Why we build for the task, not the tool
Most automation software makes you learn a new system. We think it should disappear into the work you already do.
Read postZeBlog
Know all about AI, Automation, Engineering, and the tools we're building at ZeSol.
One email a month. No spam, unsubscribe any time.
Automation
How we're turning manual, repetitive work into software that runs itself.
Mapping a workflow before you automate it
The fastest way to build the wrong automation is to skip this step. Here's the checklist we use internally.
Read postNot everything repetitive should be automated. A quick way to tell what's worth the effort.
Read postWhy we design every ZeSol tool to fail loudly, not silently — and what that looks like in practice.
Read postUnder the Hood
Engineering notes from the team building LockMind and the rest of the ZeSol stack.
Running large models on a laptop, without losing your mind
The hardware-aware decisions behind LockMind's local inference, and where we still hit limits.
Read postModel-agnostic by design: what that actually costs
Supporting any model on Hugging Face sounds simple. Here's what breaks first.
Read postNo account, no cloud calls, no usage caps. That's a harder promise to keep than it sounds.
Read postA short field guide to figuring out why someone else's model behaves strangely on your machine.
Read postRelease Notes
What shipped, what changed, and what we're working on next.
LockMind 1.2: faster model switching
Swapping between downloaded models is now near-instant. Here's what changed under the hood.
Read postA first look at what's coming to mobile
Rememberal and Supergo are still in development — here's an early look at where they're headed.
Read postA round of fixes for CUDA detection on mixed GPU setups, plus smaller performance tweaks.
Read postA rough look at the roadmap across desktop, mobile, and web.
Read postField Notes
Behind-the-scenes thoughts on building ZeSol, from the team doing it.
Why we started with desktop, not the cloud
Building LockMind to run entirely offline was the harder path. Here's why we took it anyway.
Read postWhat "do it easy" means when you're the one building it
Our mission is about making complex work simple for users. It doesn't always feel simple to build.
Read postWhat people actually asked for, versus what we expected them to ask for.
Read postSaying no to features is most of the job. Here's the filter we use.
Read post