Stacks/Sales
LangChain's GTM Agent sales artifacts
LangChain's GTM Agent, on the Deep Agents harness, builds on-brand sales artifacts from Salesforce, call transcripts, and a shared Drive folder, rendering and self-reviewing each file in a sandbox. A one-clean-review stop rule cut a one-pager from 18 to 7 minutes. It produced 1,540 artifacts in 12 weeks, traced in LangSmith and judged by sales-leader feedback.
- Use case
- LangChain's internal GTM Agent building sales artifacts (account plans, proof-of-value docs, proposal decks, onboarding plans) from deal context, and suggesting the right one before customer calls
- Category
- Sales
- Team size
- not stated
- Orchestrator / framework
- Built on the Deep Agents harness. A rep asks in Slack (or the agent offers during meeting prep); the agent resolves the account in Salesforce, reads the shared Drive folder, asks for missing context, copies a leadership-owned template into its sandbox, renders via LibreOffice to PDF, rasterizes and self-reviews pages, then converts to Google Slides/Docs/Sheets. A stop rule ships after one clean review (a one-pager went from 18 to 7 minutes). Scheduled runs power pre-meeting briefs.
- Memory
- Memory that carries context across a deal; each account has one shared Google Drive folder the agent reads before every build. Skills hold the sales process, templates, and stage rules, versioned separately from code.
- Evals
- Human taste as the main signal: weekly meetings with each artifact owner, office hours, and Slack feedback. A scorecard tracks what happens to every artifact (sent externally, hours saved vs run cost): 1,540 artifacts in 12 weeks, 47% of deals with a sent artifact advanced vs 16% matched.
- Observability
- LangSmith tracing shows every render, tool call, and review step; it's how the team found the self-review latency problem.
- Cost (stated in writeup)
- Not stated
- Cost per output (stated)
- Not stated in the writeup
- WhatWeRun estimate
WhatWeRun estimate
Not enough info to estimate
Volume is stated (100+ artifacts a week) but the model and tokens per artifact aren't.
Reviewed Nov 13, 2026
- Models
- Observability tools
- Other tools
- SalesforceGoogle Drive / Slides / Docs / SheetsSlackLibreOffice in sandboxBrand kit baked into sandbox image
- Source
- https://www.langchain.com/blog/gtm-agent-high-quality-sales-artifacts
Nick Bray · Published Nov 10, 2026 · Last verified Nov 13, 2026
Related
More like this
temporal.ioTemporal's Kepler prospecting agent
An internal sales agent that watches buying signals, finds and vets the right contact, and drafts outreach sequences for BDRs and reps to edit and send
langchain.comLangChain's Content Studio design agent
An in-house design agent in Slack that turns a brief into on-brand videos, social graphics, recreated UI images, and slide decks, all written in code
Stated: ~50¢ per request
Production sales agents with thread-as-context (HN)
A production sales context: multi-turn conversations with real prospects who may reply days apart, covering research, drafting, sending, and parsing replies