Week of July 25–31, 2026
Project-level YAML import and export 🔥
You can now import and export your entire test suite at the Project level. Previously, YAML import and export worked test by test, which meant moving or reusing a full suite required doing it piece by piece.
Export a project to hand off to a teammate, commit it to version control, or use as a starting point for a new project. Import any existing project YAML to get a full suite running without rebuilding from scratch. And if you're creating a brand new project, import works there too — you don't have to set anything up first.
Orgs and subscriptions 🔥
Teams now have an Org — a shared container for members, projects, billing, and LLM configuration. If you're working with colleagues, this is where the shared settings live.
The most useful part: LLM configuration set at the Org level cascades down to every project automatically. Set your model and API key once, and your whole team inherits it. Individual projects can still override if they need to work differently. Billing connects through Stripe from the Subscription tab in Org settings.
Invite activation
Inviting someone to a project now takes them through a real activation flow. If they're new to LitmusLab, they set up an account. Once done, they land directly in the project they were invited to and immediately take on permissions assigned by an admin.
Last run timestamp
Each test now shows when it was last run. A small addition, but it answers a question that comes up often when running tests regularly: did this actually fire, and when? You can see at a glance which tests are being actively used and which haven't been touched in a while.
Public site
The public site was rebuilt, to bring faster load times and better search indexing; and more visibility, by providing page links that visitors share on Slack, LinkedIn, or X, with a proper preview card rather than a bare URL.
In the Lab
This week we published The Speed Trap, the first in a series on AI quality at scale. It starts with a number from Google's 2026 AI Agent Trends report: 88% of early adopters are seeing positive ROI from agentic AI. Then it asks the harder question buried in the same report: how are teams supposed to verify quality at the speed AI actually produces output?