Questions teams actually ask.
Grouped for the people evaluating HexaHQ: the business and ops view, the technical and IT view, and the bridge, one technical person setting it up for a non-technical team.
Business & ops
Can I use my AI to automate a workflow?
Yes. HexaHQ connects your AI client to your real tools and runs the multi-step work end to end, then saves it as a reusable workflow your team can run with one sentence.
Will my AI automation still work without the person who built it?
Yes. A HexaHQ workflow lives in your organization, not one person's chat history. Teammates run it on their own accounts, so it does not evaporate when someone leaves.
How do I turn a one-off AI report into something my whole team runs every week?
Do it once in a conversation. HexaHQ captures it as a workflow in your voice and format; after that it is one sentence, for you and anyone you share it with.
Can non-technical teammates use it?
Yes. One technical person connects the tools once. Teammates click a link, sign in with their own account, and run the workflows. No JSON, no setup.
How do I disconnect a tool or delete my data?
Open Connectors in the Hub, select the tool, and use the Manage menu to clear your saved sign-in. To remove what HexaHQ holds, send a privacy request. See Managing and deleting your data for what disconnecting, revoking, and deleting each do.
Is my data used to train models?
HexaHQ does not train models on your data. Content is processed only to resolve and run the tasks you invoke. See the Privacy Policy for sub-processors and data handling.
Technical & IT
Isn't this just MCP?
HexaHQ is an MCP server, the one that makes the rest unnecessary. Raw MCP is single-player and stateless: a config file per person, limited to tools that ship a server. HexaHQ is one endpoint for the whole team, turns any API into a tool, and remembers the workflows. See Skills vs MCP vs workflows.
Can I turn any REST API into an MCP tool without writing code?
Yes, and you don't need to be a developer. Describe the API in plain words and your AI authors the capability for you; there is no server to build or host, and nothing to code. Tools that already ship an MCP server connect directly. More on Connect anything.
Isn't this just Zapier, or an MCP marketplace?
No. Zapier and similar tools reach a big but fixed catalog of prebuilt apps. That list sounds unbounded, but an app or internal tool that isn't on it is still a dead end. HexaHQ doesn't rely on a catalog at all: if a tool has an API, your AI authors the connector on the spot. And each action runs as the specific person who ran it, with an allow / ask / deny check before each write. See HexaHQ vs Zapier & n8n.
Isn't this what Stainless or Anthropic does?
No. Stainless and tools like it generate an MCP server from an API's spec, but someone still has to be a developer, run the generator, and host the server, and only the team that owns the API can do it. That never reaches the internal or niche tool whose owner has no reason to build a server. HexaHQ needs no server and no developer: describe the API and your AI authors the connector for you, right now.
How do you keep MCP credentials out of chat and out of git?
Credentials are entered out-of-band in the browser and held server-side. No key transits the chat, and there is nothing to commit to git. All access routes through one credential seam. See Security & governance.
Can a Claude Teams admin assign a connector to specific members?
Not in Claude today; it is an open feature request. In HexaHQ a connector is shared once but keyed per person, so each teammate connects with their own credential and acts as themselves. See HexaHQ vs Claude Teams.
What is the difference between a Claude Skill and a HexaHQ workflow?
A Skill shapes how the AI reasons; it holds no credentials, gates nothing, and leaves no audit trail. A HexaHQ workflow governs what the AI is allowed to do in your systems and records every action.
Does Claude's org Capabilities toggle control individual connectors?
It toggles whole capability classes on or off org-wide, and its network-egress allowlist explicitly does not apply to MCP connectors. HexaHQ governs the individual capability with an allow / ask / deny policy and a per-action audit trail.
Setting it up for a team
If I build a workflow for myself, can a teammate run it without seeing my credentials?
Yes. Share the workflow and they run it on their own credential. They never see yours; the approval gate and audit trail carry each person's own identity.
If I share something through HexaHQ, can the other person see everything my AI knows about me?
No. They see only the specific HexaHQ documents or Collection you share. They cannot see your AI's chat history or memory, such as your ChatGPT memory, your other HexaHQ knowledge, or your connected accounts. HexaHQ lets you review exactly what is shared and who can see it.
Can one technical person set up AI tool access for a whole non-technical team?
That is the intended path. The technical person connects the tools and authors the capabilities once; everyone else clicks a link and runs the workflows as themselves.
Which AI clients does HexaHQ support?
Use HexaHQ through ChatGPT, Codex, Claude Chat, Claude Code, or Claude Cowork today. Any other MCP-capable client can point at the same endpoint.
Still have a question?
Start free and see it on your own tools, or ask us directly.