SBL

The model is the salesperson rather than an aid to one, and the product does not exist without it.

The vendor's own description of the mechanism leaves no ambiguity. The model opens in the customer's voice, takes over the chat after the first message, keeps the conversation moving toward a meeting, handles objections, qualifies the lead, follows up around the clock and books into the calendar, asking for the human only when it decides one is needed. It writes and personalises in more than 100 languages. A lead scoring agent decides which prospects are reached at all. Signals are surfaced by monitoring rather than by a list.

Cloned voice extends the same capability into audio, generating what the vendor describes as thousands of personalised audio messages.

Strip the model out and what remains is a pool of rented accounts and an inbox. There is no workflow underneath doing the work, no sequence a person configured that would still run, and no deterministic core the assistant merely accelerates. This is the clearest case of full centrality graded in this index.

What is absent belongs on the disclosure axis rather than this one: no provider, no version, no evaluation and no accuracy measure supports any of it.

Ask what proportion of conversations reach a booked meeting without any human turn.

A real human gate exists, it is optional, and the product is sold on switching it off.

The gate is described in three places and deserves credit. A draft mode lets the operator review every reply before it sends, or let the model reply on autopilot. Manual intervention whenever you want is listed as an entry plan feature. The comment to message flow includes verifying the outreach message before it goes as a direct message. A mobile application notifies the operator when the model needs a human, so escalation reaches someone rather than queuing.

Everything around it pushes the other way. The product is titled an autonomous representative. The marketing promises zero manual work, a calendar that fills itself, and outreach on autopilot. The model asks for the human only when it decides one is needed, which places the escalation decision inside the system rather than with the operator. Follow up continues until a meeting is booked.

Two passes located no audit trail of what was sent, no permission model beyond workspace separation, no approval requirement that cannot be disabled, and no record of which messages a human actually reviewed.

The combination matters because the conversations run through accounts belonging to third parties, so nobody with a stake in the account is necessarily reading what goes out under their name.

Ask whether draft mode can be enforced at workspace level and who can turn it off.

A persuasion model is the entire technical disclosure for a product that talks to strangers unsupervised.

Two passes across the marketing site, the pricing page, the profiles page and third party listings located no model provider, no model name, no version, no statement of what customer or prospect content is transmitted, and no description of where inference runs. The capability is referred to as a persuasive chat model built on behavioural science, which describes an intended effect rather than a technology.

The cloned voice feature carries no disclosure either. Nothing states what performs the cloning, what sample it requires, how the sample is stored, whether the person whose voice is cloned must consent, or what prevents a voice being cloned by someone other than its owner.

Quality is unaddressed. No evaluation, no accuracy figure, no error rate, and no statement of what happens when the model gets a fact about a prospect wrong mid conversation.

One surface points the other way and is worth noting precisely because it inverts the usual pattern. The vendor ships a model context protocol server so customers can drive campaigns from an assistant of their choosing. That discloses the customer's tooling and still says nothing about the vendor's own.

Ask which provider runs the conversation model and what reaches it from a prospect's profile.

Customer attribution is substantial and one headline citation does not survive checking.

The attribution is real. More than a dozen customers are named with linked company sites, executives are named with titles, and several testimonials carry a monthly spend against a monthly pipeline figure: 2,000 dollars against 9,000 added at an influence institute, 1,000 against 10,000 at an elder care business, 800 against 4,500 at a creator platform, 500 against 3,500 at an education company, and a 189 dollar plan against 2,000 added. A separate page publishes per customer profile counts, showing one customer running 60 rented profiles and others on 10, 15 and 20 or more. A case study section carries named clients with campaign structure.

The checkable claim fails. The site states a 4.7 out of 5 rating on the named review platform in three places and links to it. That platform's product profile displays a single review. An average of 4.7 cannot be derived from one rating, and a buyer following the link finds a different number from the one they were shown.

The headline figures around it carry no method: a 40 percent reply rate, 15 to 25 percent booked meetings against competitors placed at 2 to 5, more than 50 million messages, zero bans, 74 percent of meetings from non repliers, and a median 11 days to meeting. A volume calculator projects 162 to 270 meetings from 21,600 messages as customer averages, with no underlying data published.

Ask for the reply rate denominator and the source of the review platform figure.

Three regulatory badges sit on the home page with nothing behind any of them.

The badges name the European, Californian and Indian data protection regimes. Two passes located no processing agreement, no compliance page, no certification and no document supporting any of the three. A privacy policy exists and governs the website and the product relationship rather than the outreach itself.

