Archive
Content on evals, review gates, auditability, permission boundaries, and the governance of AI systems.
Topics
Topic pages create a stronger internal link graph and help readers move from broad themes into the relevant stories.
Topic hubs
These hubs concentrate internal links around the themes most likely to earn repeat visits and long-tail discovery.
Stablecoins & Payments
A reading hub on stablecoin rails, payment networks, treasury workflows, and the compliance layer that now decides who captures the margin.
Agents & Governance
A reading hub on MCP, enterprise deployment, autonomous coding, evals, and the safety layer that keeps agent systems usable in production.
Power Markets & Data Centres
A reading hub on PJM capacity pricing, data-centre load growth, dedicated generation, and the macro effects of energy shocks.
Story index
6 stories
A new OpenReview benchmark pushes agents into game-engine tasks with code, visuals, and assets in one loop. The result is sobering: the best baseline solves only 49% of tasks.
OpenAI's Codex safety notes are notable because they focus on approvals, network policy, and logs rather than raw coding benchmarks. That is what production agent deployment looks like when risk is taken seriously.
Anthropic's terminal-native agent does not just assist developers - it completes software engineering tasks end to end: cloning repositories, writing tests, fixing CI pipelines, and opening pull requests.
When one model family controls more enterprise API spend than the incumbent that invented the category, the competitive dynamics of AI have structurally changed - and the reason is not just benchmarks.
The Frontier Suite matters because it turns enterprise AI into a familiar budget line: productivity software with identity, security, and oversight already embedded.
The acquisition signals that agent quality is no longer judged only by fluency - it is judged by whether organisations can test, document, and govern failure before deployment.