AI agents that see your book the way your team does. Custom agents when you need them.
Every agent runs on the same three-layer platform: your existing stack, reconciled data, AI agents.
Start with the one that solves your sharpest pain; expand as value compounds.
The four standard agents
Investment Risk Management
Investment risk, in the language you already use.
- Conversational risk exploration: "what's my JPY exposure?" → answer in seconds, no dashboard to hunt through
- Factor exposures, stress testing, what-if analysis
- 24/7 proactive monitoring with inbox alerts, daily and weekly briefings
- Works on your existing Bloomberg / MSCI / Axioma / FactSet risk data, or Medusa's native equity risk model when you don't have institutional tooling
- Full audit trail: every number traces back to the exact file, report, or feed it came from.
- Coming: hidden exposures. AI reads news and filings to flag the thematic bets in your book that your factor model can't see.
Portfolio Optimizer
Your quant co-pilot that briefs you every morning.
- Daily rebalancing suggestions that keep exposures on budget
- Allocation analysis with mandate-constraint tracking
- Scenario modeling for proposed trades
- Human-in-the-loop: it advises; you decide.
Client Reporting & Communications
Institutional-quality investor communication without a dedicated reporting team.
Client reports, RFPs, and investor questions pull the team off the portfolio.
Three use-case families on the same data layer:
- Scheduled client reports: performance, risk, position summaries on your schedule
- Inbound query handling: RFP responses for institutional investor questionnaires, investor Q&A drafts, IR / Client Portfolio Manager / Product Specialist requests
- Proactive commentary: market commentary, portfolio drift explanations, monthly investor letter drafts
Reconciliation
The platform becomes your system of record: break detection across prime brokers, fund admin, PMS, OMS, and your own books.
- Trade matching across your prime broker feeds
- Position reconciliation between your PMS, fund admin, and internal books
- Automated break detection with root-cause suggestions
- Daily P&L reconciliation (fund admin vs. internal)
When standard modules don't fit, we build on the same platform
For client-specific needs beyond the four standard modules, we build custom AI agents on the same platform data layer. Same infrastructure, same view of your book, client-specific logic.
Selected client engagements
Examples of how Medusa applies AI in real institutional settings.
CFA Society Switzerland
- AI member chatbot (trained on public content)
- AI email triage workflow (classification + routing, human-in-the-loop)
- Controlled rollout with governance and human oversight
From first call to production in weeks, not quarters
Demo & fit
Walk through your operational stack, identify the sharpest workflow, confirm fit.
Paid Pilot
6 wk boutique / 6–8 wk growth / 8–12 wk enterprise: working system on your real data, not a prototype.
Production retainer
Platform subscription + declining custom retainer as work standardizes.
Principles
- Pilots deliver a working system, not a prototype. The gap between pilot and production is configuration, not re-implementation.
- Pilot fee credited toward Q1 platform fee on conversion.
- Founder-led onboarding: Petr is in every engagement personally.
- Deployed in your cloud: Azure, AWS, or Google Cloud, or on-premises. We come to your data; your data does not come to us.
Pricing: From CHF 10K pilot + CHF 3.5K/month platform fee, replaces $100K–$500K/year of enterprise risk analytics. Larger firms: scoped on a fit call.