MCP (Model Context Protocol) Support — Connect Knowby to AI Tools and Enterprise Systems

Enable Knowby to act as an MCP server, allowing AI assistants and agentic tools (like Claude, ChatGPT, and enterprise AI copilots) to securely access and interact with Knowby content and data. This would let organisations surface step-by-step guidance directly inside the AI tools their teams already use, and allow AI agents to query, retrieve, and act on operational knowledge stored in Knowby.

Why this matters:

As enterprises increasingly adopt AI assistants and agents into daily workflows, static integrations are being replaced by dynamic, standardised protocols like MCP. Supporting MCP would position Knowby as a live, connected source of operational truth — rather than a siloed guidance tool — allowing frontline knowledge to flow directly into ERP, ITSM, and AI-driven systems without custom point-to-point integrations. This also strengthens Knowby's position as a modern Enterprise Guidance Platform, in contrast to legacy competitors still reliant on static exports or manual integrations.

Potential functionality (high level):

• Expose Knowby content (knowbys, steps, media, metadata) via a secure MCP server

• Allow permissioned read access for AI agents to retrieve guidance relevant to a task or query

• Explore write-back capability (e.g. logging execution/completion data back to Knowby via an agent)

• Authentication and governance controls to ensure enterprise data security requirements are met

Help us decide how it should be priced

Before we lock in how this gets rolled out, we wanted to ask the people who'll actually use it.  There are a few ways we could approach this, and each comes with trade-offs:

Option 1 — Included for everyone
MCP access becomes a standard part of every Knowby licence. Simple and fair, but it likely means a modest increase to the base per-user price across the board, since everyone would be contributing to its ongoing cost even if they don't use it.

Option 2 — Optional add-on, flat fee
Only organisations who want MCP pay for it, as a fixed additional cost per company (not per user). Keeps the base price untouched for everyone else, but means predicting usage/value upfront rather than paying for what you actually use.

Option 3 — Optional add-on, usage-based
Only pay for what you use — e.g. based on the number of AI queries or actions run through MCP each month. Fairest for occasional or exploratory use, but less predictable for budgeting, and adds a bit of billing complexity.

We don't have a preference locked in yet — genuinely want to know what would feel fairest and most useful from your side before we commit to a model. If you've got a view or a completely different option, drop it in the comments below.