Dynamic Planner

The vendor is candid that its AI is optional. The platform runs on a psychometric questionnaire and a forward looking quantitative asset risk model across seventy five asset classes. Those are behavioral science and quantitative finance, not inference. The inference features are recent, described by the company as amplifying automation that already existed, and in its own words entirely optional. Strip them and the platform runs unchanged. Still, live meeting transcription is shipped inference doing real work, and the open interface deliberately lets a firm's own agentic tooling operate inside the platform.

Control sits with the firm by design, and it is stated at two levels. Adoption itself is a choice, since the company describes its AI features as entirely optional. The ISO 42001 management framework the features are developed under commits the company to traceability, transparency and reliability in what it ships, and a published charter sets out how the technology will be applied. None of that specifies a gate at a point in the workflow.

No statement describes what a transcript or a generated insight passes through before it reaches a client record, or whether generated content is distinguishable from adviser written content. Nor does anything say what happens when an adviser's own agentic tooling drives processes inside the platform through the open interface, the least supervised path the product offers.

The underlying risk model has a published performance history against its own targets. The company states the model has performed within expected value at risk targets across a nineteen year history, including two named stress periods, and that all profiled holdings undergo formal quarterly review. Vendors rarely publish a revalidation cadence, and this one does.

Inference features are developed under an externally audited management framework, ISO 42001, that commits the company to traceability and reliability. Both attest to process and to the quantitative model rather than to the inference layer. No accuracy figure, error rate, benchmark or evaluation result is published for transcription or for any generated insight.

This is strong evidence. Independent parties with money at stake publish the relationship in their own name, in documents where being wrong would cost them. A large United Kingdom investment platform publishes an adviser facing risk profiling guide that describes this vendor as the market leader in the field. It states that its own managed portfolio ranges and funds are deliberately built to target risk profiles identified by this system.

A major life and pensions group publishes a table of this vendor's risk ratings against its own funds in its own adviser literature. Neither is a testimonial the vendor solicited. Both are firms embedding the vendor's output in their product design and telling advisers so, alongside a stated penetration figure across a national advice market.

Nothing published says whether client meeting audio, transcripts, psychometric responses or advice firm content train, tune or evaluate any model. Nor does it say whether any model provider retains content, or whether material from one firm can influence output shown to another. The ISO 42001 framework covers responsible use of the technology across the organization, which is a governance commitment rather than a statement about customer content. The accumulated data is large, since over a million individuals have completed the profiling process, and nothing published states what may be done with it.

The material at stake is personal and detailed. It covers psychometric responses about a client's attitude to risk, capacity for loss, financial circumstances and objectives, and now live meeting audio and transcripts. No retention schedule or deletion commitment is published, and nothing states how long meeting audio or transcripts persist once processed. The ISO 42001 framework the company holds addresses responsible use of AI rather than data handling terms, and those are separate questions.

There is no published information security standard, audit report or trust page, and a dedicated security page, if one exists, is not reflected here. The absence may be a disclosure gap rather than a missing control. The company holds an accredited ISO 42001 certification in an adjacent domain and has appointed a head of technical innovation and cyber security, which makes a corresponding ISO 27001 certification probable. Still, an appointment is not a credential and probability is not evidence.

The company holds no license and requires none, and no penalty applies for that. Its position relative to the perimeter is worth stating precisely. The risk profile this platform produces is a substantial input to a suitability assessment for which a regulated firm is accountable. Asset managers build products deliberately targeted at these profiles, so a large share of one national advice market rests on a scale defined by an unsupervised private company. That is a structural observation about the market rather than a criticism of the vendor. Nothing published addresses how a firm evidences the basis of a profile to an examiner.

The governance evidence is an AI specific certification rather than the vendor's own account of itself. In November 2025 the company certified to ISO 42001, the international AI management system standard, and says it was the first in its sector to do so. The certification is issued by an accredited third party, runs on a fixed term and covers ethics, transparency, accountability and risk management across the development and delivery of its features.

A charter governing application of the technology is published alongside it. A certified management system attests that governance processes exist and were audited, not that any model was tested for fairness. No metric, evaluation across client populations or result is published. The exposure that would matter is specific and unexamined in public. A psychometric instrument assigning a risk score determines what a client may be recommended, and such instruments are known to vary across age, language and financial literacy.

Responsibility for the output is unallocated in public. No warranty, accuracy commitment, service level or correction obligation is published, and no disclaimer places responsibility plainly on the advice firm either. The exposure has a particular shape because the output is a scale others build against. A client assigned the wrong risk profile receives recommendations drawn from a solution set constructed for a different profile.

The consequence surfaces years later in outcomes rather than immediately in a document. A client has no described route to see how their profile was derived or to contest it, and the advice firm carries the suitability obligation alone.

No provider, model family, version or inference host is named on the product pages, in the certification announcement or on the dedicated AI page. One published detail looks like a disclosure and is not one. The company states that a firm can either transcribe meetings live inside the platform or import transcripts from a named video conferencing service. That describes two paths and identifies who performs the work in the second, while leaving the first unattributed. A buyer learns which route avoids the vendor's own unnamed processing without learning what that processing is.

Two directions are published, and the outbound one is the more interesting. Fund and solution data flows inbound directly from asset managers, product providers and investment platforms rather than through an aggregator, which allows profiling on actual holdings, trading and tactics. Outbound, the company publishes an open programmatic interface.

It states explicitly that a firm can point its own existing automation or agentic tooling at the platform to drive processes inside it, a deliberate decision to let other systems operate the product. A video conferencing service is named as a transcript source. The integration estate is given as a count of more than twenty six rather than a list, and no back office, platform or planning system is named individually.

The platform is delivered as hosted software, and nothing about topology is published. There is no public hosting region, storage location, tenancy statement or contractual residency commitment. The consequence is narrower because customers and regulator sit in one jurisdiction, but the question is not answered. The open interface that allows external agentic tooling to operate inside the platform raises a version of it the published material does not reach.

No rate, tier structure or per adviser figure is published, and the vendor does not say whether the AI features are included in an existing license or charged separately. A competitor's comparison page asserts that the features sit behind a separate review license. That is a rival's characterization and settles nothing either way.

The two sided model makes the omission matter more, because asset managers paying to have solutions profiled and advice firms paying to use the platform are buying different things at prices neither can see.

Penetration is stated with a number rather than asserted, and it covers both sides of a two sided market. The company states that more than forty percent of United Kingdom investment advice firms use the platform, and that over one hundred and sixty asset managers do too. Discretionary managers, platforms and product providers also pay to have solutions profiled against the same model. Over a million individuals have completed the profiling process.

The buyer types therefore span advice firms, manufacturers and distributors rather than a single constituency. The limit is jurisdiction, since the platform, the regulatory frame it is built around and the entire published customer base sit in one country.