Intelligence for limited partners

The AI and agentic platform for limited partners in private market funds.

Limited partners’ intelligence — faster, deeper, more insightful.

One operating system that reads your fund documents, benchmarks what it finds, remembers your team’s judgement — and shows its working, line by line, back to the source page.

SOC 2 Type II No training on client data Segregated environments Every output traceable

Built by a team from

CPP Investments Macquarie Investec RMB Ashburton Investments INSEAD University of Oxford, Saïd Business School UNSW Sydney

The problem

Small teams. Vast volumes of unstructured fund data.

The signals an LP needs already exist inside their own documents. They are simply unreachable at the speed decisions have to be made.

Unstructured, inconsistent, non-standardised data from every GP
Narrative-heavy, data-intensive reporting and fund documentation
Lead indicators, signals and patterns buried inside hundreds of PDFs
Judgement scattered across emails, spreadsheets and presentations
Limited bandwidth for co-investment and continuation vehicle dealflow
Institutional knowledge leaves the firm when team members leave

The quarterly monitoring cycle, as it runs today

Four to seven weeks, every quarter
Several weeks
GP reporting
Quarterly reports, capital account statements and capital calls arrive piecemeal.
1–3 weeks
Performance analysis
Net IRR, TVPI, DPI and RVPI per fund; benchmarking by vintage and strategy.
1–2 weeks
Valuation & risk
Write-ups and write-downs, company fundamentals, cashflow forecasts.
1–2 weeks
GP calls & IC pack
Q&A, quarterly updates and the portfolio performance pack for committee.
Minutes
With Almagest
The same cycle, prepared end to end — and handed to your team for judgement, not data entry.
Legacy systems with bolted-on AI features and individual productivity tools do not compound.

Fund investors need a single operating system that optimises, remembers, reasons and compounds proprietary insight and institutional knowledge. In the future, every investor will have one end-to-end platform of this kind. We are building it today.

The platform

An operating system for fund investing and portfolio monitoring.

01

LP-specific automated workflows

End-to-end workflows built around how limited partners actually work — not a generic assistant pointed at your drive.

02

Patterns surfaced from your documents

Reads hundreds of PDFs and spreadsheets each quarter and surfaces what was buried inside them.

03

Sharper, data-based insight

Evidence for the decisions that matter: allocations, re-ups, co-invests, valuations and exits.

04

Institutional memory

Consolidates scattered knowledge and judgement so it stays with the firm, not with the individual.

05

A co-worker that reasons across the portfolio

Ask once and get an answer grounded in your entire funds portfolio, with its sources attached.

06

Decision audit

Every output is traceable to the original source document, page and line — automatically, on every run.

07

Integrated with your data stack

Connects to the third-party market, benchmarking and portfolio data platforms you already use.

08

Built to complement your team

It enhances professional judgement and intuition, and re-orients the team towards higher-value decisions.

Workflows

Automated end-to-end LP workflows. Results in minutes, not weeks.

Portfolio monitoring

From hundreds of PDF documents in GP data rooms to an Investment Committee-ready report.

FastIntelligentSecureTraceable
01Sweep GP data rooms and ingest every quarterly report, capital account and KPI pack
02Normalise metrics across managers, vintages and reporting conventions
03Flag deviations, lead indicators and narrative changes against prior quarters
04Draft the IC report — every figure linked back to its source page

How it works

Bounded, repeatable steps — with your team at the end of every one.

Almagest is not a chat window pointed at your data room. Each workflow is a defined sequence with a fixed budget, a recorded state and a human sign-off. The same inputs run the same process, every time.

Step 01

Plan

The orchestrator sets the scope, the sources and the budget before any work begins.

Step 02

Gather

Specialist agents extract and classify across your documents and connected data.

Step 03

Enrich

Findings are grounded against filings, benchmarks and third-party market data.

Step 04

Challenge & cite

A separate pass contests the conclusions and attaches a citation to every claim.

