---
title: "Workshop — Build a revenue agent"
description: "An invite-only, in-person Vasco x Claude workshop: deploy a revenue agent that reasons on your real data in one half-day."
canonical: "https://vasco.app/workshop"
---

# Build a revenue agent: a Vasco x Claude workshop

An invite-only, in-person pre-summit workshop. Your CRM may be live with Claude, but your context isn't. In four hours with the Vasco team you deploy a revenue agent that actually reasons on your data — live and ready to use by end of day.

[Apply for a seat](/request-demo)

- **99.5%** accuracy gained when Claude reasons on a Vasco Context Graph vs. raw CRM data.
- **4 hours** to a deployed, production-ready agent in one half-day session.
- **10–12 companies** per cohort, with 1:1 support in breakout rooms.

## Who it's for

For revenue operators at any stage of agentic adoption. Most B2B teams hit the same wall: the AI is connected but the data isn't ready — context is missing, quotas aren't encoded, and the agent confidently produces the wrong answer. This workshop fixes the foundation.

It's for you if you use HubSpot or Salesforce, own your revenue data and reporting, have seen AI produce a confident-but-wrong answer, and want to leave with something that runs the next day. Skip it if you're still evaluating whether AI fits your stack, don't have access to your CRM data, or want a demo rather than a build.

## The accuracy gap (why your stack lies to your agents)

Real examples from connecting HubSpot, Gong, Stripe, and Slack via MCP:

- Asked for a pipeline summary, Claude reported $1.2M; the actual number was $740K — no uncertainty flagged.
- Asked to rank rep performance, it put the wrong rep on top because closed-lost reasons weren't encoded.
- Asked to flag churn risk, it marked three accounts healthy; two churned within 30 days. The signals existed; the context to surface them didn't.

## Agenda

You don't spend the day setting up data — a free pre-event call connects your CRM, maps your stack, and scopes context beforehand, so the day is spent building:

1. **Setting the stage** — the accuracy gap, with five real scenarios where Claude had full CRM access and still produced the wrong answer; map your own stack and find the breaks. The fix: a Context Graph acting as the GTM brain for LLMs.
2. **The diagnosis** — a full GTM diagnostic with the Vasco team: funnel performance, channel/motion deep dive, and ICP discovery.
3. **Agent design sprint** — design and ship a custom agent for your specific revenue motion, with hands-on 1:1 support in breakout rooms.
4. **Cohort showcase** — present your agent (the problem it solves, the data it runs on, the before/after) and get judged.

## Facilitators

- **Justin Hudon** — Head of Sales & Customer Success at Vasco; a career across frontline sales, CS, and GTM leadership (Lightspeed), now translating what revenue teams struggle with into the Context Graph, metric definitions, and lifecycle logic that turns 21.9% accuracy into 99.5%.
- **Alec Oghassabian** — an architect of the agents you'll build; 7+ years in revenue infrastructure including six as Director of RevOps at Potloc (CRM architecture, forecasting, GTM alignment). Focused on fixing the data foundation before trusting the output.

## Locations & applications

Invite-only, competitive selection (priority to Boreal portfolio companies and existing Vasco customers); free; max 2 per company:

- **Toronto** — Monday, June 22, 2026
- **Montreal** — Thursday, June 25, 2026

More cities are being added after RevStar; not in a listed city? Join the virtual-cohort waitlist. [Apply now](/request-demo).

## What you leave with

- **Winner's prize** — two passes to RevStar Summit (2026 Toronto / 2027 Montreal); accommodation covered for Montréal attendees.
- **Judging criteria** — a clean data foundation, an agent that solves a real recurring GTM problem, and a clear before/after story.
- **Mandatory pre-session** — every accepted company completes a 60–90 min pre-workshop call (CRM connected, stack mapped, use cases scoped) so the day is spent building, not troubleshooting logins.
- **Opt-in case study** — winners can opt into a co-branded case study, sanitized and approved by you before publication.