Hippo Video

The current marketing is almost entirely agentic, with three named agents covering creation, campaigns and a video assistant, and the page title leads on automating video creation. Apply the removal test and a complete, saleable product survives: a screen and webcam recorder with hosted playback, viewer analytics, sales landing pages, interactive elements and roughly forty integrations.

That is not a hypothetical remainder, it is an entire competitive category, and the vendor publishes comparison pages against two of its best known members. The packaging evidence points the same way. The free tier ships unlimited recording, personalisation, integrations and analytics with no generative capability of consequence, and the generative features are concentrated in the top tier, which is described in the vendor's own words as generative artificial intelligence for humanising videos. Two of four tiers are sold complete with the models absent.

The vertical product states the position most plainly: four agents handle enquiry capture, proposal creation with avatar video, engagement tracking and follow up automatically, and the customer's team just closes. On the core platform a campaign agent creates and distributes personalised video campaigns at scale, and a video assistant answers customer questions in real time in a synthetic likeness.

Nothing published describes an approval step before a generated video carrying a named employee's face and voice is sent to a prospect, which is the specific review this product needs more than most, since the output is a person's likeness saying words they did not say.

No guardrail, no escalation path, no statement of what an agent may not generate, and no action level audit trail appear anywhere, including in the tier comparison where the administrative controls are otherwise enumerated in detail.

Capabilities are named extensively and the machinery behind them is not. The site lists text to video, prompt to video, document to video, script generation, avatar generation, an editor that adjusts gestures and removes filler words, multilingual voiceover, and a real time conversational video assistant, each with its own page.

No model provider is identified for any of them, no model or version is named, no inference location is stated, and no description of how the avatar or the voice is synthesised appears. For a product that reconstructs a real person's speech and gestures, the absence of any statement about the generation method is more material than it would be for a text tool, because a buyer cannot assess quality, failure modes or the likelihood of the output misrepresenting them without knowing what produces it.

The logo wall is genuinely strong and nothing behind it is. Six recognisable organisations are displayed including a flag carrier airline, a global electronics manufacturer, a major cloud retailer, a contact centre platform and a software company. Four review platform badges from a single seasonal report are shown. Scale is stated at more than 5,000 businesses and 1.5 million users, and a funding round of 8 million dollars is named with four named investors.

Against all of that, the single customer quotation on the homepage carries a first name and a generic job description with no employer attached, which is a conspicuous gap immediately beneath logos of that size. No named customer is paired with a result anywhere on the pages read, no percentage or outcome figure carries a method, period or baseline, and the thirty day results section makes time saving and engagement claims with no basis at all. Recorded as observed and relevant to how current the whole claim set is: the site footer copyright reads 2023 across a homepage marketing 2026 capabilities.

Video messages are sent as outreach to prospects at campaign scale, through the seller's own mail client or through the platform's campaign engine, and no sending regulation is named anywhere on the pages read. There is no reference to the United States commercial email statute, no consent standard, no unsubscribe mechanism, no suppression process and no position on cold contact.

Compliance language on this site is entirely about data protection and health information privacy rather than about the conduct of the outreach itself, which is a recurring split in this index and a sharp one here because the vendor sells into healthcare and insurance where the contact rules are tighter rather than looser.

One structural point is recorded in the vendor's favour: a substantial share of sending happens through the customer's own mail client and sequencing tools, so the primary compliance obligation sits with software the buyer already governs.

The structure is properly built and its contents were not read on this pass, so this row grades what is demonstrably published rather than what it says. A privacy policy, a dedicated compliance page for the European data protection regulation, and a code of ethics and conduct are each published as separate documents in the site footer, and compliance with the United States health information privacy statute is claimed on the homepage, which is a meaningful commitment for a vendor whose customers record and transmit material in healthcare and insurance settings.

What could not be located from the pages read: a data protection contact, a data subject request route, a stated retention period, an international transfer position, and a list of the other parties that process customer data. The question this product raises most sharply is also the one least likely to be covered by a standard policy, namely the handling of the biometric source material behind a digital twin. Re verify at the privacy policy and the compliance page, either of which could move this row.

No contact database is bought, built or resold, and the platform supplies no prospects, which removes the entire class of provenance problem that dominates this axis elsewhere in the index. The corpus is the customer's own recorded and generated video plus the engagement telemetry it produces.

