Our research, not a wrapper
The system on your desk is the one our lab improves. Every gain lands here without you changing a thing.
We build analysis systems that improve their own method, and put them to work on live questions. Our work spans self-improving agents, evaluation under noisy signal, and financial data at depth.
Founded and backed by people from KTH, SSE, Jane Street, Carnegie, SEB and McKinsey.
We work at the cutting edge of AI research, on self-improving models: systems that read their own work, revise how they think, and keep what holds up against the market and reality.
The system on your desk is the one our lab improves. Every gain lands here without you changing a thing.
Specialised agents plan, run and review the work in parallel. A question that would take an analyst a day is broken apart, distributed and reassembled, and every step it took stays visible.
Live equities, filings, rates, commodities and macro across 28 jurisdictions, licensed at source rather than scraped. Open any figure and see the filing, the series and the timestamp behind it.
A multi-step brief runs from the first question through to a finished document, built natively in Excel, Word or PowerPoint on your own template. Sibel works from your methods and your prior notes, without your data entering a model.