Most teams buy AI licenses, run a lunch-and-learn, and call it a rollout. Ninety days later a few power users love it, everyone else forgot it exists, and the forecast looks exactly the same. The gap isn't the model. It's the system around it — what it can reach, what it's trusted to finish, and whether your best reps' methods are written down where everyone can run them.
This is how we roll out Claude inside revenue teams: owned by RevOps, wired into the CRM, measured against pipeline — not "hours saved."
Two orgs can hand the same number of reps the same AI and get opposite results. Return scales with access (which systems it reads and writes) and trust (how much of the work it's allowed to finish). Click where you are today.
RevOps owns the CRM, the integrations, and the pipeline reporting where results get measured. Whoever owns the system of record should own the program.
If reps are uploading call transcripts by hand, you've automated nothing. Connect CRM, call recording, email and Slack so the work starts itself.
Provisioning, SSO, and admin consents for Salesforce or HubSpot and M365. Lock dates with IT before you promise a pilot date.
Data boundaries, connector permissions, audit trail, telemetry. Connectors inherit each rep's existing permissions — lead with that.
e.g. call-prep briefs generated + pipeline per rep. Baseline both before launch and put the readout dates on the calendar.
Limits by org, team and user. Track usage weekly. Give champions extra headroom — their experiments become everyone's skills.
Two or three teams with motivated managers. Provision plugins centrally so the pilot doesn't splinter into personal setups.
Connectors inherit each rep's existing permissions. The more of the stack you wire in, the more work finishes itself.
The Sales and Salesforce plugins are only part of the setup. Claude also runs inside the tools reps already open every day, so the work lands where it gets used.
Quick question, rewrite, or pre-call brainstorm
Research or a finished doc built from your files and connected tools: account briefs, the daily briefing, scheduled runs
Quota or comp model, or analysis inside a workbook
Proposal or QBR deck built in the file itself
A deal question asked in the team's Slack channel
CRM automations, integrations, data pipelines
Anthropic reports its own reps now get a call-prep brief in about 2 minutes instead of roughly 30 spent digging through CRM notes, email and recordings. After the call, Claude drafts the follow-up and writes the opportunity update into the CRM. No end-of-week data entry means the forecast, and every ad hoc pipeline check from leadership, runs on current data.
The pattern we see everywhere: half the team barely uses the AI it already pays for, top sellers hoard prompts nobody else runs, and account context is scattered across CRM, email, call recordings and Slack. Reps rebuild it before every call, often 30 minutes of prep for a 15-minute conversation. Anthropic's Sales plugin packages that work into commands and skills.
/call-summaryTurns call notes or a transcript into action items, a drafted follow-up and an internal summary.
/forecastBuilds weighted projections from pipeline data.
/pipeline-reviewScores deal health and flags risk.
account-researchCompany or contact intel: key people, recent news, hiring signals.
call-prepAccount context, attendee research, a suggested agenda and discovery questions.
daily-briefingA prioritized day: meetings, pipeline alerts, email priorities, next actions.
draft-outreachResearches the prospect first, then drafts personalized email and LinkedIn messages.
competitive-intelligenceProduct and pricing comparisons, recent releases, a differentiation matrix and talk tracks.
create-an-assetProspect-specific landing pages, decks, one-pagers and workflow demos.
At setup the plugin interviews each rep and saves their name, quota, positioning and competitor list. It runs on web search alone, but it only gets good once the CRM, calls and email are connected. JayOh preloads those settings and wires the connectors so every rep has a working command on first login.
Three phases, each with an exit gate. The pilot is where you learn; the scale phase is where the money is.
IT and RevOps in parallel.
2–3 teams, 2–3 use cases each.
Provision in waves (e.g. 25 → 150 → all).
Rule of thumb: one champion per 25–50 users. Seats scale fast; champions don't — that's where rollouts quietly stall.
"Time freed up" is capped at what that time costs. The case that gets budget is more revenue from the same headcount.
Same work, faster: call prep, follow-ups, CRM updates. Easy to show. Hard to fund on its own.
More output from the same team: more accounts covered, more pipeline per rep.
Work that never happened before: scoring every account in a book overnight, real outreach to the long tail no rep had time for.
Pilot teams and non-pilot teams sell into the same market in the same quarter. Compare pipeline per rep, cycle length and win rate side by side — a gap there is far easier to credit than a quarter-over-quarter swing.
High spend isn't the problem; high spend with low output is. A champion at the top of the usage table who ships every day is the program working. For the heavy-spend, low-output rep: coach first, cap second.
It's going fine, so nobody calls it. The sponsor moves on, and a year later you're still at 25 seats with gains too small to show up in the numbers the CFO reads.
Put the scale decision on the exec sponsor's calendar at kickoff. Name the decider. Agree in writing what a "yes" requires.
Three champions can carry 25 people. They can't carry 150. Later waves get a bad first week with nobody to help — and first impressions stick.
Hold the 1:25–50 ratio every wave. Before provisioning, have incoming managers name their champions so the current ones can train them.
Usage-based billing climbs with adoption. Without caps or a weekly read, the bill surprises finance — and the reflex is to pull access mid-rollout.
Caps by org, team and user before the pilot. A weekly usage review so the budget owner sees the curve coming.
Check what's true today. Be honest — the score is only for you.
JayOh's growth pods own the RevOps side of the rollout: wiring Claude into HubSpot or Salesforce, packaging your top performers' playbooks into shared skills, setting governance and spend controls, and building the pilot-vs-control readout your CFO actually trusts.
Sprint-based. Measured against pipeline. Owned end to end.
Talk to JayOhWe engineer growth systems. JayOh builds the RevOps architecture, GTM execution and data infrastructure that turn messy revenue operations into predictable pipeline for B2B SaaS and PE-backed teams.