deployable tool. If your requirement is market forecasts you can backtest against your own data this quarter, attimet cannot currently satisfy it, because nothing testable is published. Retail traders looking for signals have nothing to work with here at all — the roadmap's third step is asset management, not consumer tooling. The honest summary is that attimet should be evaluated as an early-stage research bet, not as software. Do your diligence in the room: ask what has actually been built, what data the model is trained or calibrated on, what the Prime Radiant produces today as output, and what a design partner receives in return for their time. The homepage gives you the ambition and the résumés; everything else has to come from the team directly.