The same ontology-backed intelligence system scoring companies across micro-drama entertainment and AI agent infrastructure. Same methodology. Same pipeline. Different dimensions. Compounding intelligence.
Every vertical we add makes every other vertical smarter. The ontology patterns transfer. The SPARQL queries port with namespace changes. The prediction engine accumulates accuracy. A single pipeline now produces intelligence across entertainment and enterprise technology from the same weekly cadence.
| Companies | 22 |
| Weeks tracked | 11 |
| Triples | 1,972 |
| Avg SBPI | 62.8 |
| Top scorer | ReelShort (88.0) |
| Tier distribution | 1 Dominant, 3 Strong, 5 Emerging |
| Prediction methods | 4 (68 active predictions) |
| Market signal | Bifurcating: platforms down, specialists up |
| Companies | 1,672 |
| Weeks tracked | 1 |
| Triples | 73,559 |
| Avg SBPI | 44.6 |
| Top scorer | Persana AI (69.0) |
| Tier distribution | 0 Dominant, 0 Strong, 80 Emerging |
| True agents | 677 (40.5% of AI companies) |
| Market signal | Pre-maturity: no breakout leaders yet |
Micro-drama has a clear leader (ReelShort at 88.0, Dominant tier). AI agents have zero companies above 70. The AI agent space is 11+ weeks behind in maturity signal accumulation.
60% of YC "AI companies" are tools, not agents. True agents score 5.6 points higher (47.9 vs 42.3). The SBPI dimensions detect structural capability gaps that marketing language obscures.
The biggest dimension anomalies in AI agents are Autonomy Depth spikes — companies with strong tech (Model Capability 85+) but low overall composites. Pattern: great AI capability, no agent architecture.
22 companies across the short-form drama vertical. 11 weeks of weekly scoring. Data from manual research + SPARQL-validated scoring.
| # | Company | Composite | Tier | CS | NO | DP | CoS | MI |
|---|---|---|---|---|---|---|---|---|
| 1 | ReelShort | 88.0 | Dominant | 95 | 90 | 90 | 85 | 80 |
| 2 | DramaBox | 79.8 | Strong | 85 | 80 | 80 | 75 | 80 |
| 3 | FlexTV | 72.0 | Strong | 75 | 70 | 75 | 70 | 70 |
| 4 | JioHotstar | 70.5 | Strong | 70 | 75 | 70 | 70 | 68 |
| 5 | COL Group / BeLive | 67.2 | Emerging | 65 | 70 | 70 | 65 | 66 |
| 6 | Disney | 65.0 | Emerging | 60 | 75 | 70 | 55 | 65 |
| 7 | CandyJar | 64.0 | Emerging | 70 | 60 | 65 | 65 | 60 |
| 8 | GoodShort | 62.5 | Emerging | 65 | 60 | 60 | 65 | 63 |
| 9 | ShortMax | 61.5 | Emerging | 65 | 55 | 65 | 60 | 63 |
| 10 | Lifetime / A+E | 59.8 | Niche | 55 | 65 | 60 | 55 | 64 |
CS = Content Strength, NO = Narrative Ownership, DP = Distribution Power, CoS = Community Strength, MI = Monetization Infrastructure
677 true AI agent companies from 1,672 AI companies classified across the YC ecosystem. First week of scoring — no weekly deltas yet.
