An installable package in your own cloud — not a service you send data to. Client data stays in your existing ledgers, document stores and mail where it already lives. No second data estate.
Agentic operations for private markets fund administration.
The agentic operating layer for closed-end fund administration — your agents working alongside your fund accountants and client-service staff, above the ledgers, document stores, portals and workflow you already use, inside your cloud estate and under your risk framework and controls environment. Models are selected and managed against risk, with cost recorded per task.
Agents do the heavy lifting·Your people deliver the judgement·Everything is evidenced
Under construction.
Two things make this hard. Neither is the models.
Many AI options.
Agent platforms, frontier agentic tools and AI inside each system. Each is capable. Each is either too generic, hosted elsewhere, or confined to one system — and none of them starts from your risk framework. A fund administrator needs its own agents: in its own estate, under its own framework, calling the models it chooses, with a person releasing what leaves.
No market infrastructure.
Public markets standardised themselves: common identifiers, shared settlement, central prices, agreed formats. Private markets never did. Each manager keeps its own conventions and the administrator absorbs the difference — by reading, every item, before any process can begin.
Your agents, alongside your staff, above your systems.
Everyone can give a fund administrator agents. Countersign gives it the control environment its agents work inside — the one its regulator, its auditor and its clients already expect of its people. Agents are consulted, not in charge: they read, retrieve, draft and check. A person releases anything that leaves the firm.
Every bolt-on makes controls harder to prove. Countersign replaces the patchwork.
Each new tool adds another place your controls have to be evidenced — and another gap between what your policies say and what your auditor can test. Countersign is one system of control running across the systems you use: work is risk-rated and prepared by your agents, checked and released under a single permission system, and evidence is produced as part of the work itself — not assembled after the fact.
Anything leaving the firm is released by a named person, per item. Low-risk internal steps can follow a policy your firm sets. Maker-checker is enforced in the database. History is append-only.
You already run this for payments — A and B signatories, thresholds, dual authorisation. Countersign applies that grammar to every task, and treats agents as a signatory class with a ceiling.
Every AI platform says a person reviews the output. Countercheck, a feature within Countersign, measures whether that review is working. How Countercheck works →
An audit log says who logged in. An evidence record says what happened to the item.
Every system keeps an audit log. Countersign keeps one too. The record is a different thing: for each item of work, what arrived, how it was rated, what the agent drafted, which documents it used, who checked, who released, to whom, and which model was called at what cost. Append-only, assembled as the work happens, readable by your auditor without a reconstruction.
The agentic layer will become core infrastructure.
A general-purpose platform sells the means to build agents. The administrator then has to build everything that makes agents usable on a regulated service: a risk taxonomy and a rating for every item; permissions by risk, amount, seniority and direction; a release gate for anything client-facing; an evidence record per item; an exam for each model and a re-examination schedule; a measure of whether the human review still holds; model cost by task — all inside its own tenancy, above its ledgers, document stores, portals and mail. That is potentially a multi-year programme, which is where enterprise systems and agent programmes typically stall. Countersign ships it as the product. You load your own framework into it.
This is how we see it, from vendors' published material, October 2026. Categories, not named products; "not stated" means we could not find it published, not that it does not exist.
AI is everywhere in the firm. The value is in the operation.
Most AI use today is personal and point-by-point: a chat window, a meeting summary, a deck, a workbook someone built for themselves. It makes a working day easier. It rarely changes the cost of running the service, and each private routine is one more place where accountability loses line of sight. Countersign is not a replacement for those tools. It is where the service itself runs — the client inbox, the preparation, the checking, the release — with every item rated, recorded and costed, so the value shows in the operation’s own figures, not in anecdotes.
Set up with your own AI, not a consultant programme.
Countersign ships its set-up as files your own AI can read and work through with your Technology, Risk and Client Service teams — ingesting your procedures, your SLAs and each client’s prior quarters of service and cycles — to deliver a configuration and risk ratings for your people to confirm. Deployed, in your control, into your own cloud. With a train-the-trainers approach, adopting Countersign is not a consultant-led engagement.
Agentic operations for fund administration.
Forty minutes: where Countersign sits in a fund administrator’s estate, a short walk through the application, how it is examined before it is adopted, and questions.
For Fund Administrators.
Countersign is built to deliver the value of agentic service efficiency — whilst maintaining service quality within your risk and control environment.
GPs are increasingly adopting AI for their businesses, including pointing it at deliverables from their fund administrator. AI is writing the requests arriving in your service inbox — while regulators keep raising the bar on controls and evidence. Countersign is the layer you own: agents on your side, under your control, enabling you to define agent-to-agent engagement, evidenced end to end.