Get your firm's data ready for AI.
Groundwork joins every source in your firm into one model of the firm. The model is governed and keeps its full history. Every person and every agent reads from it.One governed model of your firm, built from every source. Every person and every agent reads from it.
Your firm has the data. But it sits in many systems, the terms mean different things to different teams, and models guess at the numbers. Groundwork fixes each of these problems.
Facts sit in the administrator, the GL, the CRM, models and thousands of documents. These sources do not share keys.
Every source attaches to the same funds, investors, commitments and deals.
Net IRR means one thing to IR and another thing to finance. An agent picks one, and the answers drift.
Each term has an owner, an approval state, a scope and a link to its calculation.
Models estimate waterfalls, fees and allocations from documents. A plausible number can still be wrong.
Maybern calculates the fund economics. Agents cite the results. They do not calculate them again.
Groundwork is one governed model of your firm, with its full history. Every person and every agent reads from it.
Read from the bottom up. Your sources attach to one model of the firm. People and agents read from the top. Governance runs through every layer. Select a layer to see what it does.Read from the bottom up. Sources attach to one model of the firm. People and agents read from the top. Select a layer to open it.
NAV and close checks, LP and IR requests, side letter and MFN compliance, IC memo review.
Agents that your firm builds, on the same approved model.
Third-party assistants and copilots, connected through the MCP.
Analysts and finance teams ask about approved metrics.
Investor reporting, fed by a data share.
approved definitions onlyTools for agents to find terms, query as of a date and get citations.
Approved views and endpoints for analysts and applications.
Zero-copy shares to other platforms and portals.
Plain-language and spreadsheet access for business users.
one model, many readersKeyed to the Maybern calculation engine
The model of the firm. Every source attaches here.
What each term means, who owns it and which metric calculates it.
Each link has two dates. So agents can answer with what the firm knew at quarter end.
How your firm ranks its sources and ties them out.
resolved · mapped · sent for approvalStructured systems arrive by connector or secure share. The administrator, GL and CRM become live tables.
LPAs, side letters, notices and memos become structured clauses, linked to terms.
The same investor, vehicle or deal in different systems resolves to one key.
Drafted from usage and documents. A person approves each conflict.
raw facts in five formatsYour fund record. Waterfalls, fees and allocations, calculated by Maybern. Agents cite them. They do not calculate them again.
Ledgers, capital activity, NAV packages.
Terms, MFN elections, fee and carry provisions.
Notices, quarterly reports, IC memos.
Underwriting models, portfolio company data.
LP relationships, fundraising pipeline.
Correspondence, meeting and call recordings.
W1, W2 and W3 show the wave in which each source attaches. See the waves.
An IR agent asks a question for a user. Follow the question through Groundwork, step by step.
What was Fund III net IRR at Q3 close, as the firm knew it on the reporting date?
Same question, same answer, from an agent, from an analyst in Excel or from an LP portal.
Start with your fund economics. Then add documents. Then add deals, CRM, email and calls.
Maybern fund economics, the administrator and the GL. Your team approves the core terms: NAV, unfunded and net IRR.
LPAs, side letters, notices and quarterly reports. Each clause links to its terms and calculations.
Deal models, portfolio data, CRM, email and meeting transcripts. Each one resolves to entities that are already in the model.
Waves compound. Each new source attaches to entities that are already in the model. So each new source costs less to add than the first.
Meet the engineers who attach each wave →Meet the engineers →Durations are typical for a multi-fund manager with $25B or more in assets.
Groundwork runs in your firm's own data environment, such as Snowflake. Every query runs as the user, on your compute. Nothing leaves.
Each query uses the access of the person who asks. An agent sees only what its user can see.
Groundwork uses the compute in your own environment. Your data does not go to a new place.
Every answer goes back to its clause, its row or its calculation. Your auditor can follow it.
Definitions export to open formats (RDF). You keep your model.
Our engineers attach your fund economics, your administrator and your GL in the first weeks. Then your first agents go to work on numbers that your team trusts.