AI agents

A local model in OpenClaw vs a Dot on a frontier model: which jobs do you actually lose?

Casey Holt

October 7, 2026

Can a local model be the only model if the agent is limited to files on one computer?

Casey Holt

October 7, 2026

Does a local model refuse tool calls, or does it invent ones that do not exist?

Casey Holt

October 7, 2026

When should you stop a cloud agent overnight instead of letting it continue?

Casey Holt

October 7, 2026

What is the smallest access that still lets a phone-controlled agent complete a booking?

Lena Kowalski

October 7, 2026

Does a calendar connection let an agent invite other people, or only read your day?

Lena Kowalski

October 7, 2026

Which memory do you delete when a project ends: the chat, the saved routine, or the logged-in browser profile?

Lena Kowalski

October 7, 2026

What do you disconnect first if you think an agent was tricked: the chat channel, the saved login, or the API key?

Lena Kowalski

October 7, 2026

How do you see, after the fact, which account an agent actually used?

Lena Kowalski

October 7, 2026

For a research job, is a browser with no logins enough, or do you need your real session?

Lena Kowalski

October 7, 2026

When does handing work from one agent to another lose the original request?

Casey Holt

October 7, 2026

Is a written instruction file more reliable than "watch me do it once"?

Casey Holt

October 7, 2026

What should the coordinator be blocked from doing, even when the specialists are allowed?

Casey Holt

October 7, 2026

Should the agent that reads the web be a different agent from the one allowed to send mail?

Lena Kowalski

October 7, 2026

Is a phone notification enough, or can the agent take the next step before you read it?

Lena Kowalski

October 7, 2026

Does an approval button before "send" stop a bad message, or only delay it?

Lena Kowalski

October 7, 2026

If an agent can open your password manager, what can a hostile web page make it do?

Lena Kowalski

October 7, 2026

Which accounts are reasonable to connect on the first day, and which should stay human-only?

Lena Kowalski

October 7, 2026

Can you trust an agent with the login to your email, or only with a draft it cannot send?

Lena Kowalski

October 7, 2026

Can two cheap models check each other well enough to skip the frontier model?

Casey Holt

October 7, 2026

Should the expensive model see every message, or only the step that uses a tool?

Casey Holt

October 7, 2026

What goes wrong when a cheap coordinator sends the job to the wrong agent?

Casey Holt

October 7, 2026

Does the coordinator that only assigns work need an expensive model?

Casey Holt

October 7, 2026

Which step breaks first on a local model: choosing the tool, filling the arguments, or finishing a long task?

Casey Holt

October 7, 2026

Can an agent that browses and calls tools run entirely on a local model?

Casey Holt

October 7, 2026

From IE6 to AI shells: the browser stopped being a window — here’s what I’d build for now

Casey Holt

September 18, 2026

Spec-driven development at a 100-person company: what actually stuck after we tried it

Tobias Werner

September 18, 2026

Hiring in the agent era: I wouldn’t staff like it’s 2021. Here’s the smaller team I’d build

Marcus Dalton

September 18, 2026

From IDEs to AI agents: the history of dev tooling — and what I’d bet on next

Casey Holt

September 18, 2026

What a year of AI agents taught me that the demos left out

Casey Holt

August 27, 2026