The Problem
In geroscience, the volume of published research is growing exponentially — over 1.5 million papers indexed in PubMed under "aging" alone, with hundreds added daily. A researcher trying to stay current faces an impossible task: read, synthesize, connect findings across disciplines, identify gaps, and generate hypotheses — all while managing the mechanics of research: writing, formatting, grant applications, collaboration.
We asked ourselves: what if a researcher could have a copilot — an AI system deeply integrated into the research workflow, not just a chatbot, but an active collaborator that understands the full context of a project?
Enter Aubrey
Aubrey (named in honor of Aubrey de Grey, a pioneer in geroscience) is an AI-powered research copilot designed specifically for geroscience and longevity research.
Core Capabilities
1. Deep Literature Integration Aubrey ingests and indexes the full text of papers, not just abstracts. It builds a semantic knowledge graph connecting:
- Molecular mechanisms (senescence, autophagy, mitochondrial dysfunction)
- Interventions (senolytics, NAD+ precursors, caloric restriction)
- Clinical trials and outcomes
- Cross-species models
2. Multi-Phase Research Architecture Aubrey operates in three progressive phases:
| Phase | Focus | Status |
|---|---|---|
| Phase 0 | Foundation — literature retrieval, summarization, question answering | ✅ Complete |
| Phase A | Active research assistance — hypothesis generation, experimental design, data interpretation | 🔄 In development |
| Phase B | Autonomous research agent — designs and executes analysis pipelines | 📅 Planned |
3. Project-Aware Context Unlike general AI assistants, Aubrey maintains persistent project memory. It knows:
- Which papers you've already read
- Your current hypotheses
- The connections you've identified
- Your pending experiments and analyses
This means no more repeating context in every prompt — Aubrey remembers.
4. Multi-Model Orchestration Aubrey intelligently routes tasks to the optimal AI model:
- DeepSeek V4 for complex reasoning and writing
- Claude for nuanced analysis
- Gemini for broad synthesis
- Open-source models for routine tasks
Technical Architecture
┌─────────────────────────────────────────────┐
│ Aubrey Core │
├─────────────┬───────────────┬───────────────┤
│ Phase 0 │ Phase A │ Phase B │
│ (Retrieve) │ (Analyze) │ (Execute) │
├─────────────┴───────────────┴───────────────┤
│ Knowledge Graph │
│ (Papers ↔ Mechanisms ↔ Interventions) │
├─────────────────────────────────────────────┤
│ Model Router │
│ DeepSeek │ Claude │ Gemini │ Open-Source │
├─────────────────────────────────────────────┤
│ Tool Integration │
│ PubMed │ ORCID │ GitHub │ Document Gen │
└─────────────────────────────────────────────┘
Current State and Next Steps
As of May 2026, Aubrey Phase 0 is fully operational. It can:
- Search and retrieve papers from PubMed with semantic understanding
- Summarize findings across multiple papers
- Answer research questions with citations
- Maintain conversation history across sessions
Phase A is under active development with focus on:
- Hypothesis generation from literature gaps
- Experimental design assistance
- Grant proposal drafting
Call to Collaboration
We are actively seeking:
- Geroscience researchers interested in beta testing
- Collaborators for grant applications (EIC Pathfinder, SRNSFG)
- Developers with expertise in knowledge graphs and scientific NLP
- Institutions interested in deploying Aubrey for their research teams
The project is led by Georgia Longevity Alliance (Reg. №404506520), a non-profit organization dedicated to advancing longevity research in the Caucasus region and globally.
Get Involved
- 📧 Email: jaba@longevity.ge
- 📱 WhatsApp/Telegram: +995 555 185161
- 🔗 ORCID: 0000-0001-8651-7243
- 🏛️ Georgia Longevity Alliance