# Doris — The Semantic Layer for Go-to-Market

> Doris connects a company's CRM, calls, emails and documents into one typed graph, the ontology, and does the follow-up work behind what was said. The calls are one input, not the whole picture. Every dashboard, agent and rep reads the same deal record.

## What Doris Is

A sales team's truth is spread across a CRM, a recording tool, an inbox and a drive, and none of them agree. Doris assembles one record per deal from all of it, in types a machine can filter, count and compare: commitments, objections, stakeholders, competitors, meetings, documents. Each fact cites the passage it came from, so a rep can check it.

That record is the product. The notetaker fills it, the agent answers from it, flows act on it, and any MCP-compatible agent can query it.

## Products

### Notetaker — https://dorislabs.com/product/notetaker
Joins every external call on Zoom, Google Meet and Microsoft Teams from the calendar invite. No extension, nothing to start or stop. Produces a transcript with speakers resolved to real people and marked internal or buyer, then extracts typed facts onto the deal the call belongs to. Recording can be switched off per user and per meeting, and the exclusion survives the next calendar sync.

### Sales agent — https://dorislabs.com/product/sales-agent
Ask anything about the pipeline. The answer comes from the deal record — calls, emails, documents, CRM fields — with the number in the sentence and a link to the deal it came from. When a deal hits an objection that an already-won deal hit, it brings back what the rep said and whether the buyer accepted it.

### Flows — https://dorislabs.com/product/flows
Typed automations that do the follow-up work: draft the re-engagement email, write the brief, update the CRM, loop in the rep when a commitment goes past its date. Built with a Doris engineer around how a team sells.

### Ontology — https://dorislabs.com/product/ontology
The layer underneath the other three. 25 entity types, one API call for full deal context, exposed over REST and as a remote MCP server.

## The Ontology

- One typed record per deal, assembled from CRM, meetings, emails and documents.
- 25 entity types, each with properties and links, each fact carrying provenance back to its source.
- Read it over REST at `https://server.meetdoris.com/api/v1/ontology/types` (OpenAPI at `/api/v1/ontology/openapi.json`), or over MCP at `https://mcp.meetdoris.com/mcp` (Streamable HTTP, OAuth 2.1).
- Claude, ChatGPT, Cursor, Claude Code and any MCP-compatible agent can query the whole deal graph in a single call.
- Developer guide: https://dorislabs.com/blog/doris-ontology-developer-guide
- Cost measured against an equivalent CRM + notetaker + email MCP stack: https://dorislabs.com/ontology-vs-mcp-stack

## Try It Without Signing Up

The demo ontology is public and fully anonymised: 244 deals, 80 MEDDPICC assessments, 1,708 meetings.

- Connect an agent to `https://demo-mcp.meetdoris.com/mcp`
- Instructions and worked prompts: https://dorislabs.com/try

## Who Doris Is For

- **Sales reps**: the follow-up, the brief and the CRM update are written from the call, not remembered afterwards.
- **Sales managers**: see what is slipping and why, from evidence, without listening to recordings.
- **CROs**: forecast against what buyers actually said, not self-reported stage fields.
- **RevOps and enablement**: typed facts across every deal, so playbooks and coaching come from the record.

## Integrations

- CRM: HubSpot, Salesforce.
- Meetings: Zoom, Google Meet, Microsoft Teams, joined from the calendar.
- Calendar and mail: Google and Microsoft.
- Agents: any MCP client.

## Qualification Frameworks

MEDDPICC, BANT, SPICED, and custom frameworks. Deals are scored from conversation evidence rather than self-reported fields.

## How Doris Differs

- **Against Gong and Chorus**: those stop at the call — record, transcribe, score. Doris connects the call to the CRM, the inbox and the documents in one typed graph, then does the work behind the promises the rep made.
- **Against a general assistant**: ChatGPT or Claude can summarise one call and forgets it. Doris holds every commitment across every deal, so on the fourth call it knows what was promised on the first.
- **Against an MCP stack**: wiring a CRM MCP, a notetaker MCP and an email MCP together makes the agent do the joining, one tool call at a time. The measurement is on the comparison page above.

## Setup

Connect the calendar, link the CRM, and Doris starts joining calls. Most teams are running in under 15 minutes. There is no training data to supply — Doris learns the deals from the conversations.

## Company

- **Website**: https://dorislabs.com
- **Mission**: https://dorislabs.com/mission
- **Blog**: https://dorislabs.com/blog
- **Careers**: https://dorislabs.com/careers
- **Documentation**: https://docs.meetdoris.com
- **Changelog**: https://docs.meetdoris.com/changelog
- **App**: https://app.meetdoris.com
- **Terms**: https://dorislabs.com/terms
- **Privacy**: https://dorislabs.com/privacy
- **Sub-processors**: https://dorislabs.com/subprocessors
- **Contact**: help@meetdoris.com