YesPress · Jul 2026
Independent write-up summarizing Halluminate’s Westworld simulators, deterministic offline training setup, founder background, and reported backers.
Data and RL environments to automate knowledge work
Company research
Public company, workplace, funding, and market signals
Updated Jul 30, 2026
Overview
Halluminate is a YC S25 applied AI/data company building reinforcement-learning environments, realistic sandboxes, datasets, and benchmarks to train and evaluate computer-use and browser-use AI agents, with a notable focus on financial-services workflows.
Primary product
Managed RL environments, datasets, and benchmarks for training and evaluating computer-use AI agents
Founded
2024
Headquarters
San Francisco, California, United States
Team size
20-30
Work style
Onsite
Industry
Technology, Information and Internet
Sub-industry
AI agent infrastructure / AI software testing / RL environments
Offices
San Francisco, California, United States
Business model
B2B enterprise data and evaluation services
Managed RL environment platform
Benchmark and dataset products
Funding
Stage
pre-seed
Total raised
$500K
Latest round
Pre-Seed · Oct 2025
Latest amount
$500K
Oct 2025 · Y Combinator
People & investors
Work & culture
Small, in-person, research-heavy team shipping benchmarks and simulators for frontier AI; public-facing and experiment-driven, with a strong bias toward practical training data and reusable evaluation infrastructure.
Work style
Onsite
Visa sponsorship
Yes
Compensation
A YC job post lists a Founding Member of Technical Staff at $150K-$250K base plus 0.50%-0.75% equity; expert contributors are paid roughly $100-$250/hour depending on role and experience.
Benefits
Fully covered medical, dental, and vision
401(k) with 4% match
$10k+ relocation stipend
Free daily meals
Gym membership
Free transportation (Ubers/Waymos)
Product & market
Pricing
Custom enterprise pricing; not publicly disclosed
Differentiators
- —Realistic, resettable sandboxes modeled after real workflows
- —Proprietary datasets and benchmarks
- —Expert-annotated evaluations and error analysis
- —Parallelizable fully managed environments
- —Financial-services focus (investment banking, private equity, consulting)
- —Deterministic / reproducible simulator-based training
Technology
Python
Linux
Playwright
Selenium
OpenAI
TypeScript
Customers
Leading computer-use model labs
The two largest browser-agent companies
Foundation model labs
Enterprises training computer-use AI
Competitors
OpenAI Operator
Anthropic Computer Use
Skyvern
Browserbase
Convergence AI
Browser Use Cloud API
Yutori
Scale & contact
Estimated revenue
≈$1M-$2M ARR (third-party estimate: $1.3M in 2025)
Estimated monthly visits
30.3K
Latest news
Halluminate · Nov 2025
Halluminate introduced Westworld, a suite of realistic web-app simulators for training and evaluating web agents, co-developed with Yutori.
Halluminate · Nov 2025
Halluminate launched BrowserBench, a benchmark for browser infrastructure stealth with 292 tasks across 292 websites, aimed at measuring real-world agent infrastructure reliability.
Halluminate · May 2025
Halluminate published Web Bench, a benchmark covering roughly 5,750 realistic tasks across 500+ websites to evaluate browser agents and infrastructure.
Y Combinator
YC’s launch post describes Halluminate’s product as realistic sandboxes, datasets, and evaluations for training computer-use AI, along with founder background and public contact emails.
Jobs and careers at Halluminate
0 jobs at Halluminate
No open positions
Check back later for new openings