The AI teammate
that does the work.

A Flowgenx Teammate is an AI teammate on its own account. Pick one of nine ready-made Teammates or configure your own — name, title, job description, the systems it may reach — and switch it on. Colleagues address it by name in Slack, Microsoft Teams, Google Chat, Discord or FlowGenX Chat, send it email, files and voice notes, and it also works on a schedule and on events when no one is watching. It remembers, turns good runs into skills, and leaves a replayable trace of every step.

A single-task agent waits for a prompt, lives in one chat window, answers, and starts from zero next session. A Teammate works on its own schedule, is present in your channels as a named teammate, performs multi-step work across 5,000+ tools, and gets better each run by turning good runs into skills. It is a role you staff, not a task you prompt — with the identity, permissions, and audit trail that implies.

Most AI coworkers are a closed surface: one assistant, in one place, on one vendor's model. A Flowgenx Teammate runs on the same engine, connectors and governance as your embedded workflows — every endpoint of 250+ apps, any API you import, your MCP servers, and your workflows and agents as tools — with your own models and keys if you want them. You can staff a team of them, each with its own role, under the same permissions and audit trail as everything else.

It distils its own tool trajectory — what worked and the dead ends — into a reusable skill with a tool allowlist, threat-scanned before it is saved to the library. One run is enough. Measured on live runs, an invoice status query went from 29 tool calls to 2, and a colleague-context request from 108 executions to 5. Four memory surfaces are read before every turn: semantic (what is true about your business), episodic (every run it has made), history (every tool call in order), and your knowledge base with citations.

Like a human teammate. Every tool is on or off, Auto or Ask first; it asks in the thread and the run parks until you answer — approve once, always allow for team, or deny. Its scope is inherited from the account's connections and never expanded; credentials are used, never read; PII and secrets are masked in results and answers; loops are halted; every run is traced and replayable; and one Pause switch quiets every channel at once.

Yes. Teammates use the same Enterprise MCP tool layer as the rest of FlowGenX, including built-in browser automation that drives real browsers through forms, portals, and legacy UIs.

It is a named member of Slack, Microsoft Teams, Google Chat and Discord, connected from one page in minutes, and it is also reachable in FlowGenX Chat. Work reaches it three ways: an @mention in a thread (or an email, a meeting action item, a photo or a voice note), a routine on a cadence, or an event such as a discount request landing in the CRM. Say "I want a routine" in any channel and it scopes, schedules and reports back there — cron, webhook or watcher.

Both run on the same engine, and Teammate Studio lives in the left rail of FlowGenX Chat. Chat is an assistant you drive: AI that works with you. A Teammate is a teammate you hire: AI that works for you, present in your channels, remembering, learning, acting unattended.