PlannerPal

Every output the product exists to produce is generated. Transcription of recorded meetings, seven document types written in the individual adviser's style, proposed record updates derived from conversation, and the pre meeting briefing synthesised from valuations, notes and correspondence are all model driven. Remove the models and what remains is a connector to a back office system and a recording function, which is not the product anyone is buying.

The company also states the models were trained specifically for the United Kingdom advice market so that product names and industry terminology are handled correctly, which is a model level investment rather than a configuration one.

Two real mechanisms sit in the path and neither is fully described. Generated advice letters, suitability reports and review letters carry citations, stated as evidencing data completeness and transparency, which surfaces where a claim in the document came from at the point the adviser reviews it. Separately the record updater suggests changes to the back office rather than writing them automatically, so a person confirms each update to the client file. Both are the right shape.

What keeps this at the middle grade is that no confidence measure accompanies the citations, nothing states what happens when source data is missing or contradictory, no escalation or threshold is described, and the extent of citation coverage across the seven document types is not set out.

Accuracy is asserted and never measured. Marketing describes context aware accuracy and reduced errors through cross checking against live record data, but no error rate, evaluation methodology, test set or independent assessment is published for transcription, extraction or document generation. The citation feature evidences where information came from, which is provenance rather than a measure of how often the generated text is wrong, and the two are easily conflated.

Nothing describes how model changes are detected or managed over time. For a product whose output is the file evidence that advice was suitable, the absence of any published measure of error in either direction is the central gap.

Traction is real and the figures are more specific than most in this pocket. Usage is reported across more than three hundred and fifty firms, with average administration and post meeting time down fifteen to twenty percent, and individual advisers reporting a seventy five percent reduction in client administration and thirty to sixty minutes saved per meeting.

Some of this is carried in interviews published by a major back office vendor rather than only on the company's own site, and independent trade press covered the pre meeting product launch, so the vendor is visible outside its own marketing. What holds it below the top grade is attribution: not one customer firm is named anywhere located, the adviser quotations are anonymous, and the percentage figures ultimately originate with the company even where a third party published them.

Neither half is addressed, and the stewardship gap is conspicuous because two direct competitors in the same pocket do address it. Nothing located states whether recorded client meetings, transcripts or extracted financial circumstances are used to train or improve models, whether third party providers are permitted to train on that material, whether a firm can decline such use, or how long recordings and transcripts are retained.

Both Automwrite and FE fundinfo publish an explicit position on exactly this question. On the safety side there is no description of controls against fabricated content in documents that become regulatory evidence, no evaluation of output quality, and no account of what happens when source material is thin. The citation feature is the nearest thing to a control and it is aimed at completeness rather than correctness.

Nothing substantive on privacy was located for a product that records client conversations and holds synchronised client records. No privacy management certification is held, no retention or deletion period is published for recordings, transcripts or generated documents, and no position is stated on the lawful basis for processing recorded conversations or on consent from the client being recorded, which is the specific question a recording product raises.

The pre meeting pack additionally assembles a consolidated view of a client's holdings, tax position and communications history, and nothing describes how that assembled profile is stored or how long it persists.

No information security management certification, service organisation control attestation, trust centre, security page or scope statement was located. This matters in proportion to what the platform holds, which is recordings and transcripts of client conversations plus synchronised client records drawn from a firm's back office system.

Advice firms in this market are increasingly asked to evidence supplier due diligence by their networks and compliance providers, and nothing published here would answer such a questionnaire. The direct integration with a major back office vendor implies that vendor conducted some assessment, but no detail of any such review is public.

PlannerPal holds no financial licence, which is correct for a technology supplier, and the regulatory grounding comes from what the product produces. Its outputs include suitability reports and advice letters, which are the mandated file level evidence that a recommendation was appropriate, and the pre meeting pack surfaces compliance status as a standard element. The models are built around the United Kingdom advice rulebook rather than adapted from another market.

It stops short of the top grade because no regulator has assessed, tested or supervised the product, no authorisation or programme participation exists, and the compliance framing is the vendor's own characterisation rather than anything externally validated.

Nothing published addresses whether the system performs evenly across the clients it processes. The exposure is specific rather than theoretical: the product converts recorded speech into the notes, letters and suitability rationales that document a client's circumstances, and transcription accuracy is a documented source of variation across accents, speech patterns and age. A transcription failure here does not merely produce a poor summary, it produces an inaccurate regulated record. No demographic analysis, no evaluation of transcription performance across speaker populations, no bias testing methodology and no governance framework covering model fairness was located.

No accuracy guarantee, remediation commitment or allocation of responsibility between the vendor and the advice firm was located. The citation feature and the suggested rather than automatic record updates place the burden of verification on the adviser, which is appropriate for a regulated professional but creates no obligation on the supplier when a generated document misstates a client's circumstances.

The retail client is further removed again: their conversation becomes the transcript that becomes the suitability rationale, they are not told a model drafted it, and no route exists for them to establish how their circumstances were captured if the advice is later disputed.

No base model, provider, version or component is named for transcription, generation or the record update suggestions. The company states the models are trained for the United Kingdom advice market, which describes tuning rather than origin and leaves open whether the underlying models are its own or licensed, and no subprocessor list is published. There is no identifier a firm could record against a generated suitability report to establish later which system produced it.

The comparison inside this pocket is direct and unflattering: a competitor of similar size publishes a dated table naming every model and transcription provider, its processing activity and its country of processing.

The integration is deeper than a data export and is externally corroborated. PlannerPal appears in the partner directory of the largest United Kingdom adviser back office system, with a direct connection that syncs client lists, auto saves generated documents, enriches meeting notes with live record data and cross checks outputs against that data for context.

Meeting capture connects directly to the two main conferencing platforms without deploying a meeting bot into the call, which is a meaningful architectural difference for firms whose compliance teams object to third party bots joining client conversations. A competitor's published comparison credits it with four back office integrations.

It sits below the top grade because only one back office connection is independently evidenced, no partner directory or public documentation of the others was located, and no service status page or changelog exists.

Delivery is hosted software integrated into the firm's existing back office and conferencing tools. Beyond that the record is silent. No hosting region or country is named, no in country residency option is described, no data transfer mechanism is set out and no subprocessor list exists. The gap is material for a product capturing recordings of United Kingdom retail client conversations, where firms and their compliance oversight providers routinely ask where such recordings are held. A direct competitor in this same pocket publishes processing country for each of its model and transcription providers, which shows the disclosure is achievable at this company size.

No rates, tiers, seat pricing, volume bands or trial terms were located. Distribution through a major back office partner store might imply standardised commercial terms, but no figure or structure is published in that listing either. Every route resolves to a demo or contact request. Two direct competitors in this pocket publish either a per seat monthly rate or a public pricing page, so the absence here is a choice rather than a category norm.

Coverage is genuinely narrow rather than merely undocumented. The product serves one jurisdiction and one buyer type, United Kingdom financial advice firms, with roles inside those firms spanning advisers, paraplanners and support staff. Penetration within that segment is real at more than three hundred and fifty firms, but penetration is not breadth.

There is no deployment outside the United Kingdom, no adjacent institution type such as banks, discretionary managers or insurers, and the model training and document formats are built around a single national rulebook that would require rebuilding for any other market. The same grade and reasoning apply to the direct competitor Automwrite.