Part of Analyze

Research Radar

8 articles

We read the paper so you can judge it: the question, the method, what was found, the limitations that matter, and what it could mean for your work. Preprints and vendor research are labeled as such.

All Research Radar articles

Newest first.

Illustrated large teacher-model block transferring an orange signal into a smaller student-model block, representing task-specific model distillation.
Research Radar
9 min read

Model Distillation for Engineering Leaders: When a Smaller Model Is the Right Call

Model distillation transfers a larger model's task-relevant behavior into a smaller student. Research demonstrates meaningful efficiency gains in defined settings, but not a universal replacement for the teacher.

An orange-lit code review screen and stopwatch illustrate why AI coding productivity must be measured across more than one outcome.
Research Radar
12 min read

AI Coding Tools and Productivity: What the Research Actually Found

AI coding research does not produce one productivity verdict. Controlled trials, repository analyses, security tests, and developer surveys measure different outcomes—and their limits matter.

Does Size Matter? Bigger AI Isn’t Always Better
Research Radar
9 min read

Does Size Matter? Bigger AI Isn’t Always Better

Does size matter? There. I asked it. Humanity has spent an unreasonable amount of time arguing about this question, so naturally we eventually gave the problem to artificial intelligence. Except now we measure it in billions of parameters .

A Tool Result Is Evidence, Not Ground Truth
Research Radar
5 min read

A Tool Result Is Evidence, Not Ground Truth

Connecting an agent to a database does not remove hallucination risk. It gives the system a second source that can be stale, malformed, unauthorized, or simply wrong—and a protocol that may make the model trust it more than it should. Tools

Benchmark the Model–Harness Pair, Not the Model
Research Radar
5 min read

Benchmark the Model–Harness Pair, Not the Model

“Which model is best for coding?” sounds like a precise engineering question. Usually, it is missing half of the system. A coding agent is a model inside a harness that chooses tools, assembles context, serializes results, decides when to

Your Agent Passed the Eval. Its Retrieval Calls Still Tripled.
Research Radar
5 min read

Your Agent Passed the Eval. Its Retrieval Calls Still Tripled.

An AI agent can pass the task and still fail the engineering evaluation. That happens when a context-compression policy preserves the final outcome but changes how the agent gets there. The agent still completes the benchmark. It also

AI Agents Do Not Just Need More Context. They Need a Context Budget
Research Radar
5 min read

AI Agents Do Not Just Need More Context. They Need a Context Budget

A long context window sounds like an obvious advantage for an AI agent. The agent can retain more search results, tool outputs, intermediate reasoning, and evidence. Give it enough context, and perhaps it will stop forgetting what it

Is Overthinking a Red Flag? We Put AI Reasoning to the Test.
Research Radar
2 min read

Is Overthinking a Red Flag? We Put AI Reasoning to the Test.

More Thinking Doesn't Always Mean Better Answers Reasoning models are designed to think step by step, and the common assumption is simple: More thinking = better results. I expected reasoning models to dramatically outperform simpler ones.

Newsletter

New guides, comparisons and analysis, sent only when there is something worth your time.

  • New articles worth your time, not every link we publish.
  • Tools and resources when they launch.
  • No spam, no selling your address. Unsubscribe anytime.

Get useful signal by email

We're building the list before the first issue. You'll get an email when the newsletter starts.