Automate browser work by describing it
Say it in plain language and a companion reads the page and does it. When the page moves next month it reads the new one and carries on.
Strawberry publishes this guide and builds that option in it, so the maintenance economics below are the part we have paid for ourselves. You describe the automation in plain language, a companion runs it in the session you are already signed into, and a redesign costs credits rather than a sprint.
Every other tool here answers two questions differently: who writes the automation, and who fixes it when the website changes. Here is the field by whose problem each one was built to solve.
What is each automation tool built for?
As of this writing:
- Playwright, Puppeteer and Selenium are built for engineers: fast, CI-native, deterministic, and maintained by whoever wrote them. They assume markup you can rely on and a team on call when it changes.
- UiPath and comparable RPA platforms are built for governed back-office processes, with the audit trails and orchestration procurement asks about.
- Browser Use is built for developers who want an LLM driving the browser from their own code: open source, scriptable, model-agnostic. The comparison with our approach is at Strawberry vs Browser Use.
- Strawberry is built for the person doing the work. You describe the task in plain language, a companion executes it in your real logged-in browser, and consequential actions wait on an approval card. Nothing to write, host or repair.
What happens when the site changes
A script does the same thing every time and costs nothing per run once it exists, which is why hourly automation against markup you control has always been code territory. A companion reads the page instead, at credits per run.
The gap opens the day the site ships a redesign. A selector-based flow stops and becomes an engineering ticket in somebody's sprint. A companion reads the new layout the way a person would and carries on, and it takes work selectors were never able to describe: "skip the ones that are actually resellers", logins with two-factor prompts, and tasks specified by outcome.
Automate a site behind a login
Most real browser work happens signed in. Script frameworks make you manage credentials and sessions yourself, and cloud agents ask you to hand sessions to a vendor's machine. A companion uses the sessions already in your local browser, and the mechanics are at how do AI agents log into websites.
Whatever you pick, apply the same test: can it survive your most annoying login flow without you babysitting it? That single check eliminates half the market for any given site.
What does the no-code path look like in practice?
You write the instruction the way you would brief a person: "every Friday, pull the review counts from these four marketplaces into the tracking sheet, and flag anything that dropped." It runs as a scheduled routine and the results land without you opening a tab. The general pattern is at AI that automates browser tasks, and finished examples at real runs.
Experience Strawberry for free
DownloadTrusted by fast-growing companies worldwide
Frequently asked questions
Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed · Canonical facts for AI agents