What is Trawler
Trawler tests a web product the way people use it. You paste the address of something that runs in a browser, and Trawler reads the page and proposes a few people — each with a situation, a temperament and things they care about — and the goals they want to get done. Then each person uses your real product in a real browser, works through the goals, and reports what went wrong on the way.
Nothing a person reports reaches the top of your report on their word alone. Every defect is handed to a fresh agent that has never seen the report: it is given the steps and nothing about what went wrong, follows them in a new browser, and says what it saw. A judge compares that with the claim. Only a defect the replay reproduces is confirmed.
Who it is for
Section titled “Who it is for”- Teams with a staging or preview environment who want to know where a first-day user gets stuck before a first-day user does.
- Anyone who built something quickly — a prototype, a proof of concept, a new feature — and wants it used by someone who is not its author.
- People who would rather read five confirmed defects than fifty guesses. A run costs a few dollars on your own model key and ends with a report that says what was reproduced and what was not.
What a run gives you
Section titled “What a run gives you”| What it means in practice | |
|---|---|
| People and goals | Proposed from your product’s page in about half a minute, and yours to edit before anything runs. |
| Use in a real browser | Each person clicks, types and navigates your product as it is — no scripts, no recorded flows to maintain. |
| Defects and friction | A defect is something that behaved wrongly; friction is something that worked but slowed a person down. Both come with the steps that led there. |
| Blind replay | A fresh agent follows each defect’s steps without being told what went wrong, and a judge decides whether it saw the same thing. |
| A hard cap | The run is estimated before you start it, and stops once its spending reaches the cap you set. |
What it is not
Section titled “What it is not”- Not a scripted test suite. The people decide how to reach a goal. Two runs of the same plan need not take the same path.
- Not a load test or a security scan. Trawler uses a product the way one person at a time would.
- Not an opinion panel. The people are told that an opinion about the design is not a finding; they report what they saw in the browser.
- Not for real people’s data. Starting a run asks you to confirm that you are authorised to test the product and that it is not a production system with real people’s data.
Not there yet
Section titled “Not there yet”Hosted runs are in a private beta: anyone can sign in, analyse a product and edit a plan, but starting a run needs an email address on the private beta list — write to contact@usetrawler.com.
Not built yet, in case you go looking: comparing a run with an earlier one, people using the product together, handing runs from the app to a runner inside your own network, reaching staging behind basic auth or a secret header from the app, and a list of past projects and runs. The local runner covers private addresses and protected staging today, from your own machine.