OnePgr

The removal test has an unusually literal answer here, because the pre model version of this company is still published on its own other website. Strip the substrate out of the current product and there is no account scoring, no parallel hypothesis testing, no drafted outreach, no prioritised queue with reasoning attached and no compounding across campaigns.

What remains is a dialer, a meeting page and a document room, which is precisely the older product still being sold on the apex domain. The vendor states the dependency plainly rather than leaving it to inference: every send, every connection, every recorded call and every campaign verdict writes into the substrate, and it is the reason the tenth campaign is smarter than the first. A company that describes itself as an artificial intelligence native service and prices finished outcomes rather than seats has staked the entire commercial model on the model layer working.

The oversight vocabulary here is specific in a way that suggests it was designed rather than added, and one phrase carries most of the weight: every action metered, every action receipted. Alongside it, behavioural scores are described as transparent with full engagement traces, every score is said to come with reasoning a representative can defend in a deal review, outcomes land in the customer relationship system the buyer already governs rather than in a vendor black box, and one of the seven entry doors is written explicitly for the person who audits the system.

For the self serve path a human role is stated plainly: the representative reviews, edits, sends and closes, and becomes the editor rather than the producer. Held off the top band on two counts. No approval queue, escalation threshold or role definition is actually specified behind any of that language.

And the delivered service model inverts the question, since the vendor operates the machinery on the customer's behalf, and what the customer can review or stop before something goes out in their name is undescribed.

Architecture is described and the models are not. The substrate is set out properly, with four named inputs writing into it, mail sends and replies, professional network connections and direct messages, recorded calls and transcripts, and campaign verdicts, and four named outputs reading out of it, the message most likely to land, the channel worth spending on, the persona who replies and the timing. That is more structural disclosure than most vendors offer.

What is absent everywhere is any identification of what does the work: no provider, no model family, no version, no hosting arrangement and no processing terms appear on any page read, and the protocol connector that external agents call names no model either. Five separately branded products sit in the stack and none of their sites was read for this build.

One real accountability mechanism and no customer evidence at all. The mechanism is the proof engagement: five thousand dollars, a written acceptance criterion, and a stated refund if the verdict does not materially advance the buyer's targeting. Putting the fee at risk against a written criterion is a commitment very few vendors in this index make, and it belongs on the record. Everything else is assertion.

No customer is named anywhere on the pages read, no logo, no testimonial and no quoted result. The one quantified claim is a productivity comparison, eight accounts touched per representative per day before against forty engaged after at the same quality, presented with no customer, sample, period or method, alongside an unattributed statement that two days a week are lost to notes and research packaging. A case studies page exists at its own route and was not opened, so this row is flagged.

This vendor sends at volume across three channels and in one commercial model sends on the customer's behalf, which makes the absence broader than usual. Mail sequences, a parallel dialer, text messaging, professional network connection requests and direct messages all appear across the two live sites, and the delivered service explicitly includes drafting and running the outreach.

Across every page read, no electronic marketing or telemarketing statute is named, no consent position is taken, and unsubscribe, suppression, complaint handling, calling windows and recipient removal are undescribed. The gap that matters most is the one the service model creates: when a third party operates outreach in the customer's name, who holds the consent obligation and who answers a complaint are questions a buyer needs settled before signing, and nothing on either site addresses them. No terms document was located on the current site, so this row is flagged rather than closed.

The current website carries no legal surface at all. Its footer has five sections covering roles, platform, machinery, contact and copyright, and not one of them links to a privacy policy, terms of service, cookie policy or processing addendum.

For a platform that holds recorded calls and transcripts, mailbox contents, professional network messages and behavioural traces across thousands of accounts, and that in its service model handles that material on the customer's behalf, the absence is conspicuous rather than incidental.

What was verified is architectural rather than legal and it does cut in the vendor's favour: enterprise customers are stated to run a dedicated substrate instance with full data isolation, and delivered outcomes are written into the customer's own system of record. Lawful basis, retention, sub processors, transfer mechanism and data subject routes were none of them established. The older site on the apex domain was not searched for legal pages and that check is flagged.

