How Anthropic's BDR Team Uses Claude Cowork to Automate Inbound and Outbound Sales Workflows

Anthropic's business development representatives (BDRs) have integrated Claude Cowork into their daily workflows to automate manual tasks such as account research and email drafting. Previously, reps spent up to 5 hours per day manually responding to inbound interest. Now, scheduled tasks and skills handle these duties, allowing the team to focus on strategic customer engagement.

Automating Inbound and Administrative Tasks

The inbound process relies on a sales knowledge base document that Claude reads before drafting replies. An hourly inbox skill scans for threads needing responses and drafts replies using the knowledge base and a personalized voice profile. Additional skills monitor Gmail and Google Calendar for meeting no-shows and use a CRM connector to draft personalized first touches for new leads. A pipeline scanner skill proposes Salesforce updates by cross-referencing internal guidance with activity in Gmail and Gong, requiring rep approval before changes are made.

Optimizing Outbound Prospecting

For outbound work, a scheduled task runs overnight to prospect across a rep's book of business, which can include upwards of 100 accounts. Claude connects to Salesforce, Apollo, Common Room, Gong, and the data warehouse to analyze account states and validate signals against the team's ideal customer profile (ICP). Reps receive a brief, score, and outbound play for each account in the morning. A call coach skill evaluates Gong transcripts against a discovery call playbook, providing scorecards with specific feedback.

Streamlining Ad-Hoc Requests

Claude Cowork also handles ad-hoc data requests, such as generating usage trend dashboards or spend analysis reports. A specific workflow identifies undiscovered usage by finding accounts with high product usage but no matching sales opportunity. For event outreach, Claude can sweep a rep's book to find accounts matching an ICP for specific webinars, such as the Claude Code for Data Engineering webinar.

Best Practices for Implementation

The team recommends building a knowledge base before creating workflows and providing Claude with examples of successful messages and writing styles. They maintain a human-in-the-loop approach where reps review, edit, and send all drafts. Skills are shared across the team via a shared plugin, and feedback from rejected drafts is recorded to prevent repeated errors.