Pre-production · No SLA · No hosted service today · FSV §1.8
One tier. Founder-fielded pilots.
Mejepa is pre-production. Today we sell one engagement: a founder-fielded pilot on one Python repository and one agent loop. Hosted SaaS and OEM substrate pricing are fenced honestly below — they unlock when the Panel A ship-gate hits 0.95 and Panel B (cross-panel triangulation, #405) clears its first measurement .
PILOT · TODAY
Founder-fielded pilot
Steve or Chris drives the engagement end-to-end. One Python repository. One agent loop (Claude Code, Cursor, Windsurf, Cline, Continue, or Replit). The verdict shape, named failure modes, closest exemplars, and the signed witness chain are delivered against your codebase.
What's included
- Mejepa MCP wiring into one agent harness
- Calibrated verdicts on AI-written patches against your repository
- Docker-oracle ground truth for in-scope mutation categories
- Signed witness packets delivered with the engagement; offline-replay tool included
- Founder time for wiring, calibration review, and debrief
Pricing posture
Per-engagement, scoped on a discovery call. We do not publish a price-sheet because the inference cost floor is CUDA-bound and pilot scope varies materially by repository size, mutation surface, and agent loop. Expect 2-4 weeks engagement length.
Request a pilot →FENCED · POST-M2 SHIP-GATE
Multi-tenant Python verification · fenced
Multi-tenant calibration is engineering risk only , but we will not sell it until the ship-gate fires green and isolation primitives are shipped (M4 productization). Pricing posture when it lands: per-verification + seat-based for signed packet downloads, anchored to risk-of-not-having-it at the audit / customer / insurance trigger. Today, a hosted offering is on the roadmap, not on the price-sheet.
Why fenced today: running multi-tenant calibration before Panel B passes means we'd be selling production stability we have not measured. The roadmap rung is real; the rung below it has not shipped yet.
FENCED · POST-M2 SHIP-GATE
OEM substrate · agent platforms + audit firms
Agent platforms (Claude Code, Cursor, Windsurf, Cline, Continue, Replit) and audit firms building AI-assurance practices are the natural OEM channel. Co-sell pilots are open today on a per-engagement basis (Tier 2 of /verticals). OEM substrate licensing is fenced until ship-gate green. Pricing posture when it lands: per-verification + per-tenant.
Request a pilot
Tell us what your AI agent shipped, and what proof you need.
Form goes to Steve and Chris directly. Replies within 1 business day. We decline pilots where the verification target falls outside the FSV-bounded surface (non-Python, non-code, document corpora) — see the honest opportunity ladder.