Patched

Patched (Patched Codes, Inc.) has evolved from its early open-source 'patchflows' framework for AI dev-ops automation into a focused production company: 'Reliable AI for regulated operations.' The pitch is direct — most AI platforms generate outputs, Patched completes the work. The product runs end-to-end execution of regulated, high-stakes workflows like fraud and risk investigations (30k+ cases/month, 82% resolution rate, 3.4x ROI in published case studies), banking customer operations (400k+ requests/month, 93% resolution rate), and cross-system reconciliation (2M+ transactions/day, 80% effort reduction). Each workflow is a small, auditable, composable program — the original patchworks lineage — that uses LLMs as one step among many (static analysis, vulnerability scanners, dependency resolvers, ticketing systems, internal data) to produce reviewable, SLA-bound results. The framing matters: instead of giving an opaque copilot the keys to your operations, you deploy specific workflows for specific jobs with deterministic inputs, auditable steps, and Patched on the hook for SLA. SOC2-compliant and deployed on dedicated infrastructure, the product is sold to publicly listed enterprises in finance and regulated industries; pricing scales with operational volume and is contact-sales. The MCP angle: workflows can call MCP servers as tools (security scanners, ticketing, internal data services), letting you compose Patched into a broader MCP-aware tool stack.