Agentic Testing: Where Agents Fit in the E2E Testing Stack
- Source
- slack.engineering
- Author
- Sergii Gorbachov
- Date
Context
An can test whether a user goal remains achievable while adapting its route through an interface. In Slack's testing experiments, agents repeatedly observed the page and chose their next action, often taking different paths to the same outcome. The study covered more than 200 executions of two flows in non-production workspaces.
That flexibility carries recurring work: observation, reasoning and accumulated are processed again on each run. A saved deterministic test replays its code. The authors consequently place agents in exploration and debugging, alongside repeatable regression tests. The experiment supports that division of work within its tested flows, rather than a universal ranking of browser tools or models.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
- MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
Checking sign-in…
Loading comments…