On the harness, evaluations, the model factory, and why the most valuable dataset your firm produces is the one it currently throws away.
We ran TypeSafe’s Jev against one of our custom eval suites (~14.8k tasks). It came back 40x faster and 32x cheaper, a fraction of a point worse on accuracy, and slightly conservative on assigning confidence to its answers.
Read the evaluationWhitepaper · September 2026Zero-retention terms protect your documents. They say nothing about the learning generated every time your experts correct an AI, the asset firms are currently destroying within minutes of creating it, and the case for starting to keep it now.
Read the whitepaperCase study · September 2026A bespoke solution for credit documents, backed by evaluations designed by domain experts.
Read the case studyCredit research · May 2026Fertitta Entertainment’s $17.6bn take-private of Caesars may not trigger a Change of Control under the bond indentures. The merger agreement reveals an acquisition vehicle that appears designed to satisfy the Permitted Holder carve-out.
Read the analysisCredit research · May 2026Marquee brand, chronic execution issues, supportive sponsor. We examine ASTONM’s liquidity path, potential LME mechanics, and sponsor incentives through the 2026 funding trough.
Read the analysis