Open-source pipeline that crawls Instagram, YouTube, TikTok, X and transforms unstructured social data into queryable Knowledge Graphs with LLM-powered entity extraction.
Docker or local install. Only one API key needed to start.
Get the code and add your OpenAI API key.
git clone https://github.com/Intellirim/ontix-universal.git cd ontix-universal cp .env.example .env # Edit .env — minimum: OPENAI_API_KEY
One command spins up Neo4j + Redis + the app. No local installs needed.
docker-compose up -d ✓ Neo4j running on :7474 ✓ Redis running on :6379 ✓ ONTIX API running on :8000 ✓ Web UI running on :3000
Open http://localhost:3000 for the full dashboard — or use the API directly.
Crawl a social media account, or upload a CSV file you already have.
# Option A: Crawl Instagram (needs APIFY_TOKEN in .env) curl -X POST http://localhost:8000/api/v1/pipeline/run \ -H "Content-Type: application/json" \ -d '{"brand_id":"demo","platform":"instagram", "target_type":"accounts","targets":["nike"],"max_items":20}' # Option B: Upload a CSV (no Apify needed) curl -X POST http://localhost:8000/api/v1/pipeline/upload \ -F "file=@reviews.csv" -F "brand_id=demo" {"job_id":"a1b2c3","status":"pending","rows_detected":150}
Your Knowledge Graph is built. Now ask anything about your data.
curl -X POST http://localhost:8000/api/v1/chat \ -H "Content-Type: application/json" \ -d '{"brand_id":"demo", "message":"What are the most discussed topics?"}' {"message":"Based on your data, the top 3 topics are..."}
The generic pipeline works out of the box. Industry Packs unlock domain-specific intelligence.
Instagram, YouTube, TikTok, X — or upload CSV/JSON from any source. 60+ column names auto-detected.
GPT-5-mini reads every post and extracts structured entities and relationships. Generic prompt included free.
Entities and relationships are stored in Neo4j. Query with Cypher, visualize connections, find patterns.
Built-in RAG chatbot answers questions about your data with retrieval-augmented generation and 4-layer quality filters.
With Industry Packs: Extract domain-specific entities (ingredients, programs, IPs) instead of generic keywords.
With Industry Packs: 10-12 ready-to-use Cypher queries for competitive analysis, ROI tracking, and trend detection.
Each pack includes extraction prompts, brand configs, and production-ready Cypher queries tailored for your industry.
Ingredient analysis, routine tracking, and creator influence measurement for cosmetics and skincare brands.
Menu intelligence, location performance, and creator-driven foot traffic analysis for restaurants and F&B brands.
Style trend tracking, body-inclusive fit intelligence, and resale value analysis for fashion brands.
Competitive intelligence, feature sentiment tracking, and developer migration analysis for software products.
Program effectiveness tracking, supplement intelligence, and transformation evidence analysis for fitness brands.
IP performance tracking, fandom intelligence, and revenue attribution for games, music, and media brands.
Get every pack and all future updates. Perfect for agencies serving multiple industries.
Domain-specific entity types and relationship definitions. Drop-in replacement for the generic prompt — immediately improves KG quality for your industry.
Optimized retrieval, filter, and generation settings. Temperature, trust thresholds, and top-k values tuned for your domain's data characteristics.
10-12 production-ready Cypher queries with explanations. Covers competitive analysis, ROI tracking, trend detection, and creator performance.
The full pipeline — crawling, transformation, KG generation, RAG chatbot — is MIT licensed. Industry Packs are optional add-ons for production-grade results.
Star on GitHubDownload the ZIP, drop it into your project, and restart. Takes 30 seconds.
After purchase, you'll receive a ZIP file via email. Extract it:
unzip ontix-beauty-pack.zip -d packs/beauty/
Each pack contains 3 files:
packs/beauty/ ├── extraction_prompt.txt # Domain-specific entity definitions ├── brand_config.yaml # Optimized retrieval & filter settings └── sample_queries.md # 10-12 production Cypher queries
Set the pack name in your .env or API call, then restart the server:
# In .env
INDUSTRY_PACK=beauty
# Or via API
curl -X POST http://localhost:8000/api/v1/pipeline/run \
-d '{"brand_id":"demo", "industry_pack":"beauty", ...}'
# Restart to apply
docker-compose restart app
Open sample_queries.md and try the included Cypher queries in Neo4j Browser:
# Example: Top ingredients mentioned across all brands MATCH (p:Product)-[:CONTAINS]->(i:Ingredient) RETURN i.name, count(p) AS mentions ORDER BY mentions DESC LIMIT 10
Each pack contains 3 files: an extraction prompt with domain-specific entity and relationship definitions, a brand_config.yaml with optimized retrieval/filter/generation settings, and 10-12 production-ready Cypher queries with explanations. No code changes needed — just drop the files in and go.
No. Industry Packs are pure configuration. Extract the ZIP into your packs/ directory, set the pack name in .env, and restart. The pipeline automatically uses the pack's prompts and settings.
The ONTIX Universal core is MIT licensed — fully open source, use it however you want. Industry Packs are sold under a per-team commercial license: one purchase covers your entire team, unlimited projects, forever.
Yes. All purchases include lifetime updates. When we improve a pack's prompts or add new queries, you'll receive the updated ZIP via email at no extra cost.
Yes. We offer a 14-day money-back guarantee. If the pack doesn't work for your use case, email us and we'll refund you — no questions asked.
Contact us at hanso@intellirim.com. We can build custom entity schemas, relationship definitions, and queries tailored to your specific domain.
The All 6 Packs Bundle saves you 47% ($249 vs $474 individually). It's ideal for agencies or teams serving clients across multiple industries. You also get all future packs added to the bundle for free.