launchpad.hello18528.chatgpt.site

research_and_ideate

Run the same server-side AI web-research workflow as the page: search the web for the actual problem, retain citation-linked supporting and counter evidence, build one problem-specific recommendation, stress-test it, and finalize the visible blueprint. Agent-started runs pause for visible, version-bound human approval before qualifying evidence and again before finalization. A request for a specific live result may pause for essential missing details instead of inventing an answer.

Input schema

{
  "type": "object",
  "properties": {
    "problem_statement": {
      "type": "string",
      "description": "Problem to research. Omit when the user has already submitted it in the page."
    }
  },
  "additionalProperties": false
}