Archal

YC s26

Eval platform for autonomous software

[Archal @archal_labs]

Today we're excited to introduce Archal publicly, the first deterministic eval platform for AI agents.

Software testing exists because we figured out that catching bugs before users do is cheaper than catching them after. Somewhere in the last two years that lesson got lost. Agents ship without anything resembling a test suite or with simple mocks and Langfuse strapped on. Observability tools only tell you about these bugs after a customer is already affected.

We built Archal because there's no excuse for this in the big 26.

[Archal @archal_labs]

@davecyen @agentcardhq Hey dave :)

[Dave Yen @davecyen]

YC S26 kicked off this week and @orangecollectv is already swarming

🤝 @ryanbed

[Archal @archal_labs]

good stuff here!

[ben hylak @benhylak]

introducing howtoeval dot com. the no-bullshit guide to eval'ing AI agents.

from personal experience, and from working with the best companies in the world.

there's even a quiz. link below.

[Archal @archal_labs]

Thread 1/1: Archal supports 11 clones of popular SaaS services to support testing your agents. These clones are fully implemented stateful copies of MCP tools and API's from the original services.

[Archal @archal_labs]

Today we're excited to introduce Archal publicly, the first deterministic eval platform for AI agents.

Software testing exists because we figured out that catching bugs before users do is cheaper than catching them after. Somewhere in the last two years that lesson got lost. Agents ship without anything resembling a test suite or with simple mocks and Langfuse strapped on. Observability tools only tell you about these bugs after a customer is already affected.

We built Archal because there's no excuse for this in the big 26.