Sendr

Inference runs through most of the workflow and the platform underneath it would still function.

What depends on it: lipsync generation producing personalised video from a single recording, enrichment and data structuring driven by natural language rather than query building, a personality classification applied to individual prospects, lead scoring against the ideal profile, generated outreach copy, and qualification inside the unified inbox that prioritises high intent replies. Credits meter the consumption at 2,500 monthly on the entry plan, so usage is visible and bounded.

The video is the differentiator and it is genuinely central to the proposition, since the vendor positions the whole product around outreach that plays natively rather than around sequencing.

What keeps this out of the top band is that the sequencer, the contact database, the unified inbox, the hosted pages and the engagement tracking are conventional software. A customer who exhausted their credits would still have a working multi channel outbound platform with a large contact database and signal driven triggers, and would lose the personalisation layer rather than the product.

The honest reading is a competent outbound platform with a distinctive generated media layer on top, metered rather than bundled.

Ask what a credit buys and how many are consumed by one personalised video.

Real constraints exist on sending and none is described on generation.

The sending constraints are worth crediting and come from the vendor's own machine readable summary: built in sending limits, volume distribution across channels, and per channel compliance controls stated as protecting email and professional network account health. Those are enforced ceilings rather than guidance, and distributing volume rather than concentrating it is the behaviour that keeps accounts alive.

The unified inbox adds workflow oversight. Replies from all three channels land in one workspace, they can be assigned to team members, and model qualification prioritises high intent leads so nothing sits unattended. A fair usage policy is published as a separate document.

What has no described checkpoint is the generation. A personalised video is produced per prospect by substituting a word or phrase into a real recording, copy is generated, and a personality classification is applied to a named individual. Two passes located no approval step before any of that reaches a recipient, no audit trail of what was generated and sent, and no permission model beyond a workspace add on sold at 99 dollars monthly.

For a product whose output is a synthetic representation of a real employee, an approval gate is the control a buyer would expect.

Ask whether generated video can be reviewed before it sends and who may approve it.

The most consequential model output here is a synthetic video of a real person and nothing is disclosed about how it is made.

Two passes across the home page, the pricing section, the machine readable company summary and the feature descriptions located no model provider, no model name, no version, no statement of what data is transmitted for inference and no training or retention position. Lipsync generation, enrichment, copy generation, lead scoring and reply qualification all run without any technology named.

The personality classification is the sharpest instance. The interface shows a named prospect assessed with a result rendered as a colour category, presented as a completed analysis. That is a psychometric style judgement about an identifiable individual, produced from data they did not supply for that purpose, with no stated basis, no accuracy claim, no validation and no disclosure to the person classified. A buyer cannot establish what it is derived from or how often it is wrong, and the person being categorised has no idea it happened.

Nothing supports quality on the video either: no fidelity measure, no failure rate, and no statement of what a badly generated substitution looks like when it reaches a recipient.

Ask what the personality classification is derived from and how it was validated.

Two named references carrying real figures, and four headline metrics with no method behind any of them.

The references are the better half. A senior marketer at a major international newspaper is quoted by name stating that a marketing database with minimal previous traction produced 66 calls in two weeks. A named chief executive at a named company states 50 meetings from 500 product leaders, which is a denominator and a numerator rather than a percentage. A third customer appears in video. Named individuals at checkable organisations with stated denominators is meaningful attribution.

The headline metrics are weaker. Five million pages generated, a 3.1 times higher reply rate, 2.5 hours saved per campaign and twelve thousand meetings booked appear as a rotating band with no measurement window, no baseline and no sample.

A rating of 4.9 out of 5 is displayed twice, once at the top of the page, without naming the platform it comes from or linking to it. Two passes located no independent review base of any volume on any platform.

A customer spotlight page exists and was not opened in this pass.

Ask which platform the 4.9 rating comes from and over how many reviews.

The apparatus exists and its contents were not examined.

What is published, by name, in the site footer and the vendor's own machine readable summary: a data sourcing page, a fair usage policy, a privacy policy, a cookies policy and terms of service. The vendor states compliance with European data protection regulation and describes its contact data as compliant and ethically sourced, refreshed every thirty days. Built in sending limits, volume distribution and per channel compliance controls are stated as operating across email and the professional network.