The outreach specifics are absent throughout. No unsubscribe or stop mechanism is described for direct messages on either channel, no suppression list, no consent basis for contacting a person surfaced by monitoring their post activity or job change, no jurisdictional guidance, and no do not contact handling.

The messaging channel deserves separate mention. Business messaging on the messaging platform used here operates under an opt in requirement and a template approval process, and nothing in the vendor's material addresses either, while unlimited automated conversation handling on that channel is sold as a feature.

One detail from the vendor's own privacy policy compounds the picture. It directs opt out for its own site tracking to a third party identity resolution service, meaning the vendor resolves anonymous visitors to contactable identities on its own property and sends the person elsewhere to stop it.

Ask how a recipient stops contact, and what consent basis covers messaging platform outreach.

A privacy policy exists, acknowledges holding third party personal data, and stops there.

What it does say is worth recording because the acknowledgement is explicit. The vendor describes itself as operating at two levels and states it may have access to company information and to user lists comprising end users' names, telephone numbers and email addresses. That is a direct admission that the platform holds contact details of people who are not its customers and never agreed to anything with it.

What follows that admission is nothing. Two passes located no processing agreement, no supplier or subprocessor list, no retention schedule, no deletion commitment, no residency statement, no rights request process and no named privacy contact. The three regulatory badges on the home page are unaccompanied by documents.

The volume gives the gap weight. The vendor claims more than 50 million messages sent, each implying a held record for a person contacted, with monitoring of post activity, reactions, job changes and funding events layered on top.

A further detail sits in the policy itself: site tracking is operated through a third party identity resolution service, with opt out hosted on that service's domain rather than the vendor's.

Ask how long a contacted prospect's record is retained and how they request erasure.

Thousands of leads are included in the monthly price and the claim alongside them is that no lists are bought.

Both statements appear on the vendor's own surfaces. The entry plan includes 1,000 matched leads a month with an imputed value of 600 dollars and the middle plan 2,000 with a value of 1,000 dollars. Elsewhere the vendor states that it finds buyers already showing intent, with no list buying. Leads that are neither bought nor supplied by the customer have to come from somewhere, and the vendor never says where.

One route is stated openly enough to answer part of the question. Unlimited extraction from a professional network search URL is a listed plan feature, described as pulling every matching lead automatically with no exporting and no risk. That is bulk extraction from a licensed subscription product, performed at unlimited volume, and it is sold as a headline capability.

The signal layer is harvested the same way, monitoring post comments and reactions around the clock, job changes, funding events and engagement on competitors' posts.

Two passes located no data supplier, no licence, no interface agreement, no verification step and no statement of lawful basis for holding any of it.

A buyer inherits whatever the collection method turns out to be, since the records arrive in their workspace.

Ask where the included monthly leads originate and under what licence.

Circumvention is not a side effect here; it is the product, and the vendor sells it by name.

The headline on the profiles page reads: scale your LinkedIn outreach with unlimited rented LinkedIn accounts. Accounts are rented at 65 dollars monthly, 59 above ten, recruited through an affiliate route, and described as pre warmed with real activity history. Account sharing and transfer is among the clearest prohibitions in the professional network's user agreement, and the vendor's business model is to broker it.

Evasion is then marketed as a benefit rather than mitigated. The product is said to behave like a real human and respect the platform's limits, with zero bans across more than 50 million messages presented as a headline statistic, and a student facing page claiming a 0.3 percent restriction rate against an industry average of 12 to 15 percent. A vendor publishing its own restriction rate as a competitive metric is measuring how well it avoids enforcement.

Automatic sender rotation is described precisely: profiles switch when one hits 75 requests and 200 messages a day so sending never stops. Rotating accounts to stay under a per account threshold is what a rate limit exists to prevent.

Add unlimited extraction from a licensed search product, automated likes, profile visits and comments, comment capture on competitors' posts, and unaddressed automation of a messaging platform with its own business rules.

Ask who owns the rented accounts and what happens to the customer's pipeline when one is closed.

An unsupervised model holds conversations with strangers using a cloned human voice, and no guardrail is described anywhere.

The cloned voice is the sharpest element. The vendor offers to clone a voice once and then send thousands of personalised audio messages that feel human. Two passes located no consent mechanism establishing that the voice belongs to the person cloning it, no verification, no watermark or disclosure on the resulting audio, no retention statement for the voice sample and no limit on who may use a cloned voice once it exists.