Two data flows do reach third parties and both are the customer's own choice: a prospect's company website and their profile on a professional network are pulled in as personalised video backgrounds, which is public material used as decoration rather than as an assembled record. Held off the top band on two populations the public material does not reach.

Viewers who watch a video generate detailed engagement records including play rate and watch duration held against a contact profile, and nothing addresses what rights those viewers have over that. And the digital twin is built from a real person's face, voice and gestures, which is the most sensitive input the platform takes, with no statement of where that source material is held, how long for, or what may be done with it.

Every third party surface the product touches is reached through a sanctioned channel. Recording and sending run through published extensions and add ins for the major browser and both major mail clients, distributed through their own stores. Integrations into systems of record and sales engagement platforms are native where the platform offers a partner path and extension based where it does not, and the vendor states which is which rather than blurring them.

A partner certification from a major platform is displayed. There is no scraping, no credential rotation, no multiple account handling and no automation of actions from a buyer's own social profile, so the failure modes that produce the low grades on this axis are absent.

Held off the top band because no conformance position is stated for any of it, and because the professional network personalisation feature, which pulls a prospect's profile background into a video, has no described mechanism, and pulling profile content into a commercial asset is exactly the kind of use that platform's terms address.

One published item is genuinely unusual and it is left unexplained. The lower tiers of the generative product line list trust and safety for artificial intelligence as an included feature, which no other vendor in this index names as a shipped capability, and nothing anywhere describes what it does, what it blocks or who governs it.

Around that gap the standard questions are unanswered: whether customer video, scripts or transcripts train any model, whether processing crosses tenants, which providers receive the material and on what terms, and how long anything is retained. The stewardship question with the sharpest edge here is the likeness itself.

The platform holds a reconstruction of a real person's face, voice and gestures capable of producing convincing footage of them saying anything, and nothing published states how that model is secured, whether it can be deleted, what happens to it when the person leaves the customer, or what prevents its use after that.

This is the most complete synthetic presence problem the index has graded, and the vendor's own naming is what settles the band. The product generates a digital twin of a named seller that reproduces their voice, gestures and style, speaks in more than thirty languages that person may not speak, and delivers demos, outreach and support in their likeness.

It is sold on being indistinguishable: lifelike avatars with natural gestures, humanlike support, reaching thousands without losing the human touch. A section of the homepage is headed with making artificial intelligence feel human, and a shipped feature is named for humanising it.

Nothing published states that a recipient is told the person addressing them is generated, no position is taken on the European transparency obligations effective August 2026, and no watermark, label or disclosure control appears anywhere in the tier comparison. The vendor's own stated mission, published on its about page, is to help businesses engage through authentic and real video, which is the direct opposite of what the flagship capability produces.

Two counterweights are recorded honestly and neither changes the outcome: the likeness belongs to the seller and is created with their consent rather than stolen from a third party, and the original recorder product, which most customers presumably use, is entirely authentic. The deception runs toward the recipient, and it is marketed as the benefit.

The conventional catalogue is the deepest graded in this index. Roughly forty integrations are named individually across systems of record, sales engagement platforms, marketing automation, conversation intelligence, recruiting, commerce and workplace tools, and the comparison table draws a distinction most vendors hide, marking which integrations are native and which run through a browser extension, which decides whether they work outside one browser.

A partner certification badge from a major platform is displayed. Beyond integrations there is a genuine developer surface: a software development kit with four documented interfaces covering recording and import, playback and embedding, editing, and reporting, offered for embedding video into another company's product. Native applications ship for four platforms.

Held off the top band on one gap that is conspicuous given the positioning: a vendor whose own page title leads with agentic artificial intelligence publishes no support for the protocol that lets external agents query it, no marketplace, no partner catalogue and no machine readable index file. The catalogue is built for a world of tools talking to tools rather than agents talking to vendors.

Where video, engagement data and avatar models are stored and processed is not stated on the public product surface. No region, data centre, cloud provider or residency commitment was located, and no residency option is offered to a buyer. The registered address published in the footer is a Delaware one and gives no indication of where processing happens. This carries more weight than it would for a text based product for two reasons.

