Technical Capabilities
Crassus is built and operated by AENEA Global, a UK-registered technology and data intelligence company. We design, build, deploy, and maintain production systems end-to-end - from data infrastructure through to user-facing applications.
Five engineering disciplines
Each capability is demonstrated by a system currently in production, not a slide deck. Proven, not promised.
Production-grade applications built across the full stack. From schema design and API architecture through to user interface design and accessibility. Multiple deployed systems serving real users.
Original research in language model training, factual recall emergence, and tokenizer design. Practical experience deploying models into production for inference, summarisation, and intelligent automation.
Server provisioning, containerisation, continuous delivery, and operational monitoring. Production deployments tested across multiple Crassus systems on European cloud infrastructure, with engineering practices designed to keep total cost of ownership low and deployments repeatable.
API integration, ETL pipelines, real-time data processing, and multi-source aggregation. Crassus aggregates from World Bank, IMF, FRED, SEC EDGAR, FLORES-200, and 30+ news sources into a single coherent data layer.
Dashboard design, analytics platforms, and process automation tools. Designed for operational use by analysts, executives, and citizens - not just for demos. Mobile-first PWA architecture means access without installation friction.
Crassus systems are built on open standards and free public data APIs wherever possible. The result is dramatic cost reduction compared to proprietary terminals, no vendor lock-in for the host institution, and full reproducibility of analytical outputs.
Currently live and deployed
Systems running in production today - not pilots, not prototypes. The depth of these deployments demonstrates the engineering capability behind every Crassus engagement.
| System | Description | Endpoints |
|---|---|---|
| Crassus National Intelligence Terminal | Real-time intelligence platform across 250 countries with macro, commodity, news, weather, and climate data. Mobile-first PWA. | Live |
| Crassus V5.0 Financial Terminal | Full-stack FastAPI + Next.js financial analytics terminal with macro, equity, filings, and options modules. 20 technical analysis indicators. | 25+ |
| QT V.4.4 32K UltraLingo Tokenizer | Multilingual SuperBPE tokenizer trained in-house. Best cross-lingual equity ratio of any QT generation, beating Llama 3 across 11 of 12 script families. | Apache 2.0 |
| Prelude-5 Language Model | 276M-parameter language model in active training. Reaching factual crystallisation 5x faster than prior models on QT V.4.4 tokenization. | In Training |
| Crassus Research Reports | Eight published institutional research reports covering distressed credit, structured products, consumer credit cascades, and systemic risk. | 8 Live |
Every system listed above is currently deployed and accessible. We build production systems first, then offer them to clients - which means engagement discussions can begin with a working demonstration rather than a roadmap.
From these capabilities, the deliverables
Software-as-a-service licensing for the Crassus intelligence platform, with country and sector localisation.
Project-based development of bespoke digital services and analytical platforms tailored to a single client's requirements.
Consulting and integration services connecting Crassus or bespoke platforms with existing government or enterprise systems.
Deployment of language models, summarisation pipelines, and other ML components into client-facing or operational systems.
AI and data engineering training for client technical teams, with the goal of local maintenance and extension of deployed systems.
Institutional credit and macro research, both as public-facing publications and as bespoke analyses commissioned by clients.
Discussing a build
For institutions and governments considering an engagement - whether platform licensing, bespoke build, or integration - please get in touch.
Open Discussion