How to Manage Your Cap Table With AI Agents and MCP

# How to Manage Your Cap Table With AI Agents and MCP Managing a cap table with AI agents means using natural language commands — through tools like Claude Code, Cursor, or any MCP-compatible client — to create stakeholders, issue share classes, model dilution, and generate investor reports without clicking through a web UI. The Model Context Protocol (MCP) makes this possible by giving AI agents structured, authenticated access to your cap table data. This guide walks through what MCP is, how it connects to cap table management, and how to set it up with OpenCap Stack in under five minutes. ## What Is MCP (Model Context Protocol)? MCP is an open standard that lets AI assistants connect to external tools and data sources through a unified protocol. Think of it as a USB-C port for AI — instead of building custom integrations for every tool, MCP provides a single interface that any AI client can use. For cap table management, MCP means your AI assistant can: - Query your cap table directly ("Who owns more than 5% of the company?") - Create and modify stakeholders, share classes, and equity grants - Run dilution scenarios and waterfall analyses - Generate board-ready reports and export cap tables - Manage SAFE notes and track conversion terms All through conversational commands, with the same access controls and audit trail as the web interface. ## Why Cap Tables Need AI Agents Traditional cap table software requires founders to navigate complex UIs, manually enter data, and context-switch between tools. AI agents solve three problems: **Speed.** Creating a stakeholder, assigning a share class, and issuing an equity grant takes 15-20 clicks in a typical cap table tool. With an AI agent, it's one sentence: "Add Jane Chen as an advisor with 0.5% equity on a 4-year vest with a 1-year cliff." **Accuracy.** AI agents validate data as they enter it — catching issues like share counts that exceed authorized shares, vesting schedules that don't add up, or missing 83(b) election deadlines before they become problems. **Context.** When an AI agent has access to your cap table, it can answer questions that would otherwise require pulling data from multiple screens: "If we raise a $3M Series A at a $15M pre-money valuation, what will each founder's ownership look like after dilution?" ## Setting Up the OpenCap Stack MCP Server ### Step 1: Install the MCP Server ```bash npm install -g @opencapstack/mcp-server ``` ### Step 2: Configure Your AI Client Add the OpenCap Stack server to your MCP client configuration. For Claude Code, add to your `.claude/mcp.json`: ```json { "mcpServers": { "opencap": { "command": "npx", "args": ["@opencapstack/mcp-server"], "env": { "OPENCAP_API_URL": "https://api.opencapstack.com", "OPENCAP_API_KEY": "your-api-key" } } } } ``` ### Step 3: Start Using It Once connected, your AI assistant has access to 50+ cap table tools. Try: - "Show me the current cap table summary" - "Create a new Common share class with 10 million authorized shares" - "Add a SAFE note for $500K at a $10M valuation cap" - "Run a dilution analysis for a $2M seed round" ## What You Can Do With AI-Powered Cap Table Management ### Cap Table Operations | Command | What It Does | |---------|-------------| | "Show cap table summary" | Returns ownership breakdown by stakeholder and share class | | "Add a stakeholder" | Creates a new founder, investor, employee, or advisor record | | "Create share class" | Sets up Common, Preferred Series A/B/C, or options pool | | "Export cap table" | Generates a downloadable cap table in standard format | ### Equity and SAFEs | Command | What It Does | |---------|-------------| | "Create an equity grant" | Issues options or RSUs with vesting schedule | | "Add a SAFE note" | Records a Simple Agreement for Future Equity with terms | | "Calculate dilution" | Models ownership impact of a new financing round | | "Run waterfall analysis" | Shows payout distribution at different exit valuations | ### Compliance and Reporting | Command | What It Does | |---------|-------------| | "Check 83(b) status" | Shows which grants need 83(b) elections filed | | "Get latest valuation" | Returns the most recent 409A valuation data | | "Generate financial report" | Creates investor-ready financial summaries | | "List documents" | Shows all legal documents in the data room | ## Real-World Example: Modeling a Seed Round Here's what a conversation with an AI agent managing your cap table looks like: **You:** "We're raising a $2M seed round on a SAFE at a $10M post-money cap. We have two founders — Alex with 60% and Jordan with 40% — and an option pool of 10%. Show me what ownership looks like after conversion." **AI Agent:** The agent calls `calculate_dilution` and `cap_table_summary` through MCP, then returns: - Alex: 48.0% (from 60%) - Jordan: 32.0% (from 40%) - Option Pool: 8.0% (from 10%) - Seed Investors: 12.0% (new) "The $2M SAFE at $10M post-money cap creates 12% dilution across all existing shareholders. Alex drops from 60% to 48%, Jordan from 40% to 32%. Want me to add this SAFE to the cap table?" No spreadsheet. No manual calculation. No risk of formula errors. ## How OpenCap Stack's MCP Server Compares Most cap table tools offer APIs, but few support MCP. Here's the difference: | Feature | Traditional API | OpenCap Stack MCP | |---------|----------------|-------------------| | Setup time | Hours (auth, SDK, code) | 5 minutes (npm install + config) | | Interface | Code only | Natural language | | Who can use it | Developers | Anyone with an AI assistant | | Real-time queries | Build custom dashboards | Ask questions directly | | Dilution modeling | Build calculation logic | "What if we raise $3M at $15M?" | | Audit trail | Manual logging | Automatic via MCP protocol | ## Security and Access Control MCP connections to OpenCap Stack use the same JWT-based authentication and role-based access control (RBAC) as the web application. An AI agent can only access data that the authenticated user has permission to see. - **Founders** get full read/write access to all cap table data - **Investors** see their own holdings and company-level summaries - **Employees** see their own equity grants and vesting schedules - **Accountants** get read access for reporting and compliance Every action taken through MCP is logged in the activity feed, so you have a complete audit trail of what was queried or modified and by whom. ## Getting Started Today 1. **Sign up** at [opencapstack.com](https://opencapstack.com/register) (free) 2. **Install** the MCP server: `npm install -g @opencapstack/mcp-server` 3. **Configure** your AI client with your API credentials 4. **Start managing** your cap table with natural language The entire setup takes under five minutes. Your cap table data stays in OpenCap Stack's secure infrastructure — the MCP server is just the bridge between your AI assistant and your data. ## FAQ ### What is MCP and how does it work with cap tables? MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools. For cap tables, it means your AI agent can query ownership data, create stakeholders, issue equity, and run dilution models through natural language commands instead of a web UI. ### Is it safe to let an AI agent access my cap table? Yes. OpenCap Stack's MCP server uses the same JWT authentication and role-based access control as the web app. The AI agent can only access data the authenticated user has permission to see, and every action is logged in the audit trail. ### Do I need to be a developer to use MCP with my cap table? No. Once the MCP server is installed (one terminal command), you interact with your cap table through natural language in any MCP-compatible AI client like Claude Code or Cursor. No coding required. ### Which AI clients support MCP? Claude Code, Cursor, Windsurf, and any client that implements the Model Context Protocol. The ecosystem is growing rapidly — most major AI coding assistants now support MCP. ### Can I use MCP to generate investor reports? Yes. You can ask the AI agent to generate financial reports, cap table exports, waterfall analyses, and dilution summaries. The data comes directly from your cap table, so reports are always current. ### How is this different from using a cap table API? APIs require writing code to integrate. MCP lets you use natural language. Instead of building a custom integration, you install the MCP server and start asking questions. It's the difference between writing SQL queries and asking "show me all investors who put in more than $500K."

How to Manage Your Cap Table With AI Agents and MCP

Check your site's AX Score