SalesHunter AI

There is no non model product here. Every line item across four plans is an agent: a chatbot holding website conversations, a voice agent making and receiving calls and handling objections, messaging and social agents on the Meta channels, and on the automation plan an agent performing lead scoring, message personalisation and web search. Strip the models out and a customer is left with a chat widget, a phone number and an empty sequence builder.

The metering confirms it more precisely than the marketing does. Consumption is denominated in conversation credits and voice credits rather than seats, priced at roughly ten credits per chat exchange and twenty nine per voice minute, with worked examples per tier. A customer's bill moves directly with how much the models are used, which means the vendor's revenue and the model's activity are the same quantity.

The separate automation plan meters differently again, in automation credits plus a monthly allowance of enriched leads, but the enrichment and personalisation inside it are equally model driven.

What this vendor sells is model output measured by the unit. That is centrality in its strongest form, and it is why the disclosure gaps recorded elsewhere on this record matter as much as they do.

Ask what a conversation credit buys when a chat runs long, and what happens mid conversation when an allowance is exhausted.

Autonomy is high and the handoff control is real, which is the combination that holds this mid band.

Agents operate without supervision: the chatbot converses, the voice agent places and answers calls and handles objections from a knowledge base, appointments are booked into connected calendars, and the automation plan runs professional network and email sequences that branch on opens, clicks and replies.

Live chat with human handoff is included at every tier rather than gated to the expensive plans, which is the correct decision and the single most important oversight control in this product class. A supervisory chat surface and real time notifications sit alongside it, and the conditional sequence builder is authored by the customer rather than inferred by a model.

Knowledge base grounding constrains what an agent can draw on, and the vendor makes building it the setup step.

What is undescribed is the failure boundary. Two passes located nothing stating what an agent declines to answer, whether it can commit on the customer's behalf, whether it escalates on uncertainty rather than improvising, whether outbound call scripts or generated messages can be reviewed before use, and what audit record exists afterwards.

For voice agents deployed into healthcare and recruitment settings, the absence of a stated escalation rule is the gap that matters.

Ask what the agent refuses to discuss and what triggers an automatic handoff.

Two passes located no model, no provider, no version and no architecture on the retrieved pages.

One point of process honesty belongs in this note. Material crawled from this vendor's pricing page earlier in 2026 stated that on two of the plans each agent runs on a model the customer chooses, with cost depending on that choice. That would be genuine model transparency and rare in this index. It does not appear on the page as retrieved for this record, and the page carries a modification date after that crawl. It is therefore recorded as previously observed and not confirmed, and it is not credited to the grade. A buyer interested in model choice should ask directly rather than rely on either version.

What is published instead is consumption transparency, which is a different thing: credits per chat exchange and per voice minute are stated precisely, so a customer can forecast cost without knowing what produces the output.

The training commitment is narrowly worded and the wording matters. Data is stated never to be used to train public models, which leaves private, vendor or partner model training unaddressed.

No accuracy, word error rate or per language quality figure supports a claim of ninety five languages.

Ask which provider runs chat and voice, whether the customer can choose, and whether any model trains on conversations.

Eighteen testimonials are published and the pattern across them is the finding rather than any single one.

Every entry carries a personal name, a job title, an industry and a country flag, and not one names an employer. Every entry carries a dramatic quantified result: lead capture up 340 percent, reply rate from 4 to 22 percent, pipeline from 200,000 to 800,000 euros in a quarter, deliverability from 61 to 97 percent, three development representatives replaced, a rival data subscription cancelled. The set is distributed across thirteen countries at roughly one per target market, with several written in the local language, which is the composition a marketing team would design rather than the distribution a customer base produces.

One concrete error runs through it. An entry attributed to a commercial director in Mexico is displayed beside a Spanish flag. Details of that kind survive assembly and rarely survive collection.

Against eighteen glowing on site testimonials, the principal enterprise review platform shows a single review across four separate product profiles, and two passes located no case study, no named customer organisation and no methodology behind any figure.

A claimed retention rate of 94 percent appears in the questions section with nothing supporting it.

Ask for two reference customers who will take a call, with company names.

Squarely applicable across three regulated channels simultaneously, and addressed on one.

The messaging channel is handled correctly and the vendor says so plainly: social and messaging agents run through the official Meta business interface, and the questions section states that this keeps the customer's number safe and compliant while the agent handles conversations. That is the sanctioned route, named, and it is credited.

Everything else is absent. The automation plan sends professional network and email campaigns at unlimited volume, the voice agent places outbound calls, and the vendor supplies the enriched leads with phone and email attached that feed both. Two passes located no consent basis for contacting a supplied record, no suppression or unsubscribe handling, no do not call screening, no calling hour restriction, and no jurisdictional guidance of any kind.

The geographic spread makes the last omission material rather than theoretical. This vendor sells in five languages across at least thirteen countries spanning three continents, where electronic marketing consent, automated calling rules and recording requirements differ fundamentally. A customer in one market cannot infer the rules of another, and nothing here warns them the question exists.

Supplying the data and then dialling and mailing it concentrates the exposure in one place.

Ask how opt outs are captured and suppressed across voice, email and professional network, and on what basis a supplied number may be called.

Real commitments appear in the questions section and none is supported by a document.

