The meeting prep that works in 2026 is not a one-page summary an AE skims in the elevator. It is a structured brief an agent assembles from the CRM account, engagement history, recent news, and product usage signals. The AE adds one or two angles only a human would notice, and the brief stays attached to the deal so the next call builds on what was learned. The agent reads. The AE thinks. The artifact persists.
The workflow
1. Pull the account snapshot from CRM. Start in Salesforce or HubSpot. The agent reads the account record, open opportunities, contact roles, last activity, and any custom fields RevOps cares about. If the CRM is dirty, the brief will be wrong, which is why a quarterly CRM cleanup pass is upstream of every prep workflow.
2. Read the conversation history. Gong or Chorus holds the call transcripts. The agent pulls the last two calls, extracts objections raised, commitments made, and next steps agreed. A useful prompt: "Summarize what the buyer said they were trying to solve, and what was blocking them."
3. Add external signal. ChatGPT or Claude with web access pulls the last 30 days of news on the account: funding, leadership changes, earnings commentary, press releases. The agent flags anything that maps to the buyer's stated problem. No more than three items. More than three is noise.
4. Layer product usage if it exists. For PLG motions, the agent reads the workspace or product analytics: seats activated, features touched, last login. The brief calls out the gap between what was promised in the last call and what actually happened in the product.
5. The AE adds one custom angle. The agent cannot do this part. The AE writes two or three sentences: a hypothesis about why this account is moving slowly, a question to test on the call, or a reference customer to bring up. The brief is now actually useful.
The meeting itself runs through Chili Piper or a similar router, and the calendar invite links to the brief.
Worked example
An AE has a 30-minute second call with a 400-person ops team. The agent assembles: Salesforce account, two prior Gong transcripts, a TechCrunch story about a recent acquisition, and a usage report showing the champion logged in twice last week. The brief surfaces that the champion asked about SSO on the first call, never got a clean answer, and has not invited the security team to the trial. The AE adds one line: "Lead with SSO. Offer to bring our security engineer to the next call." That brief took the agent four minutes. The AE spent eight minutes on the custom angle. The call lands. Gong research shows top reps run a 46 percent talk-to-listen ratio and introduce next steps early (Gong sales statistics), which is much easier when the prep is already done.
The persistent-state problem
Here is where most teams lose the compound. The brief gets pasted into a meeting invite, the AE takes notes in a doc, and after the call the artifact is orphaned. Next quarter the same AE, or a new one, reads the Salesforce activity log and has no idea what was decided or why. The CRM holds the stage. It does not hold the reasoning.
One way to solve this is a workspace like Dock that holds the prep brief, the AE's custom angle, the post-call summary, and the next-step decision as structured rows. Salesforce and HubSpot stay the system of record for the account, contact, and stage. Dock holds what the agent interpreted around the deal. Pointers like salesforce_account_id and hubspot_contact_id link Dock rows back to the CRM record, so nothing is duplicated, just connected. The same agent identity that drafted the brief signs the post-call summary, so attribution is clean.
Why it matters
HubSpot's 2025 sales research found that only 27 percent of reps consistently hit quota (HubSpot sales statistics). The gap is rarely effort. It is leverage. AI meeting prep that compounds across calls is one of the few workflow changes that actually moves that number.
Read the full AI sales prospecting playbook for the broader pillar this fits into.
FAQ
Q: Should the agent write the brief or just draft it? A: Draft. The AE adds one or two angles and signs off. A fully agent-written brief gets skimmed the same way a generic one does.
Q: Where should the brief live? A: Not in the calendar invite. The calendar invite links to the brief. The brief lives in a workspace row that survives the meeting and gets updated after the call.
Q: How do we keep Salesforce or HubSpot in sync?
A: The CRM stays the system of record for stage, contact, and account. The brief workspace stores pointers (salesforce_account_id) back to the CRM row. Sync is one-way: CRM is canonical for the deal record.
Q: What if the rep does not trust the agent's summary? A: Show the sources. Every claim in the brief should link back to the Gong transcript, the Salesforce field, or the news article. Trust comes from auditability, not from the model being smarter.