Who We Are
Occupant owns the judgment layer for AI systems. We measure costs, architect decision infrastructure, and help institutions govern autonomous decisions.
We operate across three integrated layers: Measurement (tracking AI costs and market activity), Interpretation (advising institutions on procurement and governance), and Infrastructure (building systems that make autonomous decisions legible and governable).
We've built working trust infrastructure for social platforms (Glowrm, PBJ, Leafroll) managing cross-application reputation, resource allocation, and accountability. The architectural patterns proven in social systems—decision receipts, judgment routing, authority chains, escalation triggers—apply directly to AI governance.
The Stack
Measurement
Compute CPI tracks price deflation across AI model tiers. AI Activity Index (AEAI) measures economic volume—token throughput, spend, and energy consumption. Updated daily from 2,300+ models. Public and free.
These indices prove we understand the terrain. The measurement layer demonstrates the depth of market knowledge required to advise on procurement decisions that hold up under scrutiny.
Interpretation
Advisory services for institutional AI buyers. Emergency procurement assessments for government CIOs facing contract signature deadlines. Strategic retainers for leadership teams building governance frameworks. Vendor credibility partnerships for organizations responding to RFPs demanding responsible AI proof.
Insurance against future harm. We reduce the cost of expensive procurement mistakes before institutional lock-in.
Infrastructure
Working systems governing decisions where identity travels and behavior matters:
- Glowrm — Cross-platform trust infrastructure. API at drydown.replit.app. Portable reputation (20-80 scale) travels between applications. Good behavior on one platform carries forward. Reputation follows users across contexts.
- SPA (Stratified Preference Allocation) — Methodology transforming infinite, low-cost actions into finite, high-cost decisions. Every action carries measurable opportunity cost. Tested across dating (PBJ), professional networking (Leafroll), and roommate matching (Roomies).
- Judgment Routing — Decision architecture for AI systems. Stateful middleware evaluating agent proposals against uncertainty, stakes, authority, and novelty. Every decision produces a receipt. Authority chains stay explicit. Escalation paths remain legible.
Systems design determines who bears costs and how costs distribute. Making invisible costs legible enables governance and accountability across social platforms and AI systems.
Who We Work With
Government CIOs and procurement leads facing AI contract signature deadlines.
Foundation and nonprofit executives deploying AI into services requiring liability frameworks and exit strategies.
Vendors building for government demonstrating responsible AI practices to win contracts.
Policy teams grounding regulatory frameworks in operational reality.
Platform operators building systems where identity travels and behavior matters.
What we offer: Advisory retainers, workshops for leadership teams, expert testimony, and infrastructure architecture. Start a conversation →
How We Operate
- Vendor independence — Direct compensation only. Zero referral fees.
- Public APIs — Infrastructure documented and accessible
- Auditable methodology — Indices use open, reviewable calculations
- Open research — Free indices. Published research.
Methodology
Compute CPI aggregates pricing data into a volume-weighted basket across four workload tiers. AEAI tracks economic activity through token volumes (60%), inferred spend (30%), and energy consumption (10%). Cross-validated across sources. Discrepancies flagged before publication. January 2025 = 100 for both indices.
| Index | Composition | Weighting |
|---|---|---|
| $CPI | Composite of all tiers | Volume-weighted |
| $AIU | Token volumes, spend, energy | 60/30/10 weighted basket |
| $BULK | Budget-tier models (<$1/MTok) | Equal-weighted basket |
| $FRONT | Frontier capability models | Top-5 by Arena ELO |
| $JUDGE | Reasoning-intensive workloads | o1/o3/R1 class models |
| $LCTX | Long-context workloads | 128K+ context models |
Build Cost Indices ($START, $AGENT, $THRU) apply fixed workload mixes to current tier prices. Track costs relevant to your specific use patterns.
Data Sources
| Source | Data Type | Update Frequency |
|---|---|---|
| OpenRouter API | Primary pricing | Daily |
| LiteLLM | Cross-provider validation | Daily |
| pricepertoken.com | Independent verification | Daily |
| OpenRouter Rankings | Volume / market share | Weekly |
| Chatbot Arena | Quality ratings (ELO) | Weekly |
Price discrepancies exceeding 10% trigger manual review before publication.
Data Access
All indices are public and free. Daily values publish by 09:00 UTC. Full historical data available via JSON API. Commercial users, government agencies, and researchers welcome.
Work With Us
Facing AI contract signature deadlines? Deploying autonomous systems requiring authority chains and decision receipts? Building platforms where identity travels but reputation stays local?
We work across measurement (tracking costs), interpretation (advising on procurement and governance), and infrastructure (building decision architectures making judgment legible). When institutional choices require sustained scrutiny, we talk.