Product Trend AI

A personal Python application for combining product research, LLM analysis and human review.

Problem and approach

Product research often spreads across searches, spreadsheets and separate marketplace tools. This project brings trend collection, specialist-agent analysis and an approval queue into one application so a reviewer can compare opportunities and decide what to investigate next.

The repository contains a FastAPI backend, PostgreSQL data models, Celery jobs, Redis and a Next.js dashboard. Agent reports feed a coordinator that produces structured product recommendations.

Current status

This is a development prototype, not a verified production release. No measured revenue lift, commercial deployment, successful live marketplace publishing or repair success rate is claimed here.

• Trend discovery, agent analysis, review and monitoring code are present.
• The source configures Qwen3 32B, Llama 3.3 70B, Llama 3.1 8B and Perplexity Sonar identifiers. Provider availability and account access must be checked before running.
• Marketplace adapters include placeholder/demo responses. A returned demo listing ID is not evidence that a listing was published.
• Monitoring and repair components are experimental. Automatic rollback after failed validation is not implemented in the coordinator.
• Repair code can report success while individual actions were skipped or failed. End-to-end validation is required before relying on repair outcomes.
• A confidence threshold in code is a configuration setting, not a measured accuracy or success rate.

Older design notes may describe intended features more strongly. The source code and the limits above should govern any assessment of what currently works.

Source map

Area	Starting point
API and review workflow	backend/api/main.py
Agent configuration and analysis	backend/services/ai_analysis/agentic_system.py
Marketplace adapter scaffolding	backend/services/platform_integrations/platform_manager.py
Experimental repair coordinator	backend/agents/devops/autonomous_coordinator.py
Repair actions and result handling	backend/agents/devops/fix_engineer.py
Local service configuration	docker-compose.yml

Local development

Use an isolated development environment with Docker and Docker Compose. The repository also contains Python and frontend dependencies for local development.

1. Clone the repository and enter its directory.
2. Copy .env.example to a local .env and supply only the provider keys needed for the experiment. Never commit real keys.
3. Review docker-compose.yml before starting services. Backend and worker containers mount the host Docker socket, which can grant control of the host. The database configuration contains development defaults. Do not expose this stack to the internet or run it on a trusted production host.
4. Check configuration without starting services:

docker compose config --quiet

5. Only after reviewing the access and configuration, start the development stack on the isolated host:

docker compose up --build
docker compose exec backend python -c "from models.database import init_db; init_db()"

Configured local endpoints:

• Dashboard: http://localhost:3000
• Backend: http://localhost:8000
• API documentation: http://localhost:8000/docs

These instructions reflect repository configuration; the stack was not started or end-to-end tested during the documentation review.

Before treating this as a deployable service

• Replace marketplace placeholders with authenticated integrations and verify external publication independently.
• Add tests for approval enforcement, failed/skipped repair actions and rollback.
• Review authentication, dependency versions, rate limits and provider failure handling.
• Remove unnecessary host privileges and replace development credentials.
• Verify the data used for trend and pricing decisions, including provenance and freshness.
• Record evaluation results rather than treating agent confidence as measured performance.

Responsible use

Human review remains necessary before any external publication or business commitment. Use official marketplace APIs with approved access, and check each platform's current terms. This software does not guarantee policy compliance or provide legal advice.

This project is provided as-is for educational and business purposes. Users are responsible for ensuring compliance with all applicable laws and platform terms of service.