Two distinct provenance questions and neither is answered. The first is enrichment, where the vendor names two commercial data platforms it integrates with for finding and validating addresses, which is at least a named dependency, and says nothing about what those partners license or on what basis.

The second is more substantial and appears on the older site: the vendor offers data quality and enrichment as a service, described as ranging from filling information gaps in a customer's existing lists to procuring account and contact data against an ideal customer profile. Procuring contact data on request is a sourcing activity in its own right, and no supplier, method, coverage, refresh cadence, accuracy rate or lawful basis accompanies it. The substrate then accumulates behavioural signal about the people contacted, and where the boundary sits between a customer's own records and material assembled about third parties is drawn nowhere.

Two exposures, both real, neither carrying a position. Professional network activity is core rather than peripheral, with connections and direct messages named as one of the four inputs writing into the substrate and network outreach listed among the channels the platform synchronises, and no method, account model, rate posture or conformance statement appears anywhere.

Second, the older site advertises inbox rotation, described as spreading mail volume across multiple mailboxes to reduce spam risk, which is a mechanism this index has an established position on. Against those, the integration posture is otherwise cooperative and official: two systems of record, two marketing automation platforms, a workflow automation service, two conferencing platforms and a listing in a conferencing vendor's own application marketplace. Held at this band rather than lower because nothing rents an identity, and the rotation is framed as risk reduction rather than as evading a platform's limits.

The clearest answer to the cross client question found in this session, volunteered on the homepage rather than extracted from a contract, and stated in two parts with the trade off made explicit. First, shared learning is described as aggregate and anonymised, covering patterns about which messaging, channel and persona work, and specifically excluding the identities or content of the customer's counterparties.

Second, enterprise customers are stated to run a dedicated substrate instance with full data isolation, with the vendor's own framing that the customer's signal trains the customer's graph and the learning stays with their team rather than inside a vendor black box. Most vendors in this index either avoid the question or answer half of it. This one names what crosses accounts, names what does not, and offers a tier where nothing does.

Held off the top band because none of it is contractual on any surface read, no processing addendum or terms document was located on the current site, and anonymised carries no standard, method or verification, which is the same door left open elsewhere in this session.

The service model creates an authenticity question this index has not met in quite this form. In the delivered tiers the vendor states that it runs the experiments, scores the accounts, drafts the outreach and routes qualified opportunities, which means messages reaching a prospect were composed by a model operated by a third party and arrive appearing to come from the customer's own team. The recipient has two layers of authorship they are not told about rather than one.

Around it, a separately branded autonomous representative product sits in the stack, professional network direct messages are a named channel, and no position is taken anywhere on whether a person learns that a model wrote what they are reading, which is the European obligation now in force for systems interacting with people. Held at this band because no identity is manufactured or rented, sending runs from the customer's own connected mailboxes, and nothing in the marketing presents being undetected as a benefit.

A published protocol endpoint is the strongest item and it is live rather than announced, with its own dedicated page written for agent builders and platform engineers, describing tools an external agent calls to find contacts, enrich them, send sequences, handle replies and book meetings, with every action metered and receipted. That is an agent callable go to market stack rather than a connector list.

Around it the named integrations are broad and conventional: two systems of record, two marketing automation platforms, a workflow automation service, two conferencing platforms, two commercial enrichment platforms, and an official listing in a conferencing vendor's application marketplace. Held off the top band on documentation and coherence.

No developer documentation site, webhook description or marketplace of its own was located, and the machinery is fragmented across five separately branded products each on its own domain, which means a buyer assembling the whole picture is reading five websites rather than one.

Hosting provider, country, region and residency options are absent from every page read on either site, and no security or trust page exists where they would ordinarily sit. One architectural statement is on record and it is worth crediting even though it answers a different question: enterprise customers run a dedicated substrate instance with full data isolation, which addresses tenancy without addressing location.

