Tools I use The bin

Tools I might use

The promise of a tool and the reality often don't add up. What's in this bin is an experiment that might give value. A tool takes a while to move onto the bench. That usually happens when I commit and roll it into a client's tech stack — which means I have to defend the inclusion — or when I build it into my personal engine.

Orca

Found on Nov 19, 2026

Two builders mentioned Orca. The signals are really rough, so I’m logging it. It might not stay on this list.

Pangram

Found on Nov 17, 2026

Pangram checks, programmatically, whether the content was generated by AI. As more systems embed AI writing, the risk of sounding like AI goes up.

Back when I was on Beehiiv and Substack, they offered it as a way to catch when you had vibe-coded something and thought it was your own. I liked the service enough that I started digging into their AI.

What I need it for now is the CRM AI stacks I’ve been building. I want an AI governor that lets an organization keep its own tone dictionary — the same idea as a design dictionary — and then use an outside service to check two things: that what we’re saying does not sound like AI, and that it actually delivers the value we want.

Rex

Found on Nov 5, 2026

I got on the waitlist for Rex, from Superlogical, and I’m still hoping.

Once I’m off the waitlist, I’ll commit to a local inference box. Between now and then, I’m doing demos with other AI builders to see who includes this in their tech stack, and when.

Conductor

Found on Nov 3, 2026

A designer with a sharp AI process put this in front of me. Where a developer delivers code and features, he delivers the narrative.

He uses Conductor to run multiple AI sessions in parallel. I love that. Once I get to AI automation, I have to be careful about what I run in series and what I can run in parallel. If I can split the work that still needs my attention from the work that is already specced, a whole set of features can just run, without me watching them. I’m starting to break my own processes apart, and this one is kind of amazing.

Not long ago, Meta bought it. The terms say you can’t sign up right now. You will be able to soon.

I care about simple, direct features and services. I hate opening a product and wading through a smorgasbord of crap to find the one thing that’s actually valuable. I call that the smorgasbord of features. Or just Microsoft, which loves a button for everything so it can account for everything.

Herdr

Found on Oct 22, 2026

This is my wildcard. Two different solo operators mentioned Herdr as the tool they use to balance agents across inference setups.

I’m not running a separate inference agent or a separate inference model. I don’t have an Apple Mini sitting at home. I’m starting to think about how to segment that into the workflow. This is the one I’m digging into most.

Cursor Projects

Found on early September 2026

Cursor Projects is a new feature. I get the sense it will pick up some polish over the next few months. Right now the mobile app and the desktop app don’t really line up, so it still feels new. Setting up a project and having it act like a code-ready project manager is an interesting approach.

Cursor still has some product-manager issues to work through. The flow between Projects and Automations feels a little early for a release. I tried wiring an automation that triggered from a Linear ticket into a project, and I didn’t get the flow I was hoping for.

I did have an early win. I was in Iceland for Microconf. A speaker mentioned something, and I realized I was sitting in the conference room with my phone, and I already had a project linked to the right repos. So I prompted it to build what I wanted. I watched it sit quietly, research the features, decide where it fit, build it, and hand me a pull request. The prompt was very basic. Taking a short prompt and inflating it is, at least right now, my killer use case.