The conversational layer runs with comparable freedom. The model handles objections and qualifies prospects with draft review available but optional, and decides for itself when to involve a human. Nothing describes what it may not say, what happens when it asserts something false about a prospect or a product, or how a conversation is stopped once it has gone wrong.

Stewardship is empty. No retention schedule, no deletion commitment, no encryption statement, no access control description, no incident or breach process, no security contact and no vulnerability disclosure route were located.

The volume claimed makes the absence material rather than theoretical, and the conversations run through accounts belonging to people who are not the operator.

Ask what consent is required before a voice is cloned and how a recipient learns the audio is synthetic.

The vendor is asked directly whether the recipient will know, answers no, and offers the reply rate as proof the deception holds.

The question appears in the site's own frequently asked questions. Will my prospects know the message was automated. The published answer states that based on the reply rates they will not, and that this is the proof, reasoning that messages read as hand written and that if they read as automated such reply rates would not be possible. That is undetectability presented as a validated outcome with evidence attached, and it is the most explicit instance recorded in this index.

The identity beneath the message compounds it. Outreach runs from rented accounts belonging to third parties, pre warmed with real activity history, rotated automatically. A prospect therefore receives a message from a real named professional who does not work for the customer, was rented for 59 dollars a month, and did not write it. Replying reaches a model. Booking produces a meeting with someone else entirely. Three different identities occupy one conversation and the recipient is told about none of them.

Cloned voice notes extend the same construction into audio designed, in the vendor's words, to feel human.

Nothing described discloses automation, synthetic audio or the sender's true relationship to the offer at any point.

Ask what the recipient is told about who is messaging them and who wrote it.

The developer surface is one of the deepest in this sweep and deserves crediting on its own terms.

A public interface is offered with representational state transfer, webhooks and delegated authorisation named specifically rather than gestured at, letting a customer trigger campaigns, pull leads and synchronise signals. Webhook capability is tiered openly, static on the entry plan and custom above it.

A model context protocol server ships alongside, allowing the whole outreach motion to be driven in natural language from an assistant: creating campaigns, adding leads and checking conversations. That is the fifth such server encountered in this sweep and among the most complete in scope, since it exposes campaign creation rather than read only lookup. A public code repository accompanies it.

Connectors cover three customer record systems, a team messaging platform, a workspace tool and three automation platforms, with two way synchronisation claimed for pipeline data.

Two things hold it below the top band. The interface documentation link routes to a signup wall rather than to public documentation, so a buyer cannot evaluate the surface before creating an account. And the navigation item labelled for the interface and the protocol server points at a video channel rather than reference material.

Ask for public interface documentation and the protocol server's tool list before signing up.

Multi tenant software as a service is the entire deployment disclosure.

Two passes across the marketing site, the pricing page, the profiles page and the privacy policy located no hosting provider, no region, no residency commitment, no tenancy or isolation model, no encryption statement covering data at rest or in transit, no backup or continuity position and no self hosting option.

The absence carries weight because of what is held. The platform stores contact records for people who never agreed to anything with it, complete conversation histories from more than 50 million claimed messages, monitored activity signals, and cloned voice samples belonging to identifiable individuals. Where any of that sits is unstated.

A regional data protection badge on the home page implies one jurisdiction and the company is based there, but an implication drawn from a badge is not a residency statement and no buyer review process can act on it. Customers are claimed in more than ten countries, including markets whose transfer rules would require an answer.

The unified inbox and mobile application mean the same conversation data is reachable from multiple client surfaces, none of which carries a stated security posture.

Ask where conversation data and voice samples are stored, and under which jurisdiction.

Three regulatory badges near the top of the home page are the whole of the security disclosure.

The badges name European, Californian and Indian data protection regimes. They are graphics. None links to a document, a scope statement, an assessment or an attestation, and naming a regulation is in any case a compliance claim rather than a security credential.

Two passes located no audit report of any type, no international standard certification, no penetration test summary, no trust portal, no security page, no report request process, no encryption or access control description, no named security contact and no vulnerability disclosure route.

The positioning makes the gap harder to set aside. The platform holds conversation histories, prospect contact records, cloned voice samples and delegated access to accounts on two external platforms, and the vendor sells an enterprise tier and offers custom terms for buyers with, in its own words, compliance requirements. A buyer arriving with those requirements finds nothing to assess.

One review platform profile carries a single review, so even the informal signal a security reviewer might fall back on is thin.

