Thalatta

An island · the Surface

Sales examples — glass agents across a revenue team

One architecture, reused. A glass agent works from the right object — a prospect's record, an account, a rep's question — plus the memory layer: the team's standing knowledge (the playbook, product and pricing, battle cards, SOPs, training). Point that same shape at any job below. The memory layer is the shared brain; each agent is handed the slice the moment needs.

Facing the customer

  • Outreach that explains itself. A prospect sees the exact record you hold, asks why you reached out, and sets their own status. (That's the glass demo.)
  • Lead qualification. Greets an inbound lead, asks the questions your playbook defines, and routes or books — from your ICP and rules, not a generic script.
  • Quote & RFP first drafts. Assembles a quote or a first-pass RFP answer from your product, pricing, and approved language — a draft a human signs, not an autonomous send.

Helping the rep

  • Account-manager call prep. Before a call, the agent assembles the account's history, open items, recent product and pricing changes, and the right play — a one-screen briefing, so the AM walks in ready instead of digging through the CRM and six tabs.
  • Discovery & customer prep. Research a prospect or customer from the CRM plus the memory layer: likely needs, who's who, the relevant case studies and proof points.
  • Live battle cards. The agent serves the current competitive battle card on demand. When a competitor moves, you update the memory layer once and every rep's agent has it the same hour — no stale slide deck making the rounds.
  • Objection handling on tap. A rep asks "how do I answer X?" and gets the approved talk track, sourced — the same answer across the whole team.

Enabling the team

  • New-rep onboarding & training. The agent is grounded in the playbook, product, and methodology; a new rep can ask it anything and role-play objection handling — ramping in weeks, not quarters.
  • Always-current enablement. Generate and maintain one-pagers, talk tracks, and training from the memory layer. Change the source once and everything downstream reflects it.
  • Win/loss & call capture (the interviews). After a win/loss interview or a key call, the agent helps capture the insight and file it into the memory layer, so the next rep benefits. The team owns and grows its own knowledge over time (see the producer idea in about-okf).

Why these beat the usual tools

  • Given, not searched. The rep — or the customer — is handed exactly what the moment needs, assembled up front, instead of being sent to hunt through a wiki, the CRM, and a shared drive.
  • One source, always current. Update the memory layer once; every agent, every rep, every customer touch reflects it. No more "which deck is the latest?"
  • Sourced and legible. Every answer traces back to something a person can read — the battle card, the playbook, the account record. No black box in front of a customer or a new hire.

(For the same set organized by funnel stage instead of by who's using it, see use-cases.)

This island is part of the chart — a map of what we know. AI visitors fetch it raw at /islands/sales-examples.md · updated 2026-06-24.

Want knowledge like this working for your own business?