❓ Problem

Every account manager knows the moment. You're mid-renewal with a Fortune 1000 customer, things are going nicely, and then their VP says, "We agreed last quarter that onboarding would be included." Did you? It might be in a call note, an email thread, the signed order form or a Slack message from someone who's since left. You can say "let me circle back", which sounds weak, or you can nod along and hope, which is how discounts nobody approved end up in contracts.

The information exists, it's just nowhere useful. The average worker spends 25% of their week searching for information, and sales teams have it worse, with reps losing 60% of their time to non-selling work like hunting for decks, typing notes into the CRM and chasing approvals.

But the kicker isn't the wasted hours. It's that all of today's tools help before or after the call. The prep brief lands an hour early, the summary lands an hour late. The only moment that actually decides the outcome, the live conversation, is still run from memory. And on a $500k account, one misremembered commitment can cost more than the whole team's software budget.

✅ Solution

A live copilot for account teams that knows the full history of every customer relationship and speaks up, quietly, when the conversation drifts from the record.

• It listens and retrieves in real time. As the call unfolds, it pulls relevant commitments, past objections, pricing and figures from Salesforce or HubSpot, call notes, emails and contracts, each shown with a link to the exact source.
• It flags contradictions tactfully. When someone says something that conflicts with the record, it doesn't shout "WRONG". It suggests a graceful question: "Can we clarify the launch date? The signed plan says October."
• Every prompt shows its evidence. No citation, no prompt. This is the whole trust model: a rep will only glance at the sidebar mid-call if every card can be verified in one click.
• After the call, it closes the loop. It drafts a recap of decisions, new commitments and open questions for a human to approve, then writes the approved version back to the CRM, so the record is sharper for the next call.

The wedge is deliberately narrow: renewal calls and QBRs for US teams managing a handful of high-value accounts. The platform underneath, a live, cited memory of every customer relationship, expands naturally into sales, procurement, partnerships and anywhere promises get made on calls.

📊 Key Numbers

Market size

• The AI meeting assistant market sits at $3.1-3.5bn globally in 2025 and is forecast to add $3.46bn by 2029, with North America driving around 40% of that growth. This is a US-first market.
• The adjacent revenue intelligence category proves US buyers pay serious money for call data. San Francisco-based Gong alone passed $500m ARR in May 2026, growing over 55% year on year.
• Price anchor: Gong charges around $1,360-1,600 per user per year, so a premium seat for key account teams at ~$1,200-1,800 a year sits comfortably inside existing US sales tech budgets.

ARR potential (our estimates)

• Bottom-up: 500 US customers x 20 seats x $1,500/seat/year is ~$15m ARR.