Having a dedicated data sourcing page at all puts this ahead of every data carrying vendor built this session, none of which published one.

What could not be established in this pass, because the documents were not opened: the consent basis for contacting people in the database, whether unsubscribe is inserted and propagated across channels, whether a suppression list operates, and what the sending limits actually are numerically.

One channel is unaddressed on any retrieved surface. Business messaging on the platform named as the third channel operates under an opt in regime with template approval, and nothing in the retrieved material acknowledges either requirement.

Ask what consent basis covers messaging platform outreach and how opt outs propagate across all three channels.

Six documents published and none of them opened, with two credentials claimed on top.

The estate is real and unusually complete for an eleven person company: a privacy policy, a cookies policy, terms of service, a data sourcing page, an information security page and a fair usage policy, each as a separate document rather than a single catch all. Alongside them the vendor claims compliance with European data protection regulation and certification to the current revision of the international information security standard, displayed together in a footer block headed around certified standards and verified practices.

A European regulation claim from what appears to be a United Kingdom company carries more weight than the same claim from a vendor outside that regime, since it is the law that applies to them rather than one they are reaching for.

What could not be established: no data processing agreement, subprocessor list, retention schedule, residency statement or rights request process was located by name, and none of the six documents was opened in this pass.

The holdings warrant the questions. A contact database of several hundred million profiles, engagement telemetry down to scroll depth, personality classifications of named individuals, and reply correspondence from three channels.

Ask for the processing agreement, the subprocessor list and the retention period for personality classifications.

A dedicated data sourcing page, a stated refresh interval, and a record count that changes depending which page you read.

The good part is genuinely rare here. This vendor publishes a page specifically about where its data comes from, described in its own machine readable summary as covering compliant, ethically sourced business data refreshed every thirty days. Not one of the data carrying vendors built this session published such a page, and several published no provenance statement of any kind. A stated refresh interval is likewise concrete where competitors say regularly, and it bears directly on whether a contacted person still holds the role attributed to them.

Addresses and mobile numbers are described as verified.

The unreliable part is the count. The same asset appears as more than 500 million profiles in the feature section, more than 520 million on the pricing cards, and 479 million in a blog post from April 2026. Three figures on the vendor's own surfaces for the same database, with no reconciliation and no dated basis for any of them. A buyer cannot tell which is current, and the spread is large enough to matter.

No upstream supplier is named anywhere retrieved, and the sourcing page itself was not opened in this pass.

Ask which figure is current and which suppliers the database draws on.

Two platforms the vendor does not own carry its outreach, and it acknowledges the risk on one of them.

The acknowledgement is worth crediting because it is unusual. The vendor's own machine readable summary states that built in sending limits, volume distribution and per channel compliance controls exist specifically to protect email and professional network account health. That is a vendor naming account risk on a platform it does not control and describing a mechanism intended to manage it, rather than claiming the risk does not exist or advertising the absence of limits.

What it does not do is establish a basis. Two passes located no statement of interface access, no partnership, no licensing arrangement and no acknowledgement of the professional network's position on automated outreach. Managing the risk of enforcement is different from operating within permission.

The messaging platform is unaddressed entirely, and its business rules require opt in and template approval before commercial messages are sent.

Email runs through the customer's own sending, which is the sanctioned path, and a browser extension operates over the vendor's own hosted pages rather than a third party property.

Two passes located no scraping claim, no account renting and no undetectability marketing.

Ask under what agreement outreach reaches the professional network and the messaging platform.

The stewardship apparatus is strong for the size and the safety questions around synthetic media are unaddressed.

Stewardship first. Certification to the current revision of the international information security standard is claimed, a dedicated information security page is published, a fair usage policy exists as its own document, and an operational status indicator reports live. For an eleven person company that is a materially better position than most vendors several times the size built this session.

Safety is where the gap sits, and it is specific to what this product makes. The platform generates video in which a real, identifiable person appears to say words they did not say. Two passes located no consent mechanism establishing that the person whose likeness and voice are being modified has agreed, no limit on who may generate video from a given recording, no retention or deletion position for the source recording, and no statement of what happens to circulating videos when that employee leaves the company.

