Fund AI supplier landscape — workflows, controls and procurement questions
Fund AI is not one market. A document-ingest product, a portfolio-intelligence
workspace and an investor-reporting assistant solve different problems and require different data,
controls and buyers. This comparison records what four suppliers claim on their
own published pages; it does not treat a feature page as evidence of independent accuracy,
production scale or suitability.
How it fits: the use cases start with the job.
This page starts with the supplier and asks where the control boundary sits.
Dates: source dates are the vendors' own published dates. Where a vendor
page gives no date, the table says so.
New releases: the latest supplier launches for fund teams are in
Latest vendor releases below the comparison.
AI assistants: finance connectors, plugins and agents a fund team can use
now in Claude, ChatGPT and Grok are in Finance tools in AI
assistants.
Ownership: questions to ask when an AI-services firm changes owner are in
Who owns your AI supplier.
At a glance
Supplier
AI job, in the supplier's words
Status, in the supplier's words
Suggested sign-off (our control view)
FactSet Cobalt AI Doc Ingest
Extracts and structures private-market documents
Beta for select North American clients, with general release scheduled for March 2026 (FactSet, 2026-02-04)
A named reviewer, before extracted fields drive a record or calculation
Allvue Portfolio Intelligence
Monitors, analyses and reports on a private credit portfolio
"Available from day one"; its Deal Analytics feature is in a beta launch programme (Allvue product page, undated)
The credit, valuation and covenant owners
Dynamo DynamoAI
Connects a client's own AI to Dynamo data through MCP
The last column is our control view, not a supplier claim. The four linked pages give no
independently produced accuracy measure.
This is not one market
Private-market document intake and extraction: turning unstructured documents
into a usable, reviewable data set.
Portfolio intelligence: assembling portfolio information for monitoring and
decision workflows.
Investor, fund and operational workflow: using an existing fund-operations
system to assist drafting, retrieval or workflow handling.
Private-equity investment workflow: using portfolio or investment context to
support analysis and follow-up.
The categories overlap in procurement meetings but not necessarily in implementation. A buyer
should first decide whether the bottleneck is source-document quality, a controlled workflow,
portfolio data, or an approved knowledge base. That is an independent conclusion from the
comparison structure, not a claim about any supplier's model quality.
Vendor-claimed: document ingest for private-market documents in Cobalt.
Vendor-claimed: private-market investment professionals using Cobalt.
Documents supplied to the Cobalt workflow; mapping, retention and entitlement controls are not described in the release.
Vendor-claimed: extracting and structuring information from documents.
Document-derived fields need deterministic validation and a named reviewer before they drive a record, calculation or decision. This is an independent control conclusion.
The 2026-02-04 release described a beta and planned later availability. It is not evidence of current general availability.
Potentially connects intake to an established private-markets platform; this is a workflow fit question, not an independent performance finding.
Current availability; supported document types and languages; correction/audit trail; retention; access controls; pricing; implementation ownership.
Vendor-claimed: Portfolio Intelligence, with a Deal Analytics feature.
Vendor-claimed: private credit portfolio teams; Deal Analytics is for qualifying Credit Front Office clients.
Portfolio, deal and workflow data connected to the platform; the public page does not specify data-quality thresholds.
Vendor-claimed: intelligence and workflow support over portfolio information.
Configured calculations, approvals and exceptions remain deterministic controls; a human owns credit, valuation, covenant and reporting decisions. Independent control conclusion.
Vendor-claimed: Portfolio Intelligence is "available from day one". Its Deal Analytics feature is available to qualifying Credit Front Office clients in a beta launch program. The page gives no publication date; current terms remain to verify.
Potentially places analysis near portfolio-operating data; confirm which objects, integrations and controls are actually included.
Exact AI functions; model/data boundaries; beta eligibility and current terms; availability by module; explainability; audit exports; implementation scope; commercial terms.
Vendor-claimed: AI capabilities within Dynamo's fund and investment-management product context.
Vendor-claimed: investment-management and fund-operations users of Dynamo.
