Research and analysis.
Essays and briefings on applied AI, software economics, private markets, and the operating work behind adoption.
The Benchmarks Were Lying. We Already Knew.
DeepSWE finally measures coding agents against long-horizon engineering work, and the results match what forward-deployed teams see in real codebases.
Are SaaS Haircuts Too Short?
The SaaSpocalypse priced agentic substitution extremely loudly and the cyber-risk repricing at roughly zero. That second variable is about to matter — and the contractual risk-transfer baked into enterprise SaaS should re-rate upward, not down.