·22 min read
The shape of your AI agent team points to where it can break
Agent teams and the search behind a model can both be drawn as shapes. Three 2026 papers point to where a team can break: a hub, a single path, a late check. Seven tests.
- ·18 min read
What could go wrong with this change? Anthropic's code review post, and the video about it
Anthropic says its Claude Code team moved most code reading to machines and invested in fast undo. We read the post and a video about it and list the gaps.
- ·16 min read
Your AI agent is a text predictor in a loop. That is the floor, not the ceiling
How a 2026 AI agent is built, what has grown out of that design, why it forgets, guesses and agrees too easily, and eight habits for working with those limits.
- ·13 min read
A first build and six fix rounds: who checks an agent's work, and who writes down the verdict
OpenAI says its grader faked inputs in training. A preprint (abstract only, not peer reviewed) puts the best review setup at 71.9% of known bugs in its top five, in one benchmark.
- ·16 min read
Anthropic left git for its own code. What carries over to a team with ordinary repositories
Anthropic built its own source control after merges grew more than tenfold. What carries over to teams with agents writing code, what does not, and what the post leaves out.
- ·15 min read
AI agents can now get accounts and inboxes, and some designs name a human owner
Four AI Engineer World's Fair talks checked against primary sources: agents can register, get inboxes and run in sandboxes, and some designs name a human owner.
- ·13 min read
What our agents learn is a draft until we test it
Two AI Engineer workshops show how the check shapes what you keep. A newer 2026 recipe tests a saved lesson outside the agent. What holds, and what we do.
- ·10 min read
Which AI model is really answering? A new test is a drift alarm, not a lie detector
Tsinghua researchers can often tell which AI model sits behind an agent from how it works. What the paper shows, what it doesn't, and how to use the idea at any level.
- ·10 min read
Self-improving AI agents overfit their own tests. Two September papers each show a brake that works. Neither tests them together.
Google and Oxford researchers each published a way to keep self-improving AI agents from overfitting. What the papers do, what a popular video got wrong, and what EP takes.
- ·12 min read
Shopify's AI coding gates are built so the agent can't grade its own work. A YouTube rebuild runs looser.
Shopify built Helix to rebuild its apps with AI agents that must pass four gates. A video recreated it in Claude Code. What Shopify said, what changed, what the evidence shows.
- ·14 min read
Spec-driven development has early support. The course's biggest promises are barely tested.
A DeepLearning.AI and JetBrains course teaches writing the spec before an AI agent writes code. Benchmarks back clearer specs. Its own workflow is barely tested.
- ·10 min read
When AI agents talk to each other, structure gets lost. New research points to where it starts.
A video calls language a catastrophically lossy channel between AI models. The paper behind that claim says lossy, not catastrophic, and finds most loss starts at the writer.
- ·9 min read
Anthropic's SDLC playbook has one rule worth copying everywhere: confirm the check fails first.
The AI-native SDLC playbook asks for a test that fails before a fix. We think the field data makes the case for putting that rule on every gate.
- ·8 min read
Google's ScientistTwo was graded by two AI reviewers. Only one of them was held out, and that is the whole result.
Under the reviewer inside its own refinement loop, ScientistTwo clears the human bar. Under the held-out one, its lead depends on which papers you count.
- ·4 min read
Anthropic's new project coordinator is real. The team-sharing pitch in the video covering it is not, yet.
Anthropic's Sept 17 Claude Code Projects redesign matches its own docs closely. The video calling it a team OS describes sharing that has not shipped.
- ·9 min read
A 40-year-old search algorithm beats modern embeddings at agentic research. Only one version of it does.
A Hornet CEO says BM25 is unreasonably effective for agentic search. The paper he cites disagrees, then a newer paper agrees. The condition between them is the real finding.
- ·6 min read
Chunking is not dead. The talk that says so smooths one of its own numbers.
An AI21 researcher attacks query-dependent chunk size with a 17-year-old fusion trick. The mechanism holds. A separate 2026 benchmark found the opposite on a different corpus.
- ·5 min read
"Discovery intelligence" is a real research term. The video's argument with MIT about it is built on a claim MIT never made.
A new paper's numbers on self-improving scientific agents check out under verification. The video covering it disagrees with MIT by inverting what MIT actually says.
- ·6 min read
A talk says gamify your AI agent like a LitRPG character. A contemporaneous essay proposes something else entirely.
Daniel Miessler's record on personal AI agents holds up. His evolution-as-destiny metaphor and gamified harness design rest on contested science, a dispute the talk never names.
- ·7 min read
A Stanford professor says TCP is done for AI clusters. The industry's real money is betting on something else.
Homa is a real, peer-reviewed transport protocol with genuine 2026 momentum. It also competes against Meta and a nine-vendor Ethernet consortium, neither mentioned in the talk.
- ·7 min read
Knowledge agents need a different shape than coding agents. The harder problem is not search.
A Mixedbread researcher argues knowledge agents need a different shape than coding agents. The architecture claim holds. The numbers mostly do not, and the real gap is elsewhere.
- ·5 min read
Union Alpha was never really a mystery model. It's a procurement question wearing a mystery's clothes.
The anonymous stealth model on OpenRouter has already been unmasked. What it actually is, a best-of-N ensemble, not a router or a frontier model, matters more than who made it.
- ·6 min read
Voice agents fail the same five ways. Here's which of the numbers behind that claim actually hold up.
A Plivo talk names five voice-agent failure modes. Independent benchmarks back most of it, show the talk understates one, and add a stronger case for its latency target.
- ·8 min read
Vercel's new agent framework tells a real scaling story. A different 2026 paper argues the opposite bet.
Eve is Vercel's new filesystem-first agent framework. Hypergraph research argues almost the reverse. Both are real, dated, and unconnected in any source we found.