Step 05

Your review

No model in the loop. Your team reviews, edits and signs off — and that sign-off is recorded.

Shared plan & stateEvery agent works from one plan, so nothing is done twice and nothing is missed.
A global call budgetNo single agent can spiral. The system knows when the work is finished.
Conflict resolutionWhen two agents disagree, one place resolves it — and records why.
Model-agnosticThe harness is ours; the model underneath can be swapped as the field moves.

The output

What lands on your desk.

Live portfolio exposure roll-up

Real-time look-through across every fund, vintage, manager and underlying company.

Auditable fund performance

NAV, TVPI, DPI and IRR, each with provenance back to the document it came from.

Manager scorecards

Track record, loss ratio, marking bias, time-to-exit and peer rank in one view.

Early-warning signals

Valuation swings, narrative shifts and concentration alerts, quarter over quarter.

Scenario analysis

Multiple compression, distribution freeze and cashflow pacing, modelled on your book.

Drafted IC papers

Memos with the metrics, the drivers and the talking points — ready for your edit.

Control & trust

A system you can control, inspect and trust.

Almagest is built with a proprietary harness and context engineering. Agentic flows are tight and deterministic: they follow the process your team defines, they show their working, and they stop where your judgement begins.

Nothing is a black box. Nothing replaces your investment committee.

Deterministic by design

Constrained, repeatable flows — the same inputs produce the same process on every run.

Explainable outputs

Every number, claim and conclusion carries a citation to the source document and page.

Auditable by construction

Every model step and every human sign-off writes a record automatically — aligned to ISO 42001, not left to a developer to remember.

Human in the loop

The platform prepares the evidence. Your investment team makes the decision.

Security

Fully secure and compliant.

Bank-grade security

Encryption in transit and at rest, least-privilege access and continuous monitoring.

SOC 2 Type II

Independently audited controls, reviewed on an ongoing basis.

No training on client data

Your documents are never used to train models. Not ours, not anyone’s.

Segregated environments

Each client operates in an isolated environment with its own data boundary.

Versioned, governed config

Prompts, models and tools live in one governed store. Changes are tracked, diffable and reversible.

Auditability & traceability

Full activity logs across users, agents, documents and outputs — and every decision back to source.

What’s in the name

From human intelligence in Ancient Greece to human and machine today.

The Almagest is Ptolemy’s second-century treatise on the motions of the stars and the paths of the planets — the work that let people predict what the sky would do next, and show how they knew.

Nineteen centuries later, the throughline is unchanged: better instruments, better models, and the same human judgement deciding what to do with them. That is the company we set out to build.

1,900 years of human thought

Team

Private markets practitioners and technologists, in one small team.

Irina Luckey

Irina Luckey

Founder
  • 18+ years across private market investments and asset management.
  • Deep LP-side domain expertise: portfolio monitoring, fund diligence, secondaries and benchmarking.
  • Currently reading for the PGDip in Artificial Intelligence for Business at Saïd Business School, University of Oxford.
irina@almagest-ai.com
CPP Investments Macquarie INSEAD UNSW Sydney
Mike Wright

Mike Wright

Technical Co-founder
  • 25+ years across investment banking and asset management.
  • Builds the harness: agent orchestration, state, tooling, verification and audit.
  • PGDip in Artificial Intelligence for Business, Saïd Business School, University of Oxford (2025).
mike@almagest-ai.com
Investec RMB Ashburton Investments
We are building an advisory board. If you are a senior practitioner in private markets, or an academic working on applied AI and governance, we would like to hear from you — irina@almagest-ai.com.

Book a demo

See it run on your own fund documents.

A 30-minute walkthrough with the founders. Bring one quarterly reporting pack and we will run a live monitoring workflow against it.

Under NDA, on request
No data retained after the session
We reply within one business day

We use your details only to arrange the demo. Nothing is shared with third parties.