Build a Sales AI Agent (OpenClaw Guide)

What if your AI agent could pull a deal from HubSpot, generate a battle card for the competitor in the deal notes, and prepare a pre-call briefing before you finish your morning coffee? This guide walks through connecting OpenClaw to both the Battlecard MCP server (competitive intelligence) and the HubSpot MCP server (CRM data), then building three automations that run daily.

The Stack

**OpenClaw** is the agent runtime (orchestration). **Battlecard MCP** is the competitive intelligence layer (battle cards, objection handlers, briefings). **HubSpot MCP** is the CRM data layer (deals, contacts, meetings, notes). The agent sits in the middle, pulling deal data from HubSpot, identifying competitors, and requesting intelligence from Battlecard.

Setup

Step 1: Install Both MCP Servers

Install via OpenClaw CLI: `openclaw skill install battlecard-competitive-intelligence` and `openclaw skill install hubspot-crm`. Or configure both manually in `~/.openclaw/mcp.json` with your API keys and the server URLs (battlecard.northr.ai/mcp and mcp.hubspot.com/sse).

Step 2: Get API Keys

**Battlecard:** Sign up at battlecard.northr.ai/signup. Free tier gives 1 query to test. Starter ($49/mo) gives 50 queries/day for daily deal prep. **HubSpot:** Generate a private app access token from Settings > Integrations > Private Apps. Need read access to Deals, Contacts, Companies, and Engagements.

Step 3: Test the Connection

Ask your agent: 'List my open deals with meetings this week.' Then: 'For each deal with a meeting, check notes for competitor mentions and generate a battle card.' If both servers are connected, the agent pulls deals from HubSpot and calls Battlecard for each competitor found.

Recipe 1: Daily Deal Prep

Every morning at 7 AM, the agent scans your meetings, pulls deal context from HubSpot, identifies competitors in the notes, and generates a pre-call briefing for each competitive deal. The briefing is saved as a note on the deal. You review it over coffee and walk into every call prepared.

Recipe 2: Post-Call Intelligence Capture

After every call, the agent reads call notes from HubSpot and extracts competitive intelligence using Battlecard's capture_call_intelligence tool. It identifies competitors mentioned, objections raised, pricing discussed, and buying signals. Over time, Battlecard's get_field_intelligence aggregates patterns across all captured data.

This is the data flywheel: more captured calls create richer intelligence, which helps reps prepare better, which produces more data. See our Sales Battle Cards Complete Guide at /blog/sales-battle-cards-complete-guide.

Recipe 3: Competitive Alerts

The agent monitors deal stage changes in HubSpot. When a competitive deal moves to a risk stage (Negotiation, Proposal, or amount decreases), it generates fresh objection handlers and sends a Slack message to the deal owner with the handlers and a link to practice in a voice simulation.

Extending the Stack

• **Google Calendar MCP:** Direct meeting triggers. Briefing pushed 30 minutes before each call
• **Slack MCP:** Competitive alerts to #competitive-intel channel. Weekly summaries to manager
• **Attio MCP:** Swap HubSpot for Attio. Battlecard MCP is CRM-agnostic
• **Voice simulations:** After generating a battle card, send the rep a practice link. See our AI Sales Simulations guide at /blog/ai-sales-simulations-guide

Start Building

You need OpenClaw (free, open-source), Battlecard (free tier to start, $49/mo for daily use), and your existing HubSpot account. Install both skills. Run test queries. Set up Recipe 1 first: daily deal prep delivers value on day one.

Frequently Asked Questions