For a platform holding recorded calls, transcripts, mailbox contents and professional network messages, and operating outreach on customers' behalf across territories, the absence of any jurisdictional statement is material. Corporate identity is thin: no entity name, registered address or jurisdiction appears anywhere, and the only contact details are a named individual's mail address and calendar link.

One genuine control statement and no verification of anything. The statement is tenancy isolation for enterprise customers on a dedicated substrate instance, which is a real architectural commitment and more than several vendors this session offer. Beside it sits a data minimisation gesture worth noting, that delivered outcomes land in the customer relationship system the buyer already governs rather than accumulating solely in vendor infrastructure. Past those two, nothing.

No audit, report, certification, attestation, penetration test, control set, trust centre, status page or vulnerability reporting route was located on either of this company's two live websites. That absence sits against a platform that records calls, stores transcripts, connects to mailboxes and, in its delivered service model, operates all of it on the customer's behalf. No legal documents were located on the current site to carry a control set instead, and that check is flagged.

One published figure carrying an unusually strong guarantee, and three tiers with no figure at all. The published one is the proof engagement at five thousand dollars per programme, with the scope stated concretely, the top forty accounts, eight to twelve weeks, twenty to forty hypotheses run in parallel, and a written acceptance criterion attached, refundable in full if the verdict does not materially advance the buyer's targeting.

A vendor putting its fee at risk against a stated criterion is publishing something adverse and that is what this axis rewards. Above it, the three recurring tiers read custom per engaged prospect, custom per scored account and custom per routed opportunity, so the ongoing commercial relationship has a stated unit and no rate, which means a buyer can understand the shape of the bill and not its size.

The claim of per outcome mathematics a finance leader can model sits directly against that. The current pricing page was not opened and is flagged, and the older site's pricing page renders four billing labels with no numbers behind them to a fetcher.

The structural half is genuinely good and the published half is empty. Structurally, the vendor states that delivered outcomes land in the customer relationship system the buyer already governs, which means the thing being bought accumulates in infrastructure the customer controls rather than in a vendor store they would have to extract from, and enterprise customers hold a dedicated substrate instance rather than a share of a common one. Both reduce the practical cost of leaving.

Against that, the substrate is described as the reason the outcomes compound, accumulating every send, reply, connection, recorded call and campaign verdict across the relationship, and nothing published says whether any of that accumulated learning or its underlying record leaves with a departing customer. No export function, format, post termination right, deletion timeline or retention period was located, and no terms document exists on the current site to carry them.

The one sending mechanism this vendor describes is the one this index has an established position on. Inbox rotation is presented on the older site as spreading mail volume across multiple mailboxes to reduce spam risk and ensure higher deliverability rates, and the position on file is that rotation framed as a way to distribute volume past what one mailbox sustains is not sending discipline but its opposite.

The wording compounds it twice over, promising to ensure higher rates, which no sender can guarantee because the receiving infrastructure decides, and pairing it with a parallel dialer and a stated jump from eight to forty accounts engaged per representative per day. Around that mechanism the apparatus is absent: no warmup ramp, complaint rate threshold, bounce handling policy, authentication guidance, suppression concept or caller reputation position appears anywhere on either site. Held at this band rather than lower because the framing is risk reduction rather than evasion, and because sending runs from the customer's own connected mailboxes.

Seven buyer roles, seven doors, and each one gets its own page with the pain stated in that role's language and a distinct entry product behind it. Revenue and marketing leaders are offered three audits ending in their own numbers. Representatives are offered a free loop measurable in fourteen days. Territory owners get weekly direction with defendable scoring. Marketing leaders get parallel hypothesis testing. Operations leaders are addressed as the people who audit the system.

Founders are addressed by situation rather than title, with no representative team and no quarter to build one. And a seventh door exists for agent builders and platform engineers, which is a buyer almost nothing else in this index acknowledges. Held off the top band because the coverage half is entirely absent: no customer count, no seat or headcount band, no region, no territory, no industry, no language, and no statement of who this would be wrong for.