Conserver

The Conserver β€” a Redis-backed conversation processing engine that ingests vCons, runs them through configurable chains of links, and writes the results to any of fourteen storage backends.

The Conserver is the runtime that turns vCons into something useful. It pulls vCons off Redis ingress queues, runs them through a configurable pipeline of links (transcribe β†’ analyze β†’ tag β†’ notify β†’ record audit trail β†’ …), and writes the finished result to one or more storages (Postgres, S3, MongoDB, Elasticsearch, Milvus, SCITT transparency services, the vCon MCP server, and more).

It's an open-source project β€” Python 3.12, FastAPI for the API tier, Redis for queuing, Docker Compose for deployment. The current build ships 22 standard links and 14 storage backends, plus tracers that emit a verifiable audit trail of every chain execution.

When to use the Conserver

  • You have vCons arriving from one or more adapters (phone systems, SIPREC, email, chat, LLM exports) and need to do something with them at scale.

  • You need a transcribe β†’ analyze β†’ store pipeline that runs reliably, scales horizontally, and handles failures via dead-letter queues.

  • You want a single integration point for downstream systems (CRM, data warehouse, MCP server, blockchain audit log) so adapter teams don't each build their own.

  • You need to track lifecycle events β€” creation, enhancement, deletion, consent revocation β€” on a SCITT transparency ledger.

Documentation layout

Get started:

Configure:

Build:

Operate:

Source and support:

Last updated

Was this helpful?