Approved Dynamo records, documents and permissions; completeness of those inputs is not established by the public page.
Vendor-claimed: clients can connect their own AI through MCP directly to Dynamo data while retaining permissions and audit trail.
Permissioning, source-of-record updates, investor communications and regulated outputs require deterministic workflow controls and named human approval. Independent control conclusion.
The product page describes DynamoAI but does not provide a publication date or sufficient public evidence of availability by client, region or module.
Potentially keeps assistance within an existing operating environment; verify the actual scope rather than infer it from product branding.
Supported jobs; MCP scope; permission inheritance and audit-trail behaviour; retrieval sources; data isolation; logging; approvals; regional availability; migration effort; pricing.
Vendor-claimed: AI capabilities for private-equity investors in Chronograph.
Vendor-claimed: private-equity investors using Chronograph.
Portfolio, manager and investment information available to the Chronograph workflow; source completeness and permissions require buyer validation.
Vendor-claimed: AI-supported capabilities for private-equity investment work.
Investment conclusions, portfolio records and external reporting remain human-accountable; any calculation or threshold should remain deterministic and reviewable. Independent control conclusion.
Vendor-claimed: the 2024-04-16 release said ChronoAI was immediately available to all Chronograph LP clients at that date. It is not current independent proof of functionality, client adoption or accuracy.
Potentially aligns assistance with private-equity investor workflow; validate the exact job, data boundary and review path in a live evaluation.
Current capabilities; evidence provenance; evaluation method; user roles; audit trail; data residency; commercial terms; implementation responsibility.
Finance tools in AI assistants
Connectors, plugins and agents for finance work that a fund team can use now inside Claude,
ChatGPT and Grok. Each row links the maker's or the AI lab's own page and gives that page's
date.
Who can use it: the availability column says. For a data connector it is
usually the provider's existing customers.
Status: a tool in beta, or open only on certain plans, says so in the
availability column.
Rechecked weekly: a tool whose source page no longer confirms it comes off
the table.
Fund accountants and investor relations: workbooks, LP decks and board papers built and checked inside Office. Outlook is announced, not yet available.
Credit and compliance teams: credit memos, peer comparisons and scorecards; entity profiling, ownership mapping, adverse media and sanctions checks in one workflow.
In Claude.ai, Claude Desktop and Claude Enterprise
GP finance and investor relations: live fund performance, financials and compliance data in Claude and Excel; LP relationships and the fundraising pipeline.
Investor relations and analysts: earnings-call transcripts and event summaries (Aiera); governed search of internal data rooms and investment documents (Egnyte).
Analysts, portfolio managers and investor relations: research, models and client materials with built-in Daloopa, PitchBook, LSEG News, Crunchbase and Fiscal.ai data; separate workspaces for information barriers.
Eligible financial institutions, through OpenAI's sales team
In practice: before connecting an assistant to fund data, check what the connector can read,
whether the firm's data agreement with the provider allows it, and who reviews what the agent
produces.
Latest vendor releases
New AI products and features from fund-technology suppliers and fund servicers. Each item
names the supplier, the fund job it touches and whether it is in general release, in beta or
only announced. A release is the supplier's claim until users describe it in live work.
Who it affectsInvestor relationsFund administration
Belasko, a fund administrator with over $29 billion in assets under administration, announced the full launch of its Investor and Client Experience Platform (IXP and CXP) on 26 October 2026. Belasko describes it as a single AI-native environment that brings together fund information, a data repository, two-way communication, planning and support for managers and investors.
These are vendor claims. Chief Operating Officer Costa Barakos said the launch followed a period of extensive concept testing and pilot schemes, and that what comes next includes automation and future AI-enabled capabilities.
For funds: investor relations teams and fund managers that use Belasko as administrator should ask which portal features use AI today and which are still on the roadmap.
Which answers in the portal come from AI, and who checks them before an investor sees them?
Measure it: investor queries closed in the portal without an email chain, against the quarter before the switch, checked by the investor relations lead.
In practice: before moving investor communication onto the portal, get a written list from Belasko of the AI-enabled features that are live.
