CSEP, Layers 1 & 2 · Space Governance & Positional Analysis

Layer 1: Sovereign Space Governance Model

A conversational AI trained on the complete space governance corpus, UN COPUOS proceedings, all major space treaties, ITU regulations, IADC guidelines, Artemis Accords, U.S. executive orders, and national space legislation. Think of it as having a space law expert, a COPUOS historian, and a policy analyst available 24/7.

How to use it now:

Click any of the 6 suggested query buttons to see the model in action
Type your own question about treaty law, national positions, precedent, or governance frameworks
The conversation maintains history, ask follow-up questions to drill deeper
Responses include structured analysis formatted for diplomatic consumption

With Judy (Production)

The prototype uses a general-purpose AI with a specialized system prompt. On Judy, the model is a 13B+ parameter foundation model actually trained on the space governance corpus, every COPUOS proceeding since 1959, every treaty drafting record, every national statement. The difference: the prototype knows about space governance; the Judy model understands it. It can identify when a proposed formulation echoes language from a 1974 working paper, or when a national position contradicts that nation's statement from three sessions ago. Training requires ~5,000 GPU-hours, only possible on Judy-class compute.

SO WHAT

A diplomat preparing for the Legal Subcommittee can query 60+ years of COPUOS proceedings in seconds instead of weeks. When another delegation introduces new language, the diplomat can instantly check whether similar formulations have been proposed before, what happened, and what the implications are. That's not just faster research, it's a structural advantage in a negotiation where historical precedent determines what's possible.

Layer 2: Positional Analysis Engine

An interactive coalition mapper that models how 12 nations are likely to align on 6 key space policy topics. Click a topic and instantly see which nations support the U.S. position, which oppose it, and which are swing states.

How to use it now:

Click any of the 6 policy topic cards to see the coalition landscape
Nations sort into three columns: Likely Support (green), Swing (amber), Likely Oppose (red)
Click any country card to see their full position profile across all 6 topics
Each country shows bloc membership, Artemis Accords status, and COPUOS statement count

With Judy (Production)

The prototype uses static position data for 12 nations across 6 topics. On Judy, a graph neural network trained on the complete COPUOS behavioral record models 100+ nations across dozens of issue areas, with probability distributions rather than binary predictions. The model learns from historical voting patterns, co-sponsorship networks, delegation composition, and geopolitical context, and updates continuously as new data arrives. It can predict not just where a nation stands today, but how their position is likely to shift in response to specific U.S. actions.

SO WHAT

Before proposing a new guideline at COPUOS, the U.S. delegation knows the coalition math. They know who will support them, who they need to persuade, what concessions might bring swing states on board, and where opposition is immovable. That turns diplomatic preparation from educated guessing into data-driven strategy.

Say-Do Gap Analysis

Identifies where nations' stated positions diverge from their observed behavior, a critical intelligence indicator for diplomatic engagement. When a nation says one thing at COPUOS but does another operationally, that gap reveals either negotiating flexibility or deception.

How to use it now:

Review 5 pre-built say-do gap cases: China, Russia, India, Brazil, and the EU
Each card shows "What They Say" alongside "What They Do" with a risk rating
Risk ratings: High Divergence (red), Medium (amber), or Opportunity (green, for positive surprises like Brazil signing the Artemis Accords)

With Judy (Production)

Judy continuously monitors the gap between rhetoric and behavior for all tracked nations, automatically flagging when new operational data (satellite maneuvers, regulatory filings, industrial procurement) contradicts stated positions. The say-do dashboard becomes a living warning system rather than a static snapshot.

SO WHAT

When a delegation claims to support debris mitigation while their space program continues practices that generate debris, the U.S. diplomat has that contradiction documented and ready. It's the difference between making an argument and proving a point.

Negotiation Wargame

Input a draft U.S. proposal and see how six key delegations (China, Russia, EU, Japan, India, Brazil) are likely to respond. The AI simulates each delegation's reaction based on their modeled positions and historical behavior.

How to use it now:

Type your own proposal or click one of the 3 sample proposals (5-year deorbit, commercial SSA working group, ISAM regulatory recommendations)
Click "Run Negotiation Wargame" to simulate delegation responses
Each delegation's response includes: position (SUPPORT/OPPOSE/MODIFY/ABSTAIN), key concern, and a simulated statement
A coalition assessment paragraph summarizes allies, opponents, and recommended strategy

With Judy (Production)

The prototype generates a single round of responses. On Judy, the wargame supports multi-round interaction, you can adjust your proposal based on simulated responses and re-run, creating branching scenario trees. Judy models second-order effects (if China opposes and India joins them, does Japan's position shift?) and runs Monte Carlo simulations across hundreds of negotiation pathways to identify which approaches have the highest probability of achieving U.S. objectives. Each simulation requires ~10 GPU-hours, practical on Judy, impossible elsewhere.

SO WHAT

Diplomats practice for negotiations the way military planners wargame operations, testing strategies against realistic adversary behavior before entering the room. A proposal that looked promising in internal discussion might fail the wargame because it triggers an unexpected coalition between China and India. Better to discover that computationally than at the COPUOS plenary.