The claims are end to end encryption, strict data protection protocols, and that customer data is never sold and never used to train public models. A privacy policy and terms are published. The operating entity is Spanish, so the vendor sits under European supervision directly rather than by assertion, which is worth something.

Two passes located no data processing agreement, no subprocessor list, no retention period and no stated processing location.

The subprocessor gap is concrete rather than formal here. Checkout and the agent client portal are both served from a third party infrastructure domain unrelated to the vendor, which indicates the conversational platform is built on or resold from another provider. That provider is named nowhere. A customer cannot reconcile a claim of end to end encryption with an undisclosed third party operating the conversation layer.

The holdings raise the stakes. Voice agents are marketed into healthcare clinics and recruitment agencies, so recorded conversations may contain patient enquiries and candidate details, and the enrichment product supplies personal contact data on individuals who never dealt with the vendor.

Ask who operates the agent platform, where conversations are stored, and for the processing agreement and subprocessor list.

The vendor is a data supplier and prices the data explicitly. The automation plan bundles fifty enriched leads a month on monthly billing or five hundred a month on annual, each described as matched to the customer's stated ideal customer profile and delivered with a phone number and an email address. A separate enrichment product is marketed as checking more than thirty data points in real time.

Two passes located nothing about where any of it originates. No source, no aggregator, no provider chain, no licence basis, no consent basis, no coverage statement by geography, no refresh cadence and no indemnity.

No accuracy figure is published by the vendor either. The only number available is inside a testimonial claiming an eighty seven percent email hit rate, which is a customer's assertion rather than a vendor commitment and cannot be held to.

The contrast with a peer built earlier in this session is instructive. Another vendor in this same cohort names six enrichment suppliers and two verification suppliers on its own pricing page and charges nothing for an unverified result. That establishes the disclosure is achievable in this category rather than commercially impossible, which makes silence a choice.

A European entity supplying personal contact data has a lawful basis question to answer and does not address it.

Ask where enriched records originate, on what lawful basis, and what verification is applied before delivery.

One channel is handled properly and explicitly, another is not addressed at all, and the split is the finding.

The messaging and social channels run through the official Meta business interface, stated plainly in the plan features and again in the questions section, where the vendor explains that this route keeps the customer's number safe and compliant. Naming the official programme rather than gesturing at an integration is the correct disclosure, and it demonstrates the vendor knows how to operate inside a platform's rules where a route exists.

The professional network is the gap. Multichannel campaigns on the automation plan run professional network sequences alongside email, and two passes located no partner programme, no sanctioned interface, no described mechanism and no trademark or non affiliation disclaimer. Other vendors in this cohort publish at least a disclaimer; this one publishes nothing.

A commerce platform assistant is separately productised, which presumably runs through that platform's app framework, though nothing states so.

Voice runs through a named telephony provider integration, which is the conventional and sanctioned arrangement.

The exposure on the unaddressed channel falls on the customer, since their account and credentials connect and enforcement restricts them.

Ask how professional network sequencing is performed and under which programme, and who bears an account restriction.

The single security claim fails the credential test in the familiar way. The questions section states that the vendor uses servers compliant with a service organisation control standard, which asserts the credential for the infrastructure rather than for this vendor, and does not name the provider holding it.

The training commitment is real and narrow. Data is stated never to be used to train public models, which is a genuine undertaking as far as it goes and leaves private, vendor or partner training entirely unaddressed. For a product whose agents conduct recorded voice conversations, that distinction is the whole question.

Two passes located no security page, no trust centre, no responsible disclosure route, no incident history, no tenant isolation statement, no retention period for conversations or recordings, and no encryption detail beyond the phrase itself.

The deployment context sharpens all of it. Voice agents are sold into healthcare clinics and recruitment agencies through dedicated vertical pages, so the conversations captured include patient enquiries and candidate screening, and one testimonial references a chatbot handling patient enquiries. Neither the retention nor the access position for that material is stated anywhere.

The undisclosed third party operating the agent platform, evident from the client portal and checkout domains, compounds every question above.

Ask who can access conversation recordings, how long they are kept, and what applies in the healthcare deployments.

The vendor markets undetectability as a benefit and publishes customer testimony saying so in terms.

One displayed testimonial states that website visitors cannot tell they are talking to artificial intelligence, immediately alongside a claim that lead capture rose 340 percent. Another states that the voice agent makes outbound calls that sound completely human, alongside a conversion increase. The vendor selected both quotations and placed them on its pricing page, so these are not incidental customer remarks but chosen selling points.

The product positioning matches. The home page describes the agents as human like and the company markets conversational logic agents that engage without a person present.

Two passes located no disclosure position of any kind: no statement that an agent identifies itself, no configurable disclosure line, no default opening script, no recording notice for voice calls, and no guidance to customers on whether their jurisdiction requires any of these.

The vertical pages make the exposure concrete rather than abstract. Patients contacting a clinic and candidates contacting a recruitment agency are the recipients here, both categories where an undisclosed automated interlocutor is a materially different matter from a retail chat widget, and both where call recording rules bind the customer.

Live human handoff exists, which helps once a person is reached but says nothing about the conversation before it.

Ask whether an agent can be configured to identify itself, and what recording notice plays on calls.

Coverage is adequate for the segment and shallow on documentation.

Four customer relationship systems integrate natively, including one aimed specifically at agencies, and the vendor states connection to thousands of further applications through two general automation platforms. