People

The people surface is Dispatch's memory of the humans in your work life: colleagues, clients, candidates, investors, advisors. Each contact has a file the agent reads when operating in relation to them — drafting an email, preparing for a meeting, or summarizing recent activity.

Unlike a CRM, the people surface is not a sales pipeline. It is a per-person record that accumulates as you and the agent interact with them.

What it does

• Stores per-person context. Each contact lives at /people/{slug}.md. The file accumulates recent conversations, how the person prefers to be reached, their company, and ongoing threads.
• Enriches new contacts. When a person is first created, the agent dedupes them against existing people, then builds a short initial profile from your connected Gmail and Calendar plus a web search.
• Slug-stable identity. The slug is derived from the person's email or display name and stays stable as their information changes.
• Reads on relevant tasks. When the agent drafts an email, prepares for a meeting, or summarizes a thread, it reads the relevant people files for context.
• Writes durable facts. When the agent learns something durable about a person ("Sarah prefers 25-minute meetings"), it can update their file. The user can also edit directly.
• Surfaces in search. The agent's search tool spans entity types; "who at Acme have I been emailing?" returns matching people files.

How to use it

The people surface is mostly used implicitly. The agent reads it when it needs to; you don't have to invoke it. To work with it directly:

• "What do I know about this person?"
• "Update Sarah's file to note she's now CTO."
• "Who at Acme have I met with in the last quarter?"

What Dispatch can do here

• Contact priority: sort the people you work with into P0-P3 tiers so each tier is handled differently.
• VIP alerts: get pinged when a high-priority contact emails or messages you.
• CRM sync: pull CRM context into people profiles and write Dispatch activity back.

Limits and behavior

• One file per person. Slug collisions are resolved by the agent at creation; existing slugs remain stable.
• Auto-enrichment is on by default. New-contact enrichment is the New Person automation, which runs whenever a contact is created. You can turn it off or edit its prompt in Automations.
• Per-organization. Like all memory, people are scoped to the organization you are working in.

Related

• Meetings: the agent uses people files to enrich meeting prep.
• Inbox: drafting tone draws on the per-person file.
• Cookbook: Contact priority

For agents

• People files live at /people/{slug}.md.
• The slug is derived from email or display name and is stable.
• Files accumulate per-person context; the agent reads them when operating in relation to a contact.
• The search tool returns matching people files alongside other entity types.
• New people are enriched on creation by the NEW_PERSON triggered automation, which dedupes then gathers context from Gmail, Calendar, and web search (web_search/web_fetch) to write the initial profile. The automation is enabled by default and can be disabled or edited under Automations.
• Per-organization, like all memory surfaces.