AI for CS teams
Your board wants AI. Give them faster CSMs and fewer surprises.
AI your CS team uses every week, inside your own Claude or ChatGPT, on your own data, with a person approving every step.
CS Copilot Build · From $8,750 · Live in 30 days
30 minutes. No data needed.
Halden Freight
$214K ARR · renews in 48 days
Pulled from: Salesforce · Product usage · Support tickets · Call notes
- What changed
- The champion left six weeks ago. Weekly active users are down 38% since. One ticket open about SSO setup.
- Commercial
- Three years a customer. Took a 6% uplift last renewal without pushback. No discount on file.
- Risk call
- High. Champion gone and usage falling: the strongest churn signal in last year’s losses.
Waiting for your approval
Task for the CSM
Executive sponsor call this week. Due Friday.
Draft email
To the new VP of Operations: an intro, the three results they got last year, and a 30-minute slot.
Talking points
Usage drop by team, the SSO fix date, and what renewal at today's price looks like.
The mandate is real. The results aren't, yet.
Pilots stall on the same three things: messy data, no owner, and nothing measured. That's retention ops work, not AI work.
- 91%
- of customer service leaders say their executives are pushing them to adopt AI in 2026.
- Gartner, Feb 2026
- 95%
- of generative AI pilots showed no measurable profit impact. The cause was poor fit with real work, not the models.
- MIT NANDA, Aug 2025
Phase 1
CS Copilot Build.
Four AI routines your CSMs run every week, connected read-only to your CRM, piloted, trained and measured.
- Pre-call account brief
- Usage, open tickets, recent calls and the commercial history on one page, before every customer call.
- Weekly renewal-risk review
- Every renewal in the window, with what changed this week and the next step on each.
- First-draft QBR
- The business review deck drafted from the account's own data, for the CSM to edit.
- Monday at-risk digest
- The accounts that slipped since last week, sent to the manager with an owner on each.
- Week 0ScopeNDA, data terms and a one-page scope. We pick the four routines and agree which data the AI may read. Your company buys the seats, on its own business plan.
- Week 1Connect, read-onlyYour admin connects Salesforce or HubSpot, plus call notes and support if you use them, with me on the screen share. Each CSM signs in as themselves, so the AI only sees what they already can.
- Week 2BuildThe routines, written around your health score, your renewal stages and your words, and tested on real accounts.
- Week 3PilotTwo or three CSMs and their manager use it for a week. I log every miss and fix it.
- Week 4Roll outTwo training sessions, an AI use policy and tool register for your team, and a handoff doc.
- Day 45Prove itAn adoption report: who uses it weekly, hours back as your CSMs report them, and renewals reviewed on time.
Phase 2
Renewal agent build.
When a renewal enters the 120-day window, the account brief, risk call, task and draft email are waiting for the CSM to approve.
From $15K · About 30 days
From $5K added to a playbook build. Runs inside Salesforce or on a schedule in your own cloud account. Email stays out of the first version.
- 1A renewal enters the 120-day window
- 2The agent pulls the account and opportunity from your CRM
- 3Then product usage, support history and recent call notes
- 4It reads adoption, open issues, sentiment and the commercial history
- 5It makes a risk call and picks the next step
- 6It drafts the summary, the CSM task, the email and talking points
- 7A CSM approves or edits
- 8Only then is it written back to your CRM, and the outcome is tracked
What it costs.
From $8,750
Founding rate: from $8,750 for the first three clients, in exchange for a written case result and one reference call. From $12,500 after that.
Up to 20 seats. Each extra 20 adds $2,500.
Run by Adam: 12 years running CX operations and $100M+ in ARR retained and grown, most recently at a PE-backed SaaS company.
Live in 30 days. If fewer than 70% of seats use it weekly by day 45, I keep working free until they do.
The guarantee holds once your seats are bought and your admin has connected the tools.
- Copilot care, from $1.5K a month: I run and maintain it inside your accounts: a new routine every month, tuning as your team and products change, and a monthly report on use and how often the risk calls were right.
- Your AI seats are billed to you by Anthropic or OpenAI, not marked up.
- You keep the routines, the prompts written for your team, and the policy.
Questions before you book.
Do you need our logins?
No. Your admin creates one integration user or key per tool, with only the permissions the routine needs, and can switch it off at any time. In Salesforce I work in your sandbox; production changes go through your admin.
Do we need Claude Enterprise?
No. A business plan for Claude or ChatGPT, the team tier, is enough for the copilot, at about $20 to $25 a seat each month. Enterprise is worth it if your security team wants single sign-on, audit logs or shorter data retention. The renewal agent runs on your own API account or inside Salesforce, paid by use.
Does AI touch our customer data?
Only inside your own AI workspace, on your company's business plan, never a personal account. Your admin connects each tool with its own integration user and the narrowest permissions, read-only first. I never use anyone's login. Nothing writes to your CRM or reaches a customer until a person on your team approves it. The Rundown still runs from exports.
We already tried an AI pilot. Nobody used it.
That's the usual ending, and why this is priced on use. Routines built around how your CSMs already work, piloted with the people who'll use them, and measured weekly. If fewer than 70% of seats use it weekly by day 45, I keep working until they do.
Is our data ready for this?
If you're not sure, start with the Retention Rundown. Its AI leverage map shows which routines to build first and whether your data can carry them, and its fee credits toward the build for 90 days.
AI that keeps renewals, not slides.
Not sure where to start? The Retention Rundown maps it first.