Claude Cowork assembles sales account context
Sires reports packaging account-context skills and MCP connectors in a Claude Cowork sales plugin used at Anthropic, bringing customer evidence into pre-call research and talking points.
claude.com
Original prompt
Shared by the source author. See source notes for context.
/customer-context
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Anthropic interview with Jared Sires
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Sires reports packaging account-context skills and MCP connectors in a Claude Cowork sales plugin used at Anthropic, bringing customer evidence into pre-call research and talking points.
- Human involvement
- Sires builds and packages the skills; representatives use the evidence for customer conversations. The source does not establish automatic sends or an audited correctness rate for this task.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Install the sales plugin with the appropriate account connectors.
- Run the customer-context skill for the intended account.
- Retrieve CRM, support, meeting and communication evidence.
Jared Sires builds Claude-powered tools for Anthropic's sales team and packages skills and MCP connectors as a Claude Cowork plugin. A customer-context command assembles an account view across sales and communication systems instead of requiring a representative to inspect each application separately.
The plugin connects Salesforce, Intercom, Gong, Google Calendar, Gmail, Google Drive and BigQuery. Related daily briefing work reads the calendar, searches the people he is meeting and retrieves CRM context to prepare talking points. The source identifies installed use by the sales organisation and a customer-context skill returning its account view in roughly ninety seconds; that timing and adoption are the company's own reports, not an independent benchmark.
The specific agentic task is account preparation through tool retrieval and synthesis. Sires's earlier Gmail reply application is a separate software-building story and does not establish automatic sends by this plugin. The useful workflow makes customer context accessible before a call and keeps live connected records alongside public research. The account does not provide an audited correctness rate, a full command implementation or a measured increase in closed deals. A representative must still judge the evidence and decide how to use it in the customer conversation.
Prompt
Original prompt
Source: Blog · Anthropic interview with Jared Sires
Frequently asked questions
Which systems contribute context?
The source lists Salesforce, Intercom, Gong, Google Calendar, Gmail, Google Drive and BigQuery in the sales plugin.
Is the ninety-second timing independently measured?
No. It is the company’s reported time for its customer-context skill.
Does this case establish automatic outreach?
No. The selected task is account-context preparation; the article’s separate reply application should not be treated as proof of plugin sends.
More details
Remaining steps
- Combine connected evidence with pre-call public research.
- Review talking points before the customer conversation.
About this account
Claude Cowork uses packaged MCP connections and skills to retrieve customer context across systems and synthesise an account view for sales preparation.
Firsthand account by Anthropic interview with Jared Sires; original publication date verified against Primary article June 5, 2026. Outcomes are the author’s reports, not independently reproduced. Original installed sales-plugin command shared in the interview; the full skill implementation is not disclosed.
Credit: Anthropic interview with Jared Sires
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Cowork · MCP connectors · Salesforce · Intercom · Gong · Google Workspace · BigQuery
Account preparation · Connected customer evidence · Sales plugin
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.