The personality classification carries its own absence, with no validation, accuracy measure or review process for a judgement made about a named individual.

No model provider is disclosed, so no training or retention position exists for anything processed.

Ask what consent is captured before a recording is used, and what happens when the person leaves.

The recipient's deception is quoted back as proof the product works.

The mechanism is stated plainly in the vendor's own interface copy: generate personalised video with a replaced word or phrase. One real recording is made, then lipsync substitution produces versions that appear to address each recipient individually. The accompanying email shown on the site reads as a first person claim to have recorded a video for that named person.

A prospect receiving it sees someone apparently speaking their name, and concludes a human spent time recording for them specifically. That inference is the entire mechanism, and the vendor's own material confirms it is the intended effect, quoting prospects who reply asking whether the video was recorded just for them and presenting those replies as evidence of success. The reciprocity the vendor describes depends on the recipient believing something untrue about how much effort was spent.

Two passes located no disclosure to the recipient that the video is synthetically modified, no watermark or indicator on the output, and no requirement in any retrieved material that customers disclose it.

What distinguishes this from the sharpest records in this index is real: the person on screen exists, works for the sender, and consented to being recorded. There is no fabricated identity and no rented account. The deception is narrow and it is still deception.

Ask whether generated video carries any indication it was modified.

Published interface documentation, a changelog and a roadmap, from an eleven person company.

That combination is the substance. Interface documentation runs on its own domain, a help centre on its own subdomain, and both a changelog and a public roadmap are linked from the footer. A vendor of this size publishing what it has shipped and what it intends to ship next has chosen to be inspectable, and a buyer can judge development pace before committing rather than after.

The unified inbox is itself an integration surface, consolidating replies from email, the professional network and a messaging platform into one workspace. A browser extension extends page creation into the seller's normal working context. Hosted pages run on the vendor's own domain. A partnerships page and a partner directory name implementation agencies, and a startup programme exists.

One small oddity worth recording for a future reader: the interface documentation sits on a different top level domain from the rest of the estate, which is easy to mistake for an unrelated property.

What two passes could not establish: whether connectors reach external record systems, whether webhooks exist, and whether the interface is public or gated, since the documentation was not opened.

Ask whether the interface is public and which record systems connect natively.

A certification implying an audited scope, and no location stated anywhere.

What exists: a claim of certification to the current revision of the international information security standard, which requires a defined scope and an audited management system rather than a self assessment; a dedicated information security page; and an operational status indicator reporting all systems operational, which implies monitoring worth publishing.

Circumstantial signals point to the United Kingdom. The compliance framing leads with European data protection regulation rather than treating it as an extension, a named reference customer is a major British newspaper, and the interface mock uses a British mobile number format. None of that is a residency statement.

What two passes could not locate: a hosting provider, a region, a data centre, a residency commitment, a tenancy or isolation model, an encryption statement, a backup position or a continuity plan. The security page was not opened and may address some of it.

The holdings make the question real rather than academic. Personalised pages are hosted on the vendor's own domain with five million claimed generated, alongside a contact database of several hundred million profiles, source video recordings of identifiable employees and reply correspondence from three channels.

Ask where the platform is hosted and whether the certification scope covers it.

An eleven person company claiming the current revision of the international information security standard, which no comparable vendor in this session does.

The claim is specific rather than vague. It names the standard and its 2022 revision rather than referring loosely to being certified, and it sits in a footer block headed around certified standards and verified practices alongside a European data protection compliance claim. A frequently asked question on the pricing section addresses the certification directly, which indicates buyers ask often enough to warrant answering.

Around it: a dedicated information security page, a fair usage policy, a privacy policy, a cookies policy, a data sourcing page and terms of service, all published separately, plus an operational status indicator.

Among the smaller vendors built across this session, several published a badge image with no text behind it and several published nothing at all. This one names a standard, a revision and a supporting page.

What holds it below the top band: two passes located no certificate document, no certification body, no scope statement, no certificate number, no expiry or surveillance date and no report request process. The security page was not opened. A claim of certification is verifiable in principle through the issuing body, and none is named to check against.

