Marloo

The removal test leaves a calendar connector and a file store. Models transcribe client meetings captured on any conferencing platform or in person through a mobile application, generate tailored summaries and advice documentation including statements of advice from uploaded client files, fill forms, draft client emails, and answer free form questions across every stored transcript, document and note for a client.

That last capability is the one that could not exist without models at all, since it turns an adviser's accumulated archive into something that can be interrogated for a detail from a conversation years earlier or for how a client's goals have shifted over time.

The output includes regulated documents and no boundary on their production is published. The platform generates advice documentation including statements of advice from uploaded client files, drafts client emails and fills forms, all of which are artifacts a regulator or a client may later rely on, and nothing in public material states that an adviser must review and approve before anything leaves, describes a confidence indication on a generated document, or defines what the machine may not conclude.

The contrast within this pocket is direct and unflattering: both United Kingdom peers publish explicit non delegation statements, one architectural and one argued from the conduct rulebook, and this vendor publishes neither.

No accuracy figure, error analysis, validation evidence or confidence exposure was located for any component. The archive query capability is the sharpest gap, because answering a question by drawing on transcripts and documents accumulated over years is exactly the pattern where a plausible but wrong answer is hardest to catch, and nothing describes whether responses cite the underlying record.

Two peers show what the alternative looks like, with one running dozens of structured checks on every generated document and returning pass or fail per category, and another linking every assertion back to its source.

Founded in mid 2024, so the evidence surface is thin by age rather than by choice. Several hundred users were reported in one market in late 2025 against a team of seven, with growth stated at 45 percent month on month, and the product is live across five countries.

Funding is the strongest signal at 12.7 million dollars raised inside a year across a pre seed and a seed both led by the same institutional investor, with an angel register drawn from operators at well known fintech and platform businesses. Against that, no client firm is named anywhere, no user or meeting volume is published, and outcome evidence is anecdotal, resting on testimonial claims such as a document that would have taken thirty to sixty minutes arriving in ten seconds. The product was built from conversations with more than 900 advisers, which is research rather than adoption.

The zero data retention arrangement with the model providers is a genuine boundary at the layer where most vendors here leave the question entirely open, and stating the contractual mechanism rather than asserting a policy is what earns the grade. What is not addressed is the boundary inside the platform itself.

Nothing states whether transcripts, documents or interaction patterns from one advice firm inform anything served to another, and that matters more as the archive grows, because the product's stated value increases the longer it is used and the more client history it holds.

One commitment does real work and it sits at the most exposed point in the architecture. The company states that its third party model providers operate under zero data retention enterprise agreements, so client information passed to them during processing is not stored by those providers, which addresses the question a buyer asks first about any product sending client meeting content to an external model. Held at B because the vendor's own side is undescribed.

The whole proposition is building a complete picture of every client relationship over time, which means a persistent and growing archive of recorded advice conversations, and no retention schedule, deletion commitment, subprocessor list or statement of what the client is told about the recording was located.

No attestation, certification, trust centre or enumerated framework was located. The zero data retention arrangement with model providers is a contractual term rather than an independently assessed control, and it covers one link in the chain rather than the vendor's own environment.

For a young company this is unsurprising, and it is also the document a multi adviser firm's due diligence will request before recorded client conversations start accumulating, so it is a commercial obstacle rather than a compliance detail. Both nearest peers publish at least one named framework.

The product is described as built for the way compliance actually works in financial advice, understanding regulatory language and identifying required disclosures, and it names one regulated artifact by its proper title. Beyond that no regulator, statute, rule or guidance instrument appears anywhere.

The gap is larger than usual because the company operates across five jurisdictions with materially different advice regimes, each with its own documentation and disclosure obligations, and nothing published addresses how one product satisfies all of them. The comparison inside this pocket is stark, with AdvisoryAI citing a conduct rule at provision level.

The speech recognition variance finding applies for the tenth time in this index and lands on the client rather than the employee. Extraction works from natural conversation in advice meetings, and accuracy varies with accent, dialect, age, hearing difficulty and speech impairment, so a client the system reads less well may have their circumstances, objectives or risk tolerance captured less accurately, and that misreading flows into a document recommending what they should do with their money.

The exposure is sharper here than for the nearest peer because there is no published verification layer checking generated documents for gaps and no stated requirement that a human approve before the output is used. No per accent or per population accuracy was located.

No guarantee, indemnity or falsifiable commitment was located, and no correction or review path is described. Accountability still lands somewhere useful by default, because the adviser using the output is a regulated person answerable for the advice regardless of how the document was produced, and a client retains the ordinary complaint and ombudsman route in every market served.

But that is the regime working around the product rather than the product supporting it, and the absence of a stated approval step means nothing in the design guarantees a human read the document before the client did.

The most complete model supply chain disclosure in this index. The company names not one but two frontier model providers whose services it runs on, and it names the commercial terms governing them, stating that both operate under zero data retention enterprise agreements so client information is not stored by those third parties. Almost every vendor here leaves the provider unnamed; the handful that name one leave the contractual terms open.

Naming the providers tells a buyer whose infrastructure their client conversations traverse, and naming the retention terms tells them what happens to the data once it arrives, which together answer the fourth party question an examiner actually asks. Sixth A on this axis and the first where the disclosure covers both identity and terms.

The capture side is complete, covering calendar integration with both major providers, every significant conferencing platform, a mobile application for in person meetings and bulk upload of existing templates, client files and documents. The delivery side is where the gap sits and it is the one that matters most for this buyer.

No adviser practice management or back office system is named as an integration target, so nothing describes how a generated note, document or client update reaches the system of record the firm is obliged to maintain, leaving the adviser to move the output manually. The nearest peer names four back office platforms with automatic write back and takes an A on this axis for exactly that reason.

No hosting provider, region selection or residency commitment was located. That absence is more consequential than for a single market vendor, because the product is live across five countries spanning three continents and the recordings involved are client advice conversations, so a firm in any one of those markets cannot establish whether its client data rests in its own jurisdiction. The zero data retention arrangement addresses what the model providers keep, not where the vendor's own archive lives.

A published rate, which almost nothing in this index offers. The company states a price of 99 dollars per adviser per month, with discounts available for larger networks and support staff access included at no charge, so a firm can calculate its cost before speaking to anyone and can see that the meter runs on advisers rather than on seats, meetings or documents.

Including support staff free is a deliberate structural choice that removes the incentive to under license the people doing most of the administrative work the product exists to absorb. Second A on this axis after Sumsub, and the first from a vendor selling into advice rather than identity.

Broader than most specialists in this pocket on two dimensions. By advice type it serves wealth management, financial planning, insurance advice and mortgage advice rather than investment advice alone, and by role it covers advisers, paraplanners and support staff, with a practice tier for multi adviser firms alongside single adviser practices. Geographically it is live across Australia, New Zealand, the United Kingdom, Ireland and South Africa, which is five distinct advice regimes.

What holds it at B is depth against that breadth, since the evidenced user base is concentrated in one market and nothing demonstrates how the product handles the different disclosure and documentation requirements each of those jurisdictions imposes.