Drakkar
The intelligence layer, made operational.
Drakkar unifies research, risk, and allocation on data your firm already owns. It retrieves the evidence, reasons over it, and proposes the decision. Your team approves.
Most finance AI answers questions. Drakkar prepares decisions.
A chat window over your documents is a research assistant. Drakkar runs the full loop and puts a proposed action in front of the person who owns the call.
One platform, not a stack.
Research, risk, and allocation share the same reconciled data and the same memory. No copy-paste between tools, no version drift between teams.
A loop, not a pipeline.
When evidence is thin, Drakkar re-queries instead of guessing. Every pass is grounded in sources, not the model’s recall.
Five modules, one memory.
Research
Agentic research workflows: signal candidates, factor decomposition, and thesis assembly, validated out-of-sample and delivered with sources attached.
Research workflows in detailRisk
Regime-aware monitoring across positions and factors. Stress scenarios on demand. Alerts arrive with their reasoning.
Risk intelligence in detailAsset Allocation
Allocation beyond mean-variance: real-world constraints, scenario analysis, and regime context, with every proposal explained.
Explore Asset AllocationReporting
Attribution, commentary drafts, and review packs assembled from the same governed data. Drafts, not auto-published documents.
Data Engine
The foundation. It ingests positions, prices, documents, and alternative data, then reconciles them into one governed layer. Everything above runs on it.
Explore the Data Engine
Built on your data. Never used to train models for anyone else.
Your data stays yours.
It connects to what you run
Portfolio accounting, market data, and document stores. Drakkar arrives empty and learns your positions, mandates, and constraints.
It reconciles in the open
Conflicting records are reconciled instead of silently picking one. Discrepancies are surfaced, not buried.
Access follows your entitlements
People and agents see only what they are cleared to see.
Work your team will recognize.
Morning Position Brief
Overnight moves, exposure changes, and anything that crossed a threshold, assembled before the desk sits down.
Thesis File
A research dossier on one name or theme, with its counter-case and an explicit list of what the system could not verify.
Stress Run
Pick a scenario, historical or hypothetical. Drakkar re-prices the portfolio and reports which positions and assumptions drive the outcome.
Mandate Watch
Continuous screening of portfolios against their mandates. Drift is flagged with the specific holdings and clauses involved.
Every output can answer for itself.
Regulated firms cannot run AI they cannot explain. Drakkar is built for the examination, not just the demo.
Source-linked outputs
Select any number or statement and trace it to the documents and data behind it.
A named person approves
The system proposes. Nothing acts on a portfolio without a human decision on the record.
Decision log
Who approved what, when, on which evidence. Built for internal review and examiners.
Model governance
Documented behavior, versioned changes, monitored drift.
Security documentation, including hosting, encryption, and access architecture, is available in evaluation conversations. Ask us in a demo.
You do not have to start with the platform. You can start a rung earlier.
Advise, Build, Integrate, RunSee Drakkar on a real workflow.
A demo is thirty minutes. Bring a workflow your team runs today. We will show you how Drakkar would run it, sources and approval step included.