FactSet

The established platform shape, graded the same way as the other long lived platforms in this index: a genuine and substantial generative line sitting on a business that predates it and would survive its removal entirely. Strip the agent, the pitchbook builder and the embeddable conversational interface and a financial data and analytics platform serving thousands of institutions remains untouched.

Worth recording why this is a build rather than a reject, because the prior going in was the opposite. The reject shape for a data business is models around the edges of a research product. What the falsifying search found instead was several separately named products where inference does the work, sold to institutions rather than used internally, including an agent packaged specifically for embedding in a client's own stack. That is a product line, not a feature.

Two real controls sit in the output path rather than around it: generated output is explicitly labelled as generative throughout the interface, and every response carries in context source linking back to the underlying data, so a user can check any statement against its origin before acting on it. For an assistant serving analysts and bankers, verifiability at the point of use is the relevant control and it is present.

Held off A because agentic interfaces are named without any description of what an agent may do unattended, and no threshold, escalation route or review structure is published for the agentic path as distinct from the conversational one.

The strongest form of the citation control found in this sweep, and it is stated as universal rather than typical: regardless of which of the five or more data sources answered a question, every response carries in context source linking for verification and identification of data lineage, and generative output is labelled as such with linked references throughout the interface.

That lets a user validate any individual claim, which is the practical model risk control for a research assistant. Held off A because there is nothing about the models themselves: no evaluation methodology, no accuracy or grounding figures, no benchmark, no documented failure modes and no drift policy. Verifiability of each answer is offered in place of evidence about the system that produced it.

Scale is documented to a standard most of this index cannot match, because the client and user counts appear in investor communications from a listed company, and a figure in an investor release carries legal exposure that a marketing claim does not. That distinction is worth applying generally on this axis.

Held at B because none of it is evidence about the generative products specifically: no named institution is attached to an outcome from the agent or the pitchbook builder, and the productivity claims are stated in the abstract as hours reduced to minutes with no customer behind them.

The benchmark disclosure for this axis in the index, and it should be used as the reference example. It answers the pooled corpus question explicitly and in the negative: user prompts and responses are isolated and protected and would never influence the responses given to another firm. It names the architecture rather than gesturing at it, using an open agent protocol and vector stores over its own commercial data and retrieval augmented generation over client confidential data.

It enumerates every data source feeding the system, including the two that are optional and off by default. It states model hosting policy, that all models are private with no public endpoints. It states supplier posture, that third party suppliers follow the same approach and that zero data retention and abuse monitoring opt out are standard with the hosting platforms. And it states concrete lifecycle terms, twenty four month log retention and a sixty day purge. Almost every vendor in this index leaves this axis blank; this one closes it.

Client confidential data is treated as a named category with rules attached: retrieval based access rather than training, isolation between firms, access restricted to a defined subset of employees for accuracy evaluation only, twenty four month log retention and a sixty day purge after termination. Those are commitments a buyer can hold the vendor to. Held off A because no data processing agreement, subprocessor list or position on any specific financial privacy regime is published, and no residency commitment accompanies the retention terms.

Held at C deliberately and for consistency rather than because the posture is thin. The vendor publishes a substantive governance and security document for its generative layer describing access restriction, isolation, private model hosting and purge terms, which is more than most of this index offers.

But no certification and no trust portal were located in this pass, and the rule that an unverified credential earns nothing has to apply even when the vendor is large enough that one is near certain to exist. Banked check, and it is likely to move this grade: a listed financial data provider of this size will hold service organisation control reports and probably an international information security certification, so look for a dedicated trust or compliance page separate from the generative governance page swept here.

No financial services licence attaches to the graded products. The company is a listed issuer, which is securities registration as a company rather than supervision of what it sells, and by the same rule applied to the broking group in the insurance pocket that confers nothing here.

Banked check worth running before this is treated as settled: whether the group administers regulated benchmarks or indices anywhere, since benchmark administration is a supervised activity and would change this grade if the index business falls inside it.

A published generative governance policy exists and is genuinely detailed, but its content is data handling, transparency and labelling, all of which is credited on the stewardship and model risk rows and should not be counted twice here. On this axis specifically there is nothing: no fairness testing, no protected class analysis, no evaluation for skew in what the agent surfaces or omits.

Fair to note that the exposure is lower than for a credit or pricing model, since the user is a professional and the output is research support rather than a decision about an individual. The exposure is not zero though, because ranking and semantic search decide which companies and which themes a banker sees at all.

No allocation of responsibility for a wrong generated answer and no recourse route published. Source linking is offered as the mitigation, which places the burden of verification entirely on the user, and that is a defensible design for a professional tool but it is not recourse. Nothing states what happens when a sourced summary misrepresents the document it links to.

An inversion of the usual pattern in this index and an instructive one. Most vendors either name a supplier or say nothing about deployment; this one describes the deployment posture in detail, that models are private instances in its own cloud or private endpoints of commercial providers with no public endpoints and zero data retention at the host, and names its data infrastructure partner, while never naming which foundational models or which providers those private endpoints belong to.

A buyer therefore knows exactly how the models are run and not at all whose they are. Held at B on that missing half, with the how weighing more than it usually would because the deployment terms are the part that governs data exposure.

Integration is a named strength rather than a claim: deep wiring into the spreadsheet and presentation tools where the work actually lands, an agent exposed as an interface for embedding inside a client's own stack, packaged data feeds built to power a client's own workflows, and the open agent protocol used to reach client defined data repositories.

Held off A because the named integrations are productivity tools and the vendor's own protocol layer rather than named third party systems of record such as an order management or portfolio management platform, which is what an A requires elsewhere in this index.

The deployment posture for the generative layer is described with unusual precision: all models used are private, running either as private instances inside the vendor's own cloud or against private endpoints of commercial providers, with public endpoints ruled out explicitly, and zero data retention plus abuse monitoring opt out stated as standard with the platforms hosting those models.

Data lifecycle terms are concrete, with logs retained twenty four months for active relationships and confidential data purged within sixty days of termination. Held off A because geographic residency is not addressed at all, and for a client base spanning several supervisory regimes that is the missing half of the axis.

No published price for any product, the enterprise subscription norm. One partial and unusual offset worth noting: as a listed company it publishes annual subscription value and client counts, from which an average revenue per client can be derived, which is more than a private vendor's silence yields even though it tells a buyer nothing about what a specific configuration costs.

Among the widest in the index by any measure. Buy side and sell side institutions, wealth managers, private equity firms and corporates, more than 8,200 clients and over 218,000 individual users globally. The generative products address distinct workflows across that base: banker research and pitch production, portfolio analysis, general partner monitoring, and an embeddable agent for whatever a client builds itself. Client and user counts come from investor disclosures by a listed company rather than from marketing material.