Customer Support

How to run customer qbrs with AI

Customer Support3 AI tools6 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

A quarterly business review is how customer-facing teams stay aligned with the accounts that matter most. You pull together usage data, support history, renewal status, and open action items, then sit down with the customer to review what's working, what isn't, and what the next 90 days look like. It's a relationship-maintenance ritual that keeps churn low and expansion conversations alive — and for most operators running lean, it happens just often enough to feel like a scramble every time.

The preparation work is exactly the kind of thing people expect AI to handle. Synthesizing notes from the last three calls, identifying themes in support tickets, drafting an agenda, writing a recap email after — these are all text-heavy tasks with clear inputs and clear outputs. Nothing requires deep domain judgment. So reaching for ChatGPT or Claude makes sense: you have a pile of information and need structured, readable output from it.

ChatGPT, Claude, and Gemini are genuinely useful here. They can turn messy meeting notes into clean summaries, draft a QBR agenda from a bulleted input, write a professional follow-up email, and help you spot patterns across a handful of support tickets if you paste them in. The output quality is solid. The challenge is everything around the prompting — gathering the inputs, keeping the format consistent, and not starting over next quarter.

Skip the manual work → See the Starch version

Customer Support3 AI tools6 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Before the QBR, open your CRM, support tool, and calendar and manually export or copy the relevant data for this account: recent deal activity, open tickets, last three call summaries, and any outstanding action items.

2 Paste that data into Claude or ChatGPT with a prompt asking it to summarize account health, highlight risks, and list unresolved items. Review the output and correct anything the model missed or misread.

3 Use a second prompt to generate a QBR agenda tailored to this account — structured as a 45- or 60-minute meeting with time allocations for each section, based on the context you just summarized.

4 During the QBR itself, take rough notes in whatever tool you use — Google Docs, Notion, or even the chat window. After the meeting, paste your notes into ChatGPT and prompt it to produce a structured summary with decisions made, open questions, and action items by owner.

5 Use a final prompt to draft the follow-up email to the customer: a short summary of what was discussed, confirmed next steps, and any commitments your team made. Edit before sending.

6 Save the output documents somewhere your team can find them — usually a shared Notion page, Google Drive folder, or the notes field of your CRM deal record, pasted in manually.

Prompts you can copy

Here is 90 days of account activity for Acme Corp — recent calls, open support tickets, and deal stage history. Summarize account health, flag any risks, and list outstanding action items with owners.

Based on this account summary, write a 45-minute QBR agenda. Include sections for reviewing the past quarter, discussing current blockers, and agreeing on goals for Q3. Allocate time to each section.

Here are my rough notes from today's QBR with Acme Corp. Write a structured summary with: key decisions made, open questions, action items by owner, and any commitments our team made.

Write a follow-up email to the Acme Corp account team after our QBR. Tone: professional but direct. Include a one-paragraph summary, a bulleted list of next steps with owners, and our agreed check-in date of July 15.

I have QBR notes from six customer accounts this quarter. Paste each one below and identify common themes — what problems keep coming up across accounts, and what are customers asking for most often?

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

No live connection to your CRM or support tool — you copy-paste account history every single time, and if you forget something, the summary reflects the gap.

Nothing carries over between quarters. The format you carefully prompted last time, the structure you liked — you rebuild it from scratch the next time a QBR comes around.

Outputs drift. The agenda Claude generates this month looks different from last month's even with similar prompts, so you end up re-editing to maintain consistency across accounts.

Action items disappear into a document. There's no connection back to your CRM or task manager, so confirmed next steps get buried unless someone copies them over manually.

Preparing for ten accounts in the same quarter means running this entire process ten separate times — same steps, same copy-paste, same reformatting, no reuse across accounts.

Meeting notes are only as good as what you typed during the call. If you were focused on the conversation, the raw input is thin, and the AI-generated summary reflects that.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system. For QBR workflows, that means an agent builds a persistent app connected to your live account data — CRM history, meeting notes, action items — so each quarter's prep isn't a fresh copy-paste exercise, it's a live view that's already up to date.

The CRM starter app connects your email and LinkedIn data on a schedule, so when you ask 'who haven't I touched in 30 days?' or pull up an account before a QBR, the history is already there — no export needed.

The Meeting Notes app transcribes each QBR in real time, generates a structured summary with decisions and action items, assigns owners, and archives everything in a searchable history — so next quarter's prep starts with a full record, not blank notes.

Action items extracted from meeting notes stay in Starch and connect back to the account record, so nothing falls through the gap between 'we wrote it down' and 'someone actually did it.'

Describe the QBR prep view you want in plain English — 'Show me each account's open action items, last call date, renewal date, and any support tickets in the last 60 days' — and an agent builds that dashboard and keeps it current against your live data.

When it's time to run QBRs across ten accounts, the structure is already consistent. Each account's summary pulls from the same connected sources, formatted the same way, without rebuilding the prompt chain for every one.

The Presentation Agent (coming soon) will take your QBR summary and turn it into a polished slide deck in minutes — describe the format and your key metrics, and it generates a deck ready to share with the customer.

Get closed-beta access →

Toolkit