Litmus

Evals for humans

4.5Score#112 of 999 agents#5 of 15 in Recruiter

Why it ranks #112

Litmus, an AI recruiter from YC S26, ranks #112 of 999 AI agents and #5 of 15 in Recruiter, with a score of 4.5 of 10. Data as of 3 Oct 2026.

The score adds four parts, each 0 to 10, by weight. Every point comes from public evidence. How scoring works ·

Proof 5 × 30% 1.5
Scale 2 × 30% 0.6
Momentum 6 × 25% 1.5
Autonomy 6 × 15% 0.9
= 4.5

Proof
Do customers use it? 30% of the score
5
Claims named customers and work done at scale.
- +3
  Named customers
  “We're already helping build frontier technical teams at Mercor, Composio, Neo Scholars, and more.”
  YC page
- +2
  Work done at scale
  “3,000+ assessments evaluated.”
  Launch YC post · 9 Jul 2026
Not found: revenue or growth (+4), a customer result (+2), a customer count (+2)
Scale
Is there a real company behind it? 30% of the score
2
3 people.
- +2
  3 people on its YC page3 or 4 people
Not found: funding round in its news (+3)
Momentum
Is it shipping and growing now? 25% of the score
6
Shows a recent launch, a new claim and a new batch.
- +3
  Launched in the last 6 monthsLitmus - The industry's most accurate engineering interviewLaunch YC post · 9 Jul 2026
- +2
  A new traction claim
  “3,000+ assessments evaluated.”
  Launch YC post · 9 Jul 2026
- +1
  New: YC Summer 2026
Not found: open roles on YC (+2), funding news in the last year (+2)
Autonomy
How much of the job does it do on its own? 15% of the score
6
Supervised agent: takes actions in other systems; a person approves key steps.
- +6
  Level 2 of 4: Supervised agentTakes actions in other systems; a person approves key steps.
What the model read
- Acts in other systems: 32% likely
- Finishes whole tasks: 14% likely
- Calls itself an agent: 4% likely
- Level 1 from the model's reading of its pages; the evidence review found, with quotes, that its AI acts in the work, so the level is 2
Level 2: Takes actions in other systems; a person approves key steps.

About Litmus

Litmus is building the most accurate framework for evaluating and benchmarking human capability, starting with software.

AI will compound small differences in human capability into increasingly large differences in what people can accomplish, while making existing static benchmarks obsolete. Software is already there: AI can hill-climb any output-based evaluation, while the ability to direct it is becoming the defining advantage.

Litmus applies the same approach we already use for model evals to humans – creating world-like environments, and inspecting trajectory instead of just output.

Every knowledge industry will soon face the same problem. We build Litmus to tell you what humans are capable of.

We're already helping build frontier technical teams at Mercor, Composio, Neo Scholars, and more.