DM Faster

An agent sits at the front of the product and a complete mechanical engine sits behind it. The workspace opens with a conversational assistant that finds companies, drafts opening messages and surfaces new replies, which is a real model layer and a genuine part of the experience.

Remove it and what remains is the whole of what the platform sells: a searchable database of five million company profiles with filters for location, industry, size and technology, automated sending across four channels, and a unified reply inbox. Packaging agrees, since neither published plan lists an AI capability as a line item and the meter is denominated in credits rather than in model work.

One boundary is stated clearly and it sits after the risky step rather than before it. The vendor writes that the platform can prepare and send opening messages at a human like rhythm the operator controls, and that the customer's own team handles the conversations that follow. So a person owns every reply and no person needs to see the first message before it reaches a stranger. Pace control is a throttle rather than oversight. No approval gate, review step, withholding guardrail or audit trail of what was drafted against what was sent is described anywhere.

The agent is described by what it does and never by what it is, with no provider, model family, version or inference location named. One ambiguity is worth recording rather than resolving: two model vendor marks appear in an icon strip captioned as available and planned integrations, which tells a reader neither whether those services power the agent nor whether they are shipped at all. No accuracy or error rate is published for the company matching or for the drafted openers, on a product where a wrong opener is sent to a stranger before anyone reads it.

The proof surface is empty. No customer is named, no logo appears, no case study exists, no outcome is quantified and no independent review was located. The product illustrations use invented company names and a scripted example conversation rather than any real account. The only person named anywhere on the site is the founder, who is credited as the editor of the guide library. Recorded as an early stage fact rather than as concealment, and it leaves a buyer with nothing to reason from beyond the vendor's own description of the workflow.

The regulated surface is heavy and no regulation is named on any page read. Unsolicited opening messages are sent automatically across three social platforms and email, from a company established in the European Union, to recipients who include consumers gathered from a competitor's follower list. That engages consent rules for unsolicited electronic marketing, the lawful basis question for profiling, and each platform's own messaging policy, and none of the three is mentioned. Terms of use and a privacy notice are published at stable URLs and neither was read this pass. No consent position, suppression mechanism or opt out route was located in the product description.

One partial provenance statement exists and the harder half of the question goes unanswered. The vendor states that company profiles are compiled from public company sources and enriched with social media signals, which is more than several data vendors here disclose.

What is absent covers the people rather than the companies: no notification, lookup or removal route for the individuals profiled, no data protection officer, no sub processor list and no residency statement, on a European operator whose consumer tier assembles lists of named individuals from a competitor's following. Terms and a privacy notice are published and were not read this pass.

The second appearance in this index of the same mechanic, and here it is the headline feature of the cheaper plan rather than a capability buried in a feature list. The plan sells the ability to collect a competitor's followers into a list and message them automatically, so named individuals are taken from a platform that offers no follower export, on the basis of who they chose to follow, with no source disclosure, no lawful basis, no notification and no removal route.

The business tier's company data carries a partial provenance sentence and the follower harvesting carries none. The people gathered this way are consumers, who hold stronger protection against unsolicited approach than the business contacts this axis usually concerns.

Four elements compound and all of them are the vendor's own marketing. Opening messages are sent automatically on Instagram, Facebook and LinkedIn, none of which sanctions automated outreach of this kind. Follower lists are assembled from a competitor's page, which no platform offers as an export. The pacing control is described as a human like rhythm, which is evasion framing rather than a conformance position.

And the platforms themselves are never acknowledged: no terms reference, no rate limit statement, no automation policy and no allocation of the risk when an account is restricted. The account that carries that risk belongs to the operator, on personal social profiles they keep.

The training question goes unanswered while the platform accumulates an unusual corpus: harvested follower lists, company profiles enriched with social signals, drafted opening messages and the reply threads that follow across four channels. Whether any of that trains or improves models serving other accounts appears nowhere, and no retention period, tenancy boundary or deletion behaviour is described. The published terms and privacy notice are the likely home for any of it and went unread this pass.

Undetectability is sold as a control, which is the test for this band. The vendor states that opening messages are sent at a human like rhythm, so pacing exists to make automated contact resemble a person rather than to protect anyone receiving it. The message itself is drafted by a model and arrives from the operator's own social account, so the recipient sees a personal note from a real profile with no available signal that neither the timing nor the words were chosen by a human.

On the consumer tier the recipient did nothing but follow a competitor's account. Article 50 goes unmentioned by a European vendor. Recorded in fairness: no invented persona, no synthetic voice and no impersonation of a named individual appears anywhere.

An integration strip names eight services covering mail, messaging, a professional network, a CRM, spreadsheets, a document platform and two model vendors, and the caption underneath it reads available and planned, so a reader cannot tell which of them exist today. That ambiguity is the finding: an integration list that does not separate shipped from roadmap tells a buyer nothing they can rely on.

Beyond it, CRM integration appears as a bullet on the unpriced tier, a documentation site is published including an integration guide for one named CRM and a guide for the agents, and no public API, webhook surface or agent facing endpoint was located.

The residency question goes unanswered across every page read. Hosting provider, processing region, storage location and sub processor list are all absent. The gap is more pointed than usual because the vendor's whole positioning is national, built in Finland and marketed almost entirely in Finnish to Finnish buyers, and European buyers who choose a domestic vendor commonly do so for exactly this reason. A vendor with that position and no residency statement leaves its strongest available argument unmade.

A security page, trust centre, certification, audit, penetration test, control set and status page were all absent from the pages read, and the footer legal section carries exactly two items. Held at the bottom of the band rather than lower because no certification is claimed and left unsubstantiated.

The access the product needs is the concern: sending automated messages from an operator's accounts on three social platforms requires holding whatever credential or session makes that possible, and nothing published describes what is held, how it is protected or what happens to it on cancellation.

One tier carries a real published price with its contents enumerated: 99 euros a month, a three day free trial, a self serve checkout path, and a stated allowance of ten thousand credits a month alongside audience collection, competitor follower gathering, automatic opening messages and one workspace. That is a complete purchase path for the buyer it addresses. Off the top of the band on two counts, and the first matters most for a business buyer.

The business tier, which is the one carrying the company database, the four channels and the CRM integration, is routed to a sales conversation with no figure, band or floor of any kind, so the buyer this product is sold to has no price at all. And the credit is never defined, so a buyer cannot tell whether ten thousand is a month of work or a week.

Neither direction is documented. No export function, download route or portability statement appears on the pages read, and no post termination position either: retention period, deletion timeline and deletion artefact are all absent. The material a departing customer would want covers built audience lists, message history across four channels and the reply threads that came back, none of which is described as leaving the platform. CRM integration on the unpriced tier is the only route out implied anywhere, and it is a bullet rather than a described mechanism.

Email is one of four channels on the business tier and no sending discipline is described for it at all: warmup, authentication named by protocol, bounce or complaint thresholds, blocklist monitoring, placement testing, suppression handling and the sending architecture itself are all absent from the pages read.

The one pacing control published is the human like rhythm applied to social messages, which governs how automated contact appears rather than whether it arrives, so it belongs to the platform exposure question rather than to this one.

The pricing section opens by asking the reader whether they sell to consumers or to businesses and splits the product accordingly, which is a franker statement of market than most vendors manage and is the reason this sits at the top of the band. The rest works against it. Serving both markets from one engine means neither is described in any depth, and no headcount band, customer count, size distribution or ceiling appears anywhere.

The clearest signal of who actually buys this is indirect: the site is written almost entirely in Finnish, the guide library targets Finnish search terms, and the comparison pages are written against two domestic company data incumbents.