Jupiter Intelligence

Climate risk has become a mandatory disclosure and underwriting concern for property and casualty insurers, asset managers, and government agencies — yet the data required to quantify physical risk exposure has historically been locked behind technical expertise. Jupiter Intelligence had built ClimateScore Global, one of the most rigorous climate science platforms in the world, covering all physical perils at any point on Earth's land surface. But querying its 400+ trillion data points required fluency in data schema, query structure, and climate science terminology. This created a two-tier access problem: data scientists could retrieve answers; executives, risk officers, and regulators had to wait for them. As climate disclosure mandates accelerated, analysis that should take minutes was taking days.

AE Studio built Jupiter AI, a natural language querying platform layered over ClimateScore Global. The core is a text-to-SQL pipeline powered by NLP that translates conversational questions — such as 'which of my assets have the greatest flood risk over the next 30 years' — into precise database operations across 25,000+ climate data elements at ~90-meter global resolution. The system simultaneously resolves query dimensions spanning geography, time horizon, peril type, and financial scope, then returns structured results in seconds. AE Studio also built the full output layer: domain-specific graphs, tables, and downloadable analytics formatted for regulatory submissions and investment reporting. A proactive follow-on query suggestion engine guides users toward insights they hadn't thought to ask. Enterprise-grade security scopes each user's access to their permissions, isolates portfolio data across organizations, and provides full auditability — a non-negotiable requirement for regulated insurers and government clients.

Jupiter AI launched publicly in June 2024 and won the 2025 AI Excellence Award in the AI for Social Good category. Key outcomes include:

• 400+ trillion climate data points now accessible via natural language to any enterprise user, regardless of technical background
• 25,000+ climate data elements queryable at ~90-meter resolution for any location on Earth's land surface, with risk projections through 2100
• 22.3 billion locations worldwide analyzed through the platform
• 95% of previously uninsured climate risk regions now covered through a UNDP partnership (Jupiter Promise), extending the platform to developing nations

Decision-ready outputs that previously took days now deliver in seconds, compressing the full journey from question to stakeholder-ready deliverable into a single interaction.

• Natural language interfaces compound the value of existing expert data systems — the bottleneck in P&C climate risk was access, not data quality; rebuilding the underlying science was never necessary.
• Text-to-SQL over complex domains requires deep domain modeling, not generic NLP; climate queries span geography, time horizon, peril type, and financial scope simultaneously.
• Proactive query suggestions are a core product requirement, not a bonus feature — users who don't know the right questions will miss critical risk exposures.
• Enterprise security for conversational AI must be architected from the start: per-user access scoping, cross-organization data isolation, and full auditability are non-negotiable for regulated P&C clients.