01 / Discovery.
Detect recurring problems in public conversations without reducing them to a market score.
Agentic search
Start with a topic, workflow, or frustration.
The search agent translates a natural-language request into filters, clusters, and follow-up questions.
Recurring patterns
Compare repeated friction, not one anecdote.
Related conversations are grouped so builders can inspect frequency, affected workflows, and current workarounds.
Source evidence
Keep the original material attached.
Open every source and carry its evidence IDs forward into Analysis and the proposed solution.
02 / Analysis.
Turn the evidence set into a precise, implementation-relevant Problem Case.
Problem structure
Separate causes, symptoms, and consequences.
AI produces structured JSON for the core problem, problem units, actors, workflows, alternatives, constraints, and unknowns.
Traceability
Require evidence for every important claim.
Problem units retain citations and provenance so an engineer can distinguish observed facts from inference.
Problem analyst
Ask implementation-focused questions.
Use the analyst to clarify the problem boundary and unresolved details without market-fit, pricing, or strategy evaluation.
03 / Design.
Convert the approved Problem Case into a solution that a coding agent can execute.
72-hour scope
One flow and no more than five features.
The Solution Architect generates explicit non-goals, acceptance criteria, architecture, APIs, data entities, and tests.
Build graph
Work in dependency-ready tasks.
Each task links to a feature and includes concrete acceptance checks, dependencies, and mutable progress.
LLM / IDE MCP
Send the complete build context to the IDE.
A project-scoped MCP server lets a coding agent read context, request the next task, and report progress through release.
04 / FAQ.
What the product does and deliberately does not do.
Does Immensity evaluate market fit?
No. It does not score market fit, pricing, positioning, market size, or business strategy. It helps builders understand a documented problem and ship an MVP.
Where does the evidence come from?
From public complaints, questions, workarounds, and requests grouped around recurring problems.
Can I inspect the original sources?
Yes. Discovery and Analysis keep source records and citations attached to the project.
What does the coding agent receive?
The approved Problem Case, solution scope, architecture, task graph, tests, release checklist, and the next dependency-ready task.
Ready when you are
Put the guide into practice.
Open Discovery and turn real problem evidence into working software.