What GitHub learned when better tools made Copilot code review worse

GitHub's migration of Copilot code review to shared tools initially worsened performance. Rewriting instructions improved accuracy and reduced costs by 20%.

Paul Sawers · 8 min read · 14 Jul 2026

Not all model ‘upgrades’ are upgrades — Microsoft data shows cheaper can cost more

Microsoft data reveals that newer AI models with lower per-token costs can end up more expensive due to higher token consumption, challenging assumptions about upgrades.

Paul Sawers · 7 min read · 7 Jul 2026

Tessl Academy is live (in preview) — and there are two ways in

We just shipped the first version of Tessl Academy, a hands-on curriculum for building, evaluating, and running skills for coding agents. It's early. Two courses are up — Skill Foundations and Tuni...

Alan Pope · 3 min read · 3 Jul 2026

How context travels in a multi-agent world

Explore how context is managed in multi-agent systems, focusing on Microsoft's use of the A2A protocol to maintain coherent communication across independent AI agents.

Paul Sawers · 6 min read · 29 Jun 2026

See You at AI Engineering World's Fair 2026

Explore AI Engineering World's Fair 2026 with Tessl, featuring demos of Tessl Agent, discussions on agentic development, and the emerging field of Harness Engineering.

Jordan Sanders · 7 min read · 27 Jun 2026

How Small Can an Agent Model Get? The Nemotron Floor

The article explores the minimum model capacity needed for agent models to function effectively, highlighting the Nemotron family's performance on coding tasks.

Nicolas Fortuin, Baptiste Fernandez · 12 min read · 26 Jun 2026

Cursor's new leaderboard shows teams the most popular plugins, skills and MCPs

Cursor's new leaderboard helps teams track popular plugins, skills, and MCPs, offering insights into usage patterns and facilitating better tool management.

Paul Sawers · 5 min read · 24 Jun 2026

The new Tessl review: now you decide what "good" looks like:

The new Tessl review lets users define their own criteria for skill quality, offers agent-based accuracy, and maintains a history of review runs.

Simon Maple · 7 min read · 23 Jun 2026

AI Agent Governance: 10 Takeaways from Engineering Leaders on Agentic Development

Explore 10 insights from engineering leaders on AI agent governance, focusing on the shift from individual productivity to enterprise-level operating models.

Guy Podjarny · 26 min read · 23 Jun 2026

Why Warp is betting engineering leaders are done picking a favourite coding agent

CEO Zach Lloyd talks multi-harness orchestration, why the CFO is now in the room for AI tool discussions, and what governed autonomy actually looks like on a “factory floor.”

Paul Sawers · 17 min read · 23 Jun 2026

Open-Source Agents vs Sonnet 4.6: GLM 5.2, MiniMax M3, Kimi 2.7 and Qwen 3.7 Tested

Open-source coding models like GLM 5.2 and Qwen3.7-Plus are compared to Claude Sonnet 4.6, revealing varied performance in instruction-following and task completion.

Nicolas Fortuin, Baptiste Fernandez · 13 min read · 18 Jun 2026

How I Scan My Agent Context Across GitHub with Skill Inventory

Explore how Tessl's Skill Inventory scans GitHub to map agent skills, addressing duplication and ownership issues, and providing a clear skill estate overview.

Baptiste Fernandez · 6 min read · 17 Jun 2026