Is MatrixFlows AI priced per agent, per resolution, or something else?
Neither. MatrixFlows AI is priced by company size, the same as the rest of the platform — not per agent, not per resolution, not per conversation.
Most AI agent vendors charge per resolution or add a metered AI tier on top of a per-seat base price, so the cost climbs as usage climbs — the more your agents actually work, the more you pay. That creates a strange incentive: teams start rationing which conversations get AI involvement to control cost.
Every MatrixFlows plan includes unlimited AI usage. Build as many agents and automations as your operation needs — internal agents for team productivity, external agents for customers and partners — without a per-resolution meter running underneath them.
What's the difference between an AI agent and an AI assistant?
An Agent is the configured worker; an Assistant is a surface an Agent powers. The distinction matters because "AI assistant" gets used to mean two very different things across the industry.
An Agent is built from instructions, context, tools, and skills — it's the actual configuration that determines what the AI knows, how it behaves, and what it's allowed to do. An Assistant — a chat widget, a voice interface, a reply-drafting panel in your inbox — is just where that Agent shows up. The same Agent can power a customer-facing chat assistant, a voice assistant, and a reply assistant inside your team's inbox simultaneously, with identical instructions and guardrails in all three places.
This is why editing one Agent updates every surface it powers at once, instead of maintaining separate configurations for your chat bot, your voice bot, and your internal reply tool.
Can AI agents take real actions, or do they only answer questions?
Both — but the actions are the part most AI tools stop short of. MatrixFlows agents can query records, create and update items, search your knowledge base, and call external tools, not just generate a text response.
A support agent that can only answer questions still leaves every action — creating a ticket, checking an order, escalating a case — to a human or a separate form. That's a chatbot layered on top of the real work, not a replacement for any of it.
Give an agent tools and it can complete the task end to end: update a record, create a ticket, escalate to the right person, call an external API. Your team reviews what ships externally — the agent does the routine work, the review keeps you in control of what customers or partners actually see.
Is there a limit to how many AI agents or automations we can build?
No — every plan includes unlimited AI usage, so there's no cap on the number of agents, automations, or conversations you run.
Tools that meter by resolution or by conversation force a tradeoff: deploy AI everywhere and watch the bill climb, or ration it to the highest-value use cases and leave the rest unserved. Neither is a real answer to "should we automate this."
Build as many agents as your operation needs — a support agent, a partner-facing agent, an internal IT agent, a sales-qualification agent — each configured separately with its own instructions and knowledge scope, without a usage counter deciding which ones you can afford to run.
Does MCP integration work with tools like Claude, Cursor, or our own internal systems?
Yes — MatrixFlows works as both an MCP server and an MCP client, so the connection runs in both directions.
As a server, your workspace's AI tools are exposed to external MCP clients — Claude Desktop, IDE assistants like Cursor, or your own custom apps — authenticated by personal access token or OAuth, governed by the same permissions the requesting user already has.
As a client, your agents can call tools on external MCP servers, plus reach 1,000+ apps through Composio connectors and any external API. Your own AI can build and manage the workspace from Claude or Cursor, and your agents can reach out into the other systems you run — the connection isn't locked to one direction.
.svg)
.svg)
.svg)
