Frequently Asked Questions
What is the Flowkon MCP connector?
It's a hosted bridge, built on the open Model Context Protocol, between your Flowkon workspace and any MCP-speaking AI client - Claude, ChatGPT, or your own. It holds no database and no business logic of its own: every tool call is a live, scoped request to Flowkon's own API, so your assistant always sees exactly what's in your workspace right now.
How do I connect it?
Add https://mcp.flowkon.io/mcp as a custom connector in your AI client's settings, then sign in on Flowkon's own hosted login page - email, password, and 2FA if you have it enabled. Your credentials go straight to Flowkon; they never pass through the model or the chat window.
Which AI assistants work with it?
Claude and ChatGPT both support the hosted connector out of the box, and any other MCP-compatible client can connect the same way over OAuth. Developers can also run the same server locally over stdio for Claude Desktop, Claude Code, or the MCP Inspector.
What can my AI actually do with it?
54 tools spanning lead search and LinkedIn search, campaign creation/stats/flow-building, inbox and messaging, workspace and team management, tags, placeholders, integrations and booking. On top of that, 5 live resources (your workspaces, connected LinkedIn accounts, dashboard stats, a lead, a campaign) and 4 guided prompts that chain several tool calls into one ready-made workflow, like a campaign stats report or a workspace health summary.
What stops it from doing something I didn't ask for?
Almost every tool is read-only and runs immediately. The two that change something real - creating a campaign and pausing or resuming one - always come back first with exactly what would happen and wait for your explicit yes before anything is applied.
How is authentication handled?
The hosted connector runs a real OAuth 2.1 flow, resolving each connected client's own Flowkon identity on every single call - so one server safely serves many different Claude or ChatGPT users at once, each acting only as themselves. Self-hosted locally, you authenticate once through a small local browser login window, and that session is reused on every restart - no password is ever typed into the chat.
Is my data secure?
The server is a thin, typed wrapper around Flowkon's existing REST API - it keeps no copy of your data and Flowkon's backend stays the single source of truth. Every request is scoped to the workspace you're signed into, and the tools that change something require your explicit confirmation first.
Can I self-host it instead of using the hosted connector?
The server is a thin, typed wrapper around Flowkon's existing REST API - it keeps no copy of your data and Flowkon's backend stays the single source of truth. Every request is scoped to the workspace you're signed into, and the tools that change something require your explicit confirmation first.
What is an MCP connector for LinkedIn outreach and why does it matter?
MCP (Model Context Protocol) lets AI assistants like Claude and ChatGPT connect directly to Flowkon's data. Instead of copying campaign stats manually, your AI assistant can read live prospect lists, reply rates, and sequence performance to surface insights or draft follow-ups in real time.
How do I connect ChatGPT or Claude to my Flowkon campaigns?
Flowkon's MCP server exposes campaign data, prospect lists, and inbox activity as tools any MCP-compatible AI client can call. Connect Claude Desktop or ChatGPT with an MCP plugin, point it at your Flowkon endpoint, and your AI assistant can query sequences, draft messages, or flag stalled leads without leaving the chat interface.
How can an AI assistant prepare me for a LinkedIn sales call?
Using Flowkon's MCP integration, your AI assistant reads the prospect's campaign history — which messages were sent, what they replied, how long they took to respond — and generates a briefing. You get context on their engagement pattern, suggested talking points, and any open questions from the conversation thread, all before you dial.
Is connecting an AI assistant to outreach data safe and compliant?
Flowkon's MCP server uses authenticated, read-scoped tokens. Only the tools you explicitly expose are accessible, data stays within your Flowkon account, and no prospect information is shared externally. For teams operating under GDPR or UK GDPR, access is logged and revocable at any time.
What is Flowkon's MCP server and how does it help sales automation?
Flowkon's MCP (Model Context Protocol) server exposes campaign data, prospect lists, sequence steps, and inbox replies as callable tools for any MCP-compatible AI client. Instead of exporting CSVs or copying stats, your AI assistant reads live Flowkon data on demand — enabling use cases like AI-drafted follow-ups, stalled lead alerts, and real-time campaign analysis inside a chat interface.
How do I use Claude as an AI assistant for LinkedIn outreach via Flowkon's MCP?
Install the Flowkon MCP server into Claude Desktop or another MCP-compatible Claude client. Once connected, Claude can call Flowkon tools to list active campaigns, read reply rates, pull a prospect's message history, or draft a personalised follow-up based on their engagement pattern — without you switching out of the Claude interface.
Can ChatGPT connect to Flowkon for LinkedIn campaign management?
Yes — any MCP-compatible ChatGPT plugin or GPT action can call Flowkon's MCP server. Configure the server endpoint and authentication in the plugin settings, and ChatGPT gains access to your campaign metrics, prospect lists, and inbox threads as live tools. This lets you query 'which leads haven't replied in five days' or 'draft a follow-up for this prospect' directly in the chat.
What is an MCP server for B2B sales outreach and which tools support it?
MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools and read live data. A sales MCP server exposes actions like 'get campaign stats', 'add prospect to sequence', or 'read reply history' as callable functions. Flowkon ships a native MCP server — one of the first LinkedIn outreach platforms to do so — making it directly compatible with Claude, ChatGPT, and any MCP-capable AI agent.
Can an AI agent run or manage LinkedIn outreach campaigns automatically?
With Flowkon's MCP server, an AI agent built on Claude or ChatGPT can read campaign performance, identify underperforming sequences, suggest message edits, and flag prospects who haven't replied — all through tool calls rather than manual exports. The agent doesn't launch campaigns autonomously (a deliberate safety boundary), but it handles the analysis and drafting layer, freeing reps for high-value conversations.
How does Flowkon's MCP integration fit into a broader CRM and sales workflow?
Flowkon's MCP server sits between your AI assistant and your outreach data. The AI reads live campaign and prospect information from Flowkon, cross-references it with CRM context (via a separate CRM MCP connector), and produces insights or draft messages that a rep can review and act on. This creates a connected AI layer across outreach and CRM without moving data manually between systems.
How do I use AI to personalise LinkedIn messages at scale without sounding generic?
Flowkon's AI personalization generates message variations from dynamic fields — role, industry, recent post, company news — so each message reads as individually written. For deeper personalisation, the MCP integration lets an AI assistant like Claude read a specific prospect's engagement history and draft a follow-up that references exactly what they've interacted with.
.webp)
