top of page
MCP
I have ideas I haven't even thought of yet
Be prepared for the next generation of investment and operation teams.
AI will reshape how your firm operates. You don't need to know exactly what for — you just need to be ready. ATHENA MCP is there when the ideas arrive.
Governed by design
Every AI interaction logged and traceable
Single-tenant
Your data environment is isolated
Read Only
ATHENA MCP retrieves. It never modifies.
Your AI client
Claude, ChatGPT & MCP-compatible tools
ARCHITECTURE
Three layers. One secure connection.
ATHENA MCP is designed so that each layer has a single, clear role — keeping your data protected, your AI client flexible, and your compliance team informed.
DESIGNED FOR TRUST
Governed at every layer. Auditable at every step.
Every layer has a single, clear role — keeping your data protected, your AI client flexible,
and your compliance team informed.
LAYER 1 - DATA
ATHENA Platform
Your authoritative source. Holdings, transactions, benchmarks, analytics, party master, and more. The MCP layer reads from ATHENA; it cannot modify data.
Read-only access
LAYER 2 - CONTROL
ATHENA MCP Server
The gatekeeper. Translates AI requests into structured queries, enforces your user permissions, and logs every interaction. Hosted on ICS Azure infrastructure within your tenancy.
Fully auditable
LAYER 3 - INTELLIGENCE
Your AI Client
Interprets your question, selects the appropriate ATHENA tools, coordinates data retrieval, and compiles a plain-language answer. Compatible with Claude, ChatGPT, and any MCP-compatible client.
Client agnostic
Five types of questions you'll want to ask.
The questions that matter most often arrive before you know how to ask them. ATHENA's MCP layer is ready when they do — spanning datasets, time periods, and domains simultaneously, in plain language, against your live governed data.
TEMPORAL
Questions that traverse time
Not 'what is X now' — but 'when did X happen, how often, and what else changed around it.' Requires date-series traversal across holdings, transactions, and analytics. The kind of question that only becomes possible when your data is truly live and connected.
What is the relationship between days where our fixed income crosses 25% of AUM — and the inflow/outflow of the fund that day?
CAUSAL
Questions that explain why
Separating signal from noise. Was performance driven by a decision or by market luck? Was a breach triggered by a trade or by drift? Multi-dataset reasoning to isolate the contributing variable.
Was the manager's outperformance skill — or did they just happen to be overweight in the best sector?
HIDDEN RISK
Exposure beyond a single report
Risk that exists in the data but has never been formally assembled into a view. Look-through aggregation, cross-portfolio overlap, FX decomposition — surfaced by joining data that has never been joined before.
Provide our real exposure to China including everything held both directly and indirectly via international managers, derivatives or related parties.
CROSS-DOMAIN
Insights that live at intersections
Answers that only exist when three or more datasets are joined. ESG controversy scores against position size against liquidation cost. The insight doesn't exist until the question is asked — and ATHENA is ready to assemble it the moment it arrives.
Which top-20 holding has the worst ESG profile — and what does it cost us to exit out of the position?
OPERATIONAL
Operational data as insight
Data pipeline health, NAV validation patterns, data quality events correlated with transaction volume. Converts maintenance data into a business diagnostic — turning a manual ops question into a governed, repeatable answer.
Provide an overview of all operational activities that have been delayed over the last quarter, including their reasons and service providers that breached their SLA.
From thought to dashboard in minutes
And that is just scratching the surface
WHAT IT IS
AI that understands your investment organisation
ATHENA MCP is a secure bridge between your AI assistant and your ATHENA investment platform. Ask a question in plain English — the AI works out what data it needs, calls ATHENA, and returns a structured answer. All within a normal conversation. Your data never leaves your environment.
You ask it questions
In plain English, through any MCP-compatible AI client — Claude, ChatGPT, or your own tooling.
The AI calls ATHENA
The MCP server translates your request into a structured query, enforces permissions, and logs every interaction for compliance.
You receive a structured answer
A plain-language response compiled from your authoritative investment data — with every step traceable and auditable.
Your data never leaves your environment
ATHENA MCP is read-only. All queries run within your Azure tenancy. Nothing is shared, cached, or used to train external models.
bottom of page