Superforms is human-input infrastructure for AI agents. An agent creates a structured form, sends the public link to someone outside the agent session, and receives the response asynchronously when that person is ready.
The product is for builders and teams using Codex, Claude Code, Claude, ChatGPT, Cursor, and other MCP-capable clients who need decisions, feedback, research answers, approvals, or missing context from real people.
Why Superforms exists
Agents are good at continuing work when the necessary context is already present. Real workflows often stop because the answer belongs to a client, customer, teammate, stakeholder, or subject-matter expert who is not inside the current chat.
Superforms makes that boundary explicit. Questions leave the agent as a clear, human-facing form. Answers return as durable data instead of disappearing in a message thread or requiring someone to copy and paste them back into the original session.
How the workflow works
- The agent drafts exact questions and response settings.
- The creator reviews a private draft and publishes it deliberately.
- External people answer through a public link without creating an account.
- Voice, text, and supported attachments are stored with the response.
- The owning agent watches, polls, or reconnects later to retrieve the answers.
- The agent presents raw submissions and asks before acting on them.
What is available today
Superforms supports agent-created forms, private draft review, public respondent links, multiple respondents, selective AI follow-ups, voice transcription, website embeds, durable response retrieval, OAuth-connected account history, MCP tools, and an authenticated HTTP API.
Public links can be shared through email, messaging, documents, or any other channel. Dedicated automated routing, escalation rules, and customer-configurable outbound webhooks are not currently exposed as product features.
Build with Superforms
Start with the Superforms developer portal, connect an agent using the setup guide, or explore practical workflows in use cases.