Video is heavy and served from a content delivery network, so a buyer with residency obligations needs to know where the edge sits as well as where the origin does. And the platform is marketed into healthcare and insurance, where a buyer's own regulator will ask the question directly. The security page was not read on this pass and is where any of this would sit.

The certification set is named with the detail that separates a real disclosure from a badge wall. A service organisation control report at type two is stated with its type, an international standards certification is displayed with the certifying body identifiable on the badge itself, and compliance is claimed with both the European data protection regulation and the United States health information privacy statute, which matters because the vendor sells into healthcare and insurance verticals.

Behind those sit five distinct published documents: terms of service, a privacy policy, a dedicated compliance page for the European regulation, a code of ethics and conduct, and a standalone security page. Access controls appear in the tier comparison rather than only in marketing, with role based access, single sign on, user groups, video expiry and password protection, and restriction of viewing, each shown against the tier that carries it.

Held off the top band because none of it can be verified from outside: no audit period, no auditor named in text, no route to a report, no trust centre, no penetration test statement, no status page and no list of other processors were located. The security page was not read on this pass and should be re verified.

This is among the most complete pricing disclosures in the index and it covers two separate product lines rather than one. The video messaging line publishes a free tier and three paid tiers at 20, 60 and 80 dollars per user per month annually, with the monthly rate shown alongside every annual one. The text to video line publishes its own free tier plus tiers at 24 and 69 dollars, with a custom top tier.

Beneath that sits everything a buyer needs to compute the real bill rather than the sticker. A ten seat minimum is printed on the tier it applies to instead of being discovered at quote. A per unit overage rate of 5 dollars per video creation is published. A bandwidth ceiling of 200 gigabytes a month is published in the questions section along with how to check current usage and what happens on exceeding it.

Both add ons carry prices, at 1,000 dollars one time each, with the ownership terms stated. Cross line economics are published too, so a buyer of the top messaging tier is told they receive one text to video tier at no cost and a named discounted rate on another. A comparison table runs to roughly seventy rows across nine sections with every gate enumerated as a count or a cap. Payment methods, the absence of a setup fee, and the right to upgrade, downgrade or cancel at any time are all stated.

Terms of service are published, which keeps this above the bottom band, and they were not read on this pass. On the product side one real portability feature is enumerated in the tier comparison: video download is available, so a customer can retrieve the assets they created rather than only stream them. Cancellation at any time is stated plainly with no penalty described. What is missing is the whole post termination half.

No retention period, no deletion timeline, no deletion confirmation, and no statement of what happens to hosted video, sales pages, viewer engagement history or the avatar model once an account closes. The last of those is the one a buyer should press hardest on, since a customer leaving the platform is leaving behind a working synthetic likeness of their employees. Links already embedded in emails and pages also depend on hosting continuing, and nothing addresses what happens to them.

Video messages are sent at campaign scale, with automated personalised video counts published per tier running to several hundred a month, so the axis applies squarely. No control is named against it: no sender authentication standards, no warm up, no volume pacing, no bounce or complaint handling, no blocklist monitoring and no placement measurement.

Video outreach carries its own specific deliverability hazards that nothing addresses either, since animated thumbnails, embedded media, tracked links and custom sending domains all bear on whether a message reaches an inbox, and the product ships all four.

The mitigating structure is real and worth stating: a large share of sending runs through the customer's own mail client or their existing sequencing platform, so the buyer keeps their own sending reputation and the discipline sits with tools they already govern rather than with shared infrastructure here.

Segmentation is expressed as shipped product rather than as landing pages. Seven industries each carry their own page covering software services, travel, hospitality, real estate, healthcare, agencies and insurance, and five buyer personas are separately addressed across selling, pre sales, enablement, customer success and support.

The commitment goes further than most: an entire vertical product exists for hotel group sales, with its own brand, its own domain, its own agents and its own return on investment calculator, which is a far stronger signal of segment focus than a page. A ten seat minimum on the top tier is a published qualifying condition, and a separate pricing page exists for buyers outside the sales use case.

Held off the top band because the same site simultaneously claims sales, marketing, customer support, internal team communication, training and learning and development, which is close to claiming every function in a business, and because no headcount band, revenue band or ceiling is published anywhere. A buyer is told which industry they are in and never told whether they are the right size.