Where agents sit in the lifecycle, what records they may change, and how they are switched off. RevOps decides placement and write access; the function that owns each stage owns the result.

GTM Operating Model › Revenue Operations › AI agents
AI agents: where they sit and what they may touch
An agent here is software that reads records and acts in the GTM systems, from drafting a message to holding a live conversation, inside rules RevOps sets.
Capture
↗
Enrichment and research
fills firmographics and contacts and merges duplicates; researches target accounts and writes first-draft outbound messages; a person reviews and sends
Qualify
↗
Inbound chat or voice agent
on the demo-request fast lane: asks the qualifying questions and books on the owner’s calendar under the routing rules
Sell
↗
Notes and record-keeping
writes meeting notes and activity to the CRM; flags missing next steps and deal risks for the pipeline review
Onboard
↗
Kickoff prep
builds the kickoff brief from the handoff record; guides the customer through setup
Keep and grow
↗
Health and renewal support
summarizes health signals for the CSM; drafts renewal and expansion outreach; deflects routine support questions
CONTROL RULES
What agents may write
activity, notes, drafts, enrichment fields, and tasks for a person
What they may never write
stage, amount, close date, forecast category, ownership, the sourced channel; those stay with a person or a system rule
Who presses send
anything external in outbound is drafted and reviewed before it goes; inbound agents act live inside a script and hand off to a person on anything outside it
How they are measured
the same funnel metrics as the people doing the job, conversion and speed, plus an error rate from a weekly sampled audit
HOW THEY RUN
Its own CRM user
each agent signs in as a named system user, so every action is logged and attributable to it
↗
A kill switch per agent
one switch turns it off; the work falls back to the human queue it came from
Same change control as routing
changes to an agent’s instructions (its prompt) or its script go through the same review and log as routing rule changes
What did not hold up: fully autonomous outbound, where an agent finds prospects, writes to them and follows up with no person in the loop, was widely tried and mostly pulled back. What holds up is agents doing the research, drafting and record-keeping around a human conversation, and live agents on narrow inbound tasks with a clear handoff. How far each company lets agents act live varies with segment and risk; the write rules above do not.
owner · RevOps decides where agents sit and what they may write; the function that owns the stage owns the outcome

Boxes marked ↗ open the page for that part of the model.

How to read this

  • The useful split is by write access, not by task. Agents may write activity, notes, drafts, enrichment and tasks. Anything that moves pipeline or the forecast stays with a person or a system rule.
  • Outbound agents draft and a person sends. Inbound agents act live, but only inside a script, and hand off to a person the moment a conversation leaves it.
  • An agent is judged on the same funnel metrics as the person it supports, plus an error rate from a weekly sampled audit, so a fast agent that writes bad records shows up.
  • Each agent is its own CRM user with its own kill switch, and its instructions change through the same review as routing rules. That is what makes it auditable and safe to turn off.

Where it connects

Diagram source: 23-ai-agents.drawio 23-ai-agents.svg