Ask which body issued the certificate, what scope it covers and when it expires.

Two tiers with figures, add ons priced separately, and a managed service published as a range rather than hidden.

The published structure: 119 dollars monthly carrying 2,500 credits, 250 lipsync videos, unlimited campaigns, the contact database and the enrichment studio. An annual bundle at 1,499 dollars carrying 30,000 credits, 3,000 videos, the automation builder, two stated months free and four named inclusions covering live onboarding, a full campaign build, a private messaging channel and the workspace add on. A free trial requires no card.

Two disclosures raise this above the ordinary. The workspace add on is priced separately and exactly at 99 dollars monthly rather than folded into a tier, and outreach channels appear as a modular add on with a counter, so a buyer can see that channels are purchased rather than assumed. And the managed service, where the vendor's own staff build the whole motion inside the customer workspace, is published as a range from 350 to 1,350 dollars monthly. Vendors almost always hide managed service pricing behind a call; publishing even a wide band lets a buyer decide whether to have the conversation.

Two questions on the pricing section address what a credit is and what happens when credits run out, which is the right pair to answer.

The deductions: the per action credit consumption rate sits behind interface tooltips rather than in readable text, and the outreach channel add on shows a zero price whose escalation is unexplained.

Ask how many credits one personalised video consumes and what channels cost beyond the first.

An interface exists and the hosted pages are the problem.

The route out is real in principle. Published interface documentation on its own domain implies programmatic extraction of contacts, campaigns and engagement data, and a customer with development capacity has somewhere to start. That is more than most records at this size offer.

What two passes could not locate: an export mechanism described in product terms, a format, a scope statement, a timeline, a deletion commitment or a retention position after cancellation.

The specific issue is the hosted pages. Personalised pages are served from the vendor's own domain, the vendor claims five million generated, and links to them are embedded in emails already sent, in messages on two other channels and in collateral shared onward by recipients. A customer who cancels does not merely lose access to a tool; every link they have ever sent stops resolving, including ones sitting in prospects' inboxes and forwarded internally at target accounts. Nothing addresses whether pages can be exported, redirected to a customer domain or kept alive.

Source video recordings raise a smaller version of the same question.

Ask what happens to hosted pages and their links after cancellation.

Sending limits and volume distribution are stated, and the payload creates a problem nothing addresses.

What exists: built in sending limits, volume distribution across channels and per channel compliance controls, stated in the vendor's own machine readable summary as protecting account health. Volume distribution is rotation described in different words and it is a genuine mechanism. A fair usage policy is published separately. Contact data refreshed every thirty days with verified addresses reduces bounces at source, which is deliverability work done upstream of sending rather than after it.

What two passes could not locate: sender authentication guidance, warmup, bounce handling, address verification at send time, inbox placement testing, reputation monitoring or any measured deliverability figure.

The unaddressed issue is specific to this product. The differentiator is animated previews embedded in the message and links to externally hosted video pages, which is precisely the payload profile that filtering systems weight against, and every page link points at the same vendor owned domain across the entire customer base. A shared sending domain reputation problem would affect every customer at once, and nothing describes how that domain's standing is monitored or protected.

Ask how the shared page domain's reputation is monitored and what the sending limits are numerically.

Five motions with dedicated pages, and a managed service for buyers who would rather not run it themselves.

The motions are named and separated: sales for consistent outbound at scale, marketing for moving quickly without a large stack including account based work, event follow up and database reactivation, agencies for standardising workflows and reporting, recruitment, and electronic commerce. Each carries its own page rather than a shared template, and the marketing description in particular names three distinct jobs rather than one.

The commercial ladder is coherent with that. A free trial with no card, an entry plan at 119 dollars monthly, an annual bundle, a workspace add on for teams, a startup programme, a partner directory of implementation agencies, and a managed service where the vendor's own demand generation staff build the motion inside the customer's workspace. A buyer can enter at any level of involvement from self serve to fully delegated.

What is missing: no geography is stated anywhere despite the database being described as global, no coverage breakdown by region, and no company size band. For a product whose data is its foundation, regional coverage is the question a buyer outside the vendor's home market needs answered first.

Ask how the contact database covers regions outside Europe and North America.