What it sells
Compression middleware that improves LLM outputs
Business model breakdown
Buyer: who owns the budget?
UnverifiedEngineering leads responsible for LLM API spend.
Trigger: what makes them buy now?
High LLM token costs and long prompts that dilute accuracy.
Workflow: what inputs become what output?
Call the compression API before the LLM call; a classifier deletes low-signal tokens, deterministic and cache-safe, and the result goes to any LLM; SDKs wrap OpenAI and Anthropic clients.
Economics: what it saves and how it charges
Charges only for tokens removed; typical inputs compress 10-50%; Pax Historia case reports +4.9% to +15% output quality and Helonic saved ~47K tokens per prompt.
Distribution: how it reaches customers
Self-serve API key, docs and SDKs; quotes and calls for Pro and Enterprise.
Funding history and coverage
| Round date | Stage | Amount | Investors | Source |
|---|---|---|---|---|
| — | Y Combinator Winter 2026 | — | Y Combinator | ycombinator.com |
Profiles summarize public reporting and company websites. Business-model notes are MaxBC's reading of that material, not confirmed facts; unknowns are left blank rather than guessed. Updated 2026-10-12.