Performativ

Embedded agents automate reconciliation, data validation and operational processes across the investment lifecycle, a builder lets firms compose custom agents for regulated workflows, and the company describes itself as artificial intelligence native rather than artificial intelligence enabled, with the permission architecture designed from the outset to accommodate agents alongside people.

Against that, the removal test leaves a substantial product: a cloud platform consolidating portfolio management, trading, performance attribution, risk analytics, compliance, reporting and multi custodian aggregation is a real operating system for a wealth manager, and firms would buy it to escape legacy tooling whether or not agents ran on top. This sits between additiv, where the platform long predates the models, and a pure agent business.

The most architecturally coherent answer to agent governance recorded in this index, because it does not restrict agents specially, it subjects them to the controls that already govern staff. Data, workflows and permissions are structured so that humans and agents operate within the same controlled, auditable and compliant environment, and rules, permissions and audit trails are enforced directly within daily workflows rather than applied around them.

An agent therefore cannot reach data or take an action a person in that role could not, and every action it takes lands in the same record. That is a fifth distinct architecture in this index after Recordsure prohibiting the decision, AdvisoryAI justifying the prohibition from the rulebook, Vouched bounding the grant of authority and Saturn mandating review.

Auditability is a structural property rather than a promise, with audit trails enforced within daily workflows and the environment described as controlled, auditable and compliant for humans and agents alike, so any automated action can be reconstructed after the fact. One agent function is itself a control, since data validation run by model driven workflows checks the accuracy of the information everything else depends on.

What is missing is measurement: no accuracy rate, exception rate or validation result is published for the agents, and for a platform reconciling holdings across multiple custodians the rate at which it resolves breaks correctly would be the most informative figure available.

European firms managing more than 80 billion euros in assets run on the platform. The Series A of around 14 million dollars in April 2026 was led by a major European exchange group, whose executive board member is quoted describing it as strengthening their own investment management solutions proposition for the buy side, and joined by the investment arm of a large Dutch bank alongside existing backers including a specialist fintech fund and Denmark's state export and investment fund.

That is two significant financial institutions putting capital into a supplier serving their own market. Two major custodians are named as integrated data sources, one client is identified with manual administration reduced by up to 75 percent, and total funding stands near 20 million dollars.

No data boundary statement covering model training was located. The permission architecture governs who and what can access which data inside a firm, which is access control rather than a boundary between customers, and the platform serves many wealth managers competing for the same clients in the same European markets.

Nothing states whether workflow patterns, reconciliation exceptions or agent behaviour learned in one firm's environment inform the agents serving another, or whether custom agents built by one customer remain that customer's.

A residency commitment is published and framed as a differentiator rather than a footnote, with full European data residency offered against competing platforms said to lack it, and the company argues that absence exposes firms to cyber and compliance risk. For a platform holding consolidated client holdings, transactions and private asset records across multiple custodians, knowing the data never leaves the region is the substantive privacy assurance a European wealth manager needs. Held at B because no retention schedule, subprocessor list or data processing terms were located, and consolidated household wealth data is among the more sensitive payloads in this index.

No attestation, certification, trust centre or enumerated framework was located. The gap is notable because the company positions itself around the European operational resilience regime, which imposes specific requirements on how financial entities assess and monitor technology suppliers, and a platform holding consolidated client wealth data for firms subject to it will face exactly that assessment. Backing from an exchange group and a large bank means the diligence has been done privately and none of it is published.

Four European instruments are named rather than gestured at, covering the operational resilience regime, the sustainability disclosure rules, the revised alternative fund managers directive and, most notably, the European artificial intelligence regulation, with custom agents stated to operate in regulated environments in compliance with it.

Naming that regulation in connection with a specific agent capability is rare in this index and directly relevant, since it is the instrument that will govern how these systems may be deployed. What holds this below the top grade is provenance and depth: the naming reaches the reader substantially through third party reporting rather than the company's own detailed material, and no article or obligation is cited at provision level.

No consumer credit or insurance decision applies, so the axis adapts, and the governance question is what agents are permitted to do inside a regulated firm. The company engages that directly by claiming alignment with the European artificial intelligence regulation for agents operating in regulated environments, which implies risk classification and human oversight obligations have at least been considered, and the permission architecture provides the mechanism.

What is absent is anything about outcomes: no analysis of how automated reconciliation, validation or reporting performs across client types or portfolio structures, and nothing describing how the platform handles clients whose holdings fall outside common patterns, where automated processing is likeliest to fail quietly.

No commercial guarantee or indemnity was located, and accountability is nonetheless reconstructable by design. Because permissions and audit trails apply identically to agents and to people, a firm can establish exactly which actor took an action, under whose authority and when, which converts a dispute about an automated reconciliation or report into a documented chain rather than an argument. The wealth manager remains the regulated party and retains responsibility to its clients. What is missing is anything covering the vendor's own failure: no correction process, notification path or remediation is described where the platform itself produces a wrong result.

The data ingestion layer is partly named, with the company's own pipeline technology identified as the route through which custodian data arrives and two major custodians named as sources, which tells a buyer where holdings information originates.

Everything on the model side is undisclosed: no provider is named for the embedded agents or the agent builder, no subprocessor list appears, and for a platform claiming alignment with the European artificial intelligence regulation, identifying the underlying model providers is part of what that regime is designed to surface.

Multi custodian aggregation is the hardest problem in wealth technology and it is the platform's foundation, with two major European custodians named as integrated sources through the company's own data pipelines, alongside stated integration of custodians, data providers and third party technologies into one ecosystem.

Consolidation runs front to back across portfolio management, trading, reporting, compliance and analytics, which is what allows it to replace fragmented tooling rather than sit beside it. The agent builder extends the surface further by letting a firm compose its own workflows on top of the same connected data.

A residency position is published and used competitively, with full European data residency offered explicitly against incumbents said to lack it, which is a concrete commitment about where client data rests rather than a general assurance. For European wealth managers facing operational resilience obligations that require knowing exactly where processing occurs, that is the answer the compliance function needs. Held at B because no hosting provider or region selection detail is published, and no private or dedicated deployment option is described for institutions that require one.

No pricing, packaging or basis of charge was located. The platform spans front, middle and back office functions that are conventionally licensed separately, and the agent builder adds a further dimension, so the commercial structure is visibly layered and entirely undescribed. Nothing indicates whether charge scales with assets under administration, users, custodian connections or agent usage.

Four institution types are addressed, spanning independent wealth managers, asset managers, family offices and private banks, with a stated push into larger institutions requiring more complex data structures, higher volumes and more demanding reporting. Functional coverage is genuinely front to back rather than one layer, which is the whole proposition.

The limit is geographic: this is a European platform serving European firms, and its principal competitive argument, regional data residency, is one that only matters inside that market and does not travel.