Who it affectsFinance and reportingFund administrationFund operations
Maybern has made its fund agent, Mai, available to every customer. Maybern's software holds each private fund's waterfall, fee and allocation terms for the GP's finance team. Where a manager uses a third-party fund administrator, Mai takes the administrator's results in any format. It then reconciles them against Maybern's own figures, with no need to reformat files.
Breaks: for each break, Mai names the likely cause, such as a missed fee offset, a different allocation basis or a timing difference.
Controls: the maths stays fixed to the fund's configured terms. Mai explains each step, and any action that would change data needs approval.
MCP server: any AI agent, on any model, can pull Maybern's figures as of a date or run the fund's terms. Each result is logged with the request behind it.
These are the vendor's claims. For a GP finance team that shadows its administrator, the quarterly tie-out moves from finding breaks to reviewing reasons. The controller still has to agree each fix with the administrator.
Who confirms the cause the agent proposes before the administrator is asked to correct a figure?
In practice: keep the agent's break notes with the quarter-end reconciliation workpapers, and record who accepted each one. The auditor then sees a human decision behind every adjusted figure.
Three questions test a release:
What data does it need? And can the firm give it that data under its
confidentiality and outsourcing rules?
Is the output a draft or a decision?
Who signs off? And does the tool keep the record of that sign-off?
In practice: ask the supplier for one client reference running the feature in live work
before it goes on a procurement shortlist.
Who owns your AI supplier
Two AI-services firms that list financial services among their sectors have recently
changed owner:
Accenture and Faculty: Accenture agreed to acquire Faculty, a UK applied-AI
firm (Faculty, 2026-01-06).
Accenture completed the acquisition on 2026-03-16. More than 400 Faculty professionals joined
Accenture, Faculty's Frontier decision-intelligence product sits alongside Accenture's own, and
the terms were not disclosed
(Accenture, 2026-03-16).
Cinven and Artefact: Cinven entered exclusive negotiations to acquire
Artefact, subject to regulatory approvals
(Cinven, 2025-07-28).
Ardian was the seller. The founders' counsel described the
investment as valued at over €1bn and Artefact's clients as including finance
(Gide, 2025-07-29).
The lenders' counsel described the financing of the acquisition and Artefact's more than 1,700
people in 31 offices across 25 countries
(Clifford Chance, 2025-12-19).
Artefact now describes itself as "Backed by Cinven"
(Artefact, 2026-09-15).
In practice: when an AI supplier changes owner, ask:
Does the contract require notice of a change of control, and does it give a right to exit?
Will the product roadmap, pricing or support terms change, and from when?
Does fund data move to new subcontractors or sub-processors, and how is it returned or
deleted on exit?
Operator evaluation checklist
Name the one workflow and the accountable owner before comparing features.
Inventory the source records, permissions, document retention and correction path.
Mark each output as draft, recommendation, record update, calculation input or external communication.
Keep rules, calculations, thresholds and approval gates deterministic where they are material.
Test a representative exception set: missing fields, conflicting versions, restricted records and a required correction.
Ask for evidence of availability, auditability, implementation ownership and commercial scope before treating an announcement as a product commitment.
What remains unknown
The four official sources establish that their publishers describe these products or capabilities.
They do not, on their own, establish comparative accuracy, total cost, current entitlement,
implementation time, customer outcomes, regulatory suitability or fit with a particular fund's
data estate. No prices, rankings, performance numbers or customer claims are inferred here.
To verify
FactSet availability — confirm from an updated official FactSet source whether the beta/planned-later status in the 2026-02-04 release has changed.
Allvue and Dynamo source dates and terms: obtain vendor-published update or release dates. For Allvue, confirm beta eligibility and current terms; for Dynamo, the MCP scope, permission inheritance and audit-trail documentation. Until then, the product wording is not time-specific.
Procurement evidence — for each evaluated workflow, obtain official, current documentation for deployment model, regional availability, audit logging, data handling, integrations, support and commercial terms.
ChronoAI availability — confirm whether ChronoAI remains available beyond the 2024-04-16 claim.