# Criticality / IdeaGenerator

> Agent-facing counterpart to the [human project page](/projects/criticality-ideagenerator/).

## Record metadata

- Record: 036
- Slug: criticality-ideagenerator
- Domain: Business
- Domain code: BUS
- Type: Product lesson
- Status: Archived
- Period: 2026
- Portfolio role: Archived experiment
- Publication state: Public retrospective
- Case-study readiness: Draftable
- Compendium edition: 0.4

## Summary

An abandoned project-idea feed whose failure led to a better question: can agents find grounded problems worth solving?

## Overview

Criticality began with an attractive abundance problem: generate a stream of project ideas and present them in a feed so builders never run out of things to make. It was stopped when abundance became the problem rather than the solution.

A TikTok-style feed of generated project ideas, now retained mainly as a record of why unconstrained ideation can become a slop generator.

Purpose: Preserve the lesson from an idea generator that began producing volume more readily than value.

## The problem behind the project

The original product sought to help builders discover unusual projects, but its narrow audience and low grounding made more output feel less useful.

An idea can sound novel without identifying a real beneficiary, a verified need, or a path to evidence. As the feed produced more polished concepts, it began to resemble a slop generator for a small audience instead of an engine for useful work.

Open-source contributors and learners were the intended audience. They lose time when polished ideas are detached from real people and evidence.

## How it took shape

A feed-oriented concept and partial implementation using AI-generated project prompts and lightweight evaluation.

The partial product explored AI-generated prompts, a feed interaction, and lightweight evaluation. Josiah's most important contribution was deciding not to confuse output volume with value and reframing the underlying opportunity.

Josiah originated the product, evaluated its outputs, recognized the quality failure, and reframed the underlying opportunity.

A partial project exists; it was intentionally stopped before becoming a meaningful product.

## What the project means now

A stronger successor would send agents toward reality: investigate problems experienced by actual people, publish only a few opportunities, and attach sources, constraints, beneficiaries, and a falsifiable first step. The archived project records why grounding must precede ideation.

Generated novelty is not demonstrated usefulness, and an endless feed can optimize for engagement rather than impact.

The stronger successor is an agent that investigates real problems for real people and publishes a few evidence-backed open-source opportunities.

If revived, build a daily public-good research agent whose ideas include a beneficiary, evidence, constraints, and a falsifiable first step.

## Publication and interpretation notes

- Current classification: Archived
- Portfolio readiness: Draftable
- Publication boundary: Public retrospective

## Additional agent context

Present the abandonment and slop concern as central evidence, not as an embarrassing footnote.

## Related project records

- [OpScore](/projects/opscore/llm/) — A concept for interviewing a business owner and turning operational context into a concrete AI implementation guide.
- [Forever Habits](/projects/forever-habits/llm/) — A small app experiment used to study how easy cloning has changed the economics of consumer software.
- [Economy Map](/projects/economy-map/llm/) — A recursive knowledge graph and reasoning workbench for tracing how industries, firms, resources, and constraints connect.

## Navigation

- [Complete project index](/projects/llm/)
- [Human version of this record](/projects/criticality-ideagenerator/)
- [About Josiah's working method](/about/llm/)
- [Agent discovery map](/llms.txt)