| # | Company | Domain | Composite | Tier | MC | MT | PE | AD | CD |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Persana AI | Productivity | 69.0 | Emerging | 75 | 80 | 65 | 55 | 70 |
| 2 | Fiber AI | Software | 68.0 | Emerging | 70 | 75 | 70 | 55 | 70 |
| 3 | Warmly | Software | 66.0 | Emerging | 65 | 75 | 70 | 50 | 70 |
| 4 | Fini | Productivity | 65.0 | Emerging | 70 | 70 | 65 | 55 | 65 |
| 5 | Athelas | Health | 64.5 | Emerging | 60 | 75 | 60 | 55 | 73 |
| 6 | careCycle | Health | 64.5 | Emerging | 65 | 70 | 60 | 60 | 68 |
| 7 | Mutiny | Software | 64.5 | Emerging | 65 | 75 | 65 | 45 | 73 |
| 8 | Daily | Software | 63.2 | Emerging | 70 | 70 | 70 | 40 | 66 |
| 9 | Inkeep | Software | 63.0 | Emerging | 70 | 65 | 65 | 55 | 60 |
| 10 | QueryPie AI | Software | 63.0 | Emerging | 65 | 65 | 65 | 55 | 65 |
MC = Model Capability (20%), MT = Market Traction (25%), PE = Platform Ecosystem (20%), AD = Autonomy Depth (20%), CD = Capital & Defensibility (15%)
Agent density varies dramatically across domains. Transportation and Finance have the highest agent rates — workflow-heavy industries where autonomous operation delivers immediate value.
| Domain | Total AI | Agents | Agent % | Avg Score |
|---|---|---|---|---|
| Transportation | 34 | 21 | 62% | 46.1 |
| Finance | 172 | 104 | 60% | 44.9 |
| Productivity | 296 | 166 | 56% | 44.4 |
| Real Estate | 50 | 23 | 46% | 43.6 |
| Legal | 36 | 14 | 39% | 43.5 |
| Health | 222 | 82 | 37% | 43.4 |
| Software | 631 | 226 | 36% | 46.2 |
| Government | 20 | 6 | 30% | 41.0 |
| E-Commerce | 70 | 16 | 23% | 43.7 |
| Media | 78 | 13 | 17% | 40.6 |
| Education | 45 | 3 | 7% | 40.2 |
The dominant anomaly pattern in AI agents: high Model Capability but low Autonomy Depth. These companies have strong underlying AI technology but haven't built true agent architectures — they're AI tools wearing agent marketing.
| Company | Domain | Anomaly Dimension | Dim Value | Composite | Gap |
|---|---|---|---|---|---|
| Mica AI | Software | Autonomy Depth | 100 | 50.8 | +49.2 |
| Decisional AI | Productivity | Autonomy Depth | 90 | 46.8 | +43.2 |
| Yuma AI | E-Commerce | Autonomy Depth | 100 | 57.8 | +42.2 |
| ContraForce | Software | Autonomy Depth | 100 | 61.2 | +38.8 |
| Klavis AI | Software | Platform Ecosystem | 95 | 56.8 | +38.2 |
Both verticals share the same structural foundation. The system was designed so that adding a new vertical requires only: dimension definitions, ETL adapter, and namespace configuration. Everything else — queries, prediction engine, editorial generation — transfers directly.
| Component | Micro-Drama | AI Agents | Portable? |
|---|---|---|---|
| Ontology | sbpi.ttl (OWL 2) | ai-agent-sbpi.ttl (OWL 2) | Template |
| SHACL Shapes | sbpi-shapes.ttl (6 shapes) | ai-agent-sbpi-shapes.ttl (6 shapes) | Template |
| Namespace | sbpi: | asbpi: | Config |
| Dimensions | 5 (equal weight) | 5 (weighted: 20/25/20/20/15) | Per-vertical |
| Composite Formula | Weighted avg of 5 dims | Weighted avg of 5 dims | Identical |
| Tier Boundaries | 85/70/55/40 | 85/70/55/40 | Identical |
| ETL Pipeline | sbpi_to_rdf.py | yc_to_rdf.py | Adapter |
| SPARQL Queries | 11 .rq files | 12 .rq files | Namespace swap |
| Weekly Digest | Manual + nightly-insights.py | weekly_analysis.py | Template |
| Prediction Engine | 4 methods, 68 active | Not yet started | Direct port |
| Data Store | Oxigraph | rdflib (Oxigraph planned) | Identical target |
| Editorial Site | 7-tab IBM Plex theme | This page | Template |
Micro-drama has a mature competitive structure: one dominant player, a clear upper tier, and a long tail. AI agents have no dominant or strong players. The entire vertical is compressed into Niche tier (73.7%). This is a pre-maturity market — the equivalent of micro-drama in its first 2-3 weeks of tracking before leaders emerged. The first company to cross 70 will be the breakout signal.
