Lehvel / Labs
We test AI on real work and write down what held up.
Each entry records the question, what we built, what the evidence shows and where it falls short. Failed tests are published too.
Where we are working.
- Organizing what a business knows
AI gathers scattered knowledge into one checked record, with a source for each fact.
In use
- Choosing the right AI
Which model takes which task, judged by finished work and cost.
Running experiment
- What AI says about a business
We ask the AI platforms about a business and check each claim against confirmed facts.
Running experiment
- Website changes by voice
An owner says what changed. Lehvel updates the site and checks the live page.
In production
Systems and experiments
Software we built to run Lehvel or deliver client work.
- Sales Desk
Our client sites run on checked facts, so our sales record does too. One system carries a customer from first conversation into delivery.
In use
- How Lehvel's AI systems hand work to each other
We got tired of carrying prompts and status between AI systems by hand. So we built a shared system that does it, with GitHub as the record.
Running experiment
Bounded tests with one question and a measured result.
- Audit Lab
Audit Lab is a website check that ties every finding to something it measured, and says could not check when it could not. In October 2026 one report broke that rule.
Running experiment
- What if the expensive AI is doing the wrong job?
We are testing whether a small AI model can handle the many small judgments around big-model work, without being allowed to act.
Running experiment
- Running operations by voice
Owners do not update forms. We are testing whether an owner can say what changed and have it become a change to the business record.
Prototype
Research and field notes
Studies we ran to answer one question.
Research
Short notes from real client work, backed by evidence.
Field notes
Archive notes.
Earlier industry notes, weekly reviews and paused work. Each stays live and dated. None is current work.
Field notes
Field notes
Field notes
Field notes
Systems
Field notes
Field notes
Field notes
Field notes
Field notes
Research
Research
Research
Research
Field notes
Get Upstream
One email a week: what we asked the AI, what it got wrong, and how we checked.
Our research reaches 17,000+ followers across our channels. Upstream is the weekly version.
Weekly. Unsubscribe anytime. Privacy
Contact
Talk to Lehvel.
Tell us where product information, quoting or workflows break down. You get a written scope before any work starts.