Ask whether any independent security assessment exists and what it covered.

The pricing page is among the most granular in this index, and two of the vendor's own surfaces contradict each other on what the product costs.

What is published: three tiers at 99, 189 and 997 dollars monthly, a full feature comparison table across them, outreach seats included per tier with the incremental seat priced exactly at 39 dollars monthly, monthly lead volumes per tier, team members included per tier, webhook type per tier, and rented profiles priced separately and openly at 65 dollars monthly falling to 59 above ten profiles. A seven day trial requires no card, cancellation is stated as available at any time with no lock ins, and an enterprise tier is marked custom with a contact route. A public calculator converts a target message volume into the number of profiles needed.

Unit level disclosure of that depth is rare here and it is the reason this sits in the upper band.

Four things pull against it. A student facing page states that the same platform is what enterprises pay 8,000 dollars for, which does not reconcile with a published top tier of 997. The top tier is priced while marked coming soon and available by invitation only. A 90 percent money back guarantee appears with an asterisk and terms that were not located. And a countdown timer offering a discount runs at zero across page loads, which is manufactured urgency on a page whose value is otherwise its candour.

Ask what the 8,000 dollar enterprise arrangement includes that the published tiers do not.

Export exists, sits behind a tier, and stops at the boundary of what a departing customer would need.

What is offered: a built in customer record system with export, listed as a middle tier feature, exportable per account analytics described as ready to drop into a report, a public interface with webhooks for programmatic extraction, and two way synchronisation into three external record systems, which gives a customer running a connector a live copy of pipeline data outside the platform throughout.

Commercially the exit is clean. Cancellation is stated as available at any time with no lock ins and no minimum commitment, on monthly terms.

The gaps are the usual ones and one unusual one. Export is not listed on the entry plan, so the cheapest customers have no stated route out. No format, no scope statement covering whether conversation histories and monitored signals travel with the contacts, no timeline and no deletion or retention commitment after cancellation were located.

The unusual gap concerns the rented accounts. The conversation history, the relationships built and the connection graph accumulate inside profiles the vendor controls and the customer never owned, so a departing customer cannot take the network with them even in principle. That is a structural exit cost the pricing page does not mention.

Ask what happens to conversations and connections held in rented profiles after cancellation.

The discipline described here is more specific than most records in this index, and its stated purpose is avoiding enforcement rather than protecting recipients.

The specifics are real and published. Per profile ceilings are given as 75 connection requests and 200 messages a day. Profiles are pre warmed with real activity history before use. Sender rotation switches accounts automatically once one reaches its daily limit. Sequencing warms a prospect with automatic likes, profile visits and comments before any pitch. The system is said to warn an operator before a limit is reached.

Publishing a numeric daily ceiling is more than most vendors on this axis manage, and warming before contact is a genuine practice rather than a slogan.

What the mechanism is for is stated just as plainly. The section carrying these features is headed around never getting banned, and rotation exists so that sending never stops once a single account is exhausted, which is the operation a per account limit is designed to prevent. Warming establishes plausibility rather than reputation.

On the messaging channel nothing equivalent appears: no volume guidance, no template rules, no quality rating handling.

No measured outcome is published on either channel beyond the reply rate claims, and nothing addresses what a recipient experiences when several rented profiles contact the same person.

Ask what prevents two rented profiles contacting the same prospect in one campaign.

Three buyer types are drawn with distinct propositions rather than listed as labels.

For founders the pitch is running a full outbound motion without hiring a representative, with a first booked meeting in the first week and a personal account left clean. For sales teams it is five to ten times more meetings per representative with record system synchronisation and the model handling replies and objections. For agencies it is a separate workspace and profile per client under one flat fee, unlimited profiles, and every client run from a single dashboard. Each carries its own stated outcome, which is more useful than a persona list.

A separate student programme runs on its own subdomain with its own pricing and contact route, and an invitation only top tier gates the largest deployments.

Coverage claims are 5,000 teams across more than ten countries with personalisation in over 100 languages.

What is missing keeps it below the top band. No industry vertical is named, no company size band is given, and no geography is stated beyond the country count, while the named customer roster is concentrated in one market across education, healthcare, immigration services and technology. A buyer elsewhere cannot tell whether the rented profile pool, the lead supply or the language handling suits their market.

The one hard prerequisite goes unstated: the model only works where the buyer accepts outreach from an identity that is not theirs.

Ask which markets the rented profile pool actually covers.