Measured from the AI agent vertical buildout. These numbers are real, not projected.
| Phase | Work | Time | Artifacts |
|---|---|---|---|
| 1. Dimension Design | Define 5 domain-specific SBPI dimensions with weights and scoring rubrics | 1 day | Design doc + ontology extension |
| 2. Ontology | OWL 2 classes + SHACL shapes for the new vertical namespace | 2 hours | .ttl files (ontology + shapes) |
| 3. ETL Pipeline | Data source adapter → RDF triples with dimension scoring heuristics | 4 hours | Python ETL script |
| 4. SPARQL Library | Port existing queries with namespace swap + dimension-specific cross-correlations | 2 hours | 12 .rq files |
| 5. Weekly Digest | Generate markdown report from SPARQL results | 1 hour | Python digest generator |
| 6. Classification | Run 2-stage filtering on company dataset (depends on dataset size) | 1-4 hours | JSON classifications |
| 7. Editorial Site | Cross-vertical page update + vertical-specific tab | 2 hours | HTML page |
Total: 2-3 days per vertical. Most of the time is dimension design — the technical infrastructure ports directly.
Both verticals share the same 6-stage pipeline. Each stage is independently portable — a new vertical needs only Stage 1 (data source) and Stage 2 (dimension config) customized.
STAGE 1 Data Source → JSON/CSV/API (per vertical) │ STAGE 2 Classification → 2-stage filter: regex pre-filter → LLM structured output │ STAGE 3 ETL → RDF → Python + rdflib → Turtle (OWL 2 + SHACL validated) │ STAGE 4 SPARQL Queries → 12 .rq files (movers, anomalies, predictions, cross-correlations) │ STAGE 5 Weekly Digest → Markdown + JSON (auto-generated from query results) │ STAGE 6 Editorial Site → Self-contained HTML (Cloudflare Pages, global CDN)
| Vertical | Prefix | Namespace URI | Status |
|---|---|---|---|
| Micro-Drama | sbpi: | https://shurai.com/ontology/sbpi# | Production (Oxigraph) |
| AI Agents | asbpi: | https://shurai.com/ontology/ai-agent-sbpi# | Production (rdflib) |
| Next vertical | *sbpi: | https://shurai.com/ontology/*-sbpi# | Template ready |
| Query | Micro-Drama | AI Agents | Purpose |
|---|---|---|---|
| weekly-movers | Yes | Yes | Biggest delta changes week-over-week |
| tier-transitions | Yes | Yes | Companies crossing tier boundaries |
| dimension-anomalies | Yes | Yes | Dimension vs. composite gap detection |
| cross-correlation | Yes | Yes | Dimension co-movement patterns |
| anomaly-compound | Yes | Yes | Multi-signal overlap detection |
| predictive-signals | Yes | Yes | Momentum pattern recognition |
| attestation-coverage | Yes | Yes | Data quality governance |
| cross-vertical-summary | No | Yes | Standardized cross-vertical feed |
| domain-stack-rank | N/A | Yes | Per-domain leaderboard |
| agent-vs-tool | N/A | Yes | Agent/tool structural comparison |
| batch-cohort | N/A | Yes | YC batch year analysis |
| autonomy-spectrum | N/A | Yes | Autonomy depth distribution |
| platform-vs-pureplay | Yes | N/A | Platform vs specialist comparison |
| prediction-accuracy | Yes | Planned | Prediction method accuracy tracking |
| cross-graph | Yes | N/A | InfraNodus ↔ SBPI bridge |
| annotation-timeline | Yes | N/A | Manual annotation tracking |