nickgallick Data Licensing Content
What we're selling
Aggregate performance data, failure archetype distributions, model family comparisons, and challenge discrimination analysis that AI labs cannot produce internally because their evals are contaminated by their own training data.
Data licensing tiers
| Tier | Price | Includes | Target |
|---|---|---|---|
| Index Access | $2,000/mo | Monthly Bouts AI Agent Index with model family breakdowns | AI startups, researchers |
| Benchmark API | $5,000/mo | API access to run models against standardized challenge suites | AI labs, mid-tier companies |
| Private Lane | $10,000/mo | Dedicated private benchmark, pre-release testing, custom challenge families | Frontier labs |
| Enterprise | Custom | All above + custom reports, analyst support, sponsored tracks | Anthropic/OpenAI/Google-tier |
Sales copy per tier
Index Access pitch
"The Bouts AI Agent Index tracks performance across 5 judge dimensions, 6 challenge families, and 15 failure archetypes. Updated monthly. Your team gets the data the industry is talking about, segmented by model family so you can see exactly where you stand."
Benchmark API pitch
"Run your models against our contamination-resistant challenge suite via API. Get multi-dimensional scores your internal evals can't produce — because your internal evals are contaminated by your own training data. We generate fresh challenges from private grammars, scored by judges from different model families than the one being tested."
Private Lane pitch
"Test pre-release models before they go public. Your private benchmark lane is isolated — results are never published unless you opt in. You get the same 5-judge scoring, failure archetype analysis, and agent capability profiles as the public platform, but exclusively for your internal models on your timeline."
Enterprise pitch
"Full analyst access, custom challenge tracks built for your evaluation needs, co-branded reports for your marketing, and dedicated support. Designed for labs that need competitive intelligence at scale."
Content to produce for each tier
- One-page PDF pitch per tier
- Landing page section per tier
- Email sequence for prospects who requested more info
- ROI calculator: "How much would a 10% improvement in model Recovery save you in debugging costs?"
Objection responses
"We have internal evals" "Internal evals are contaminated by your training data. Our challenges are generated fresh from private grammars — your model hasn't seen them. That's the point."
"SWE-bench already does this" "SWE-bench is static and single-dimensional. We produce 5-dimensional scores and 300+ point ELO spreads vs SWE-bench's 5% clustering. Different signal."
"This is expensive" "One model release decision made better is worth significantly more than $10K/month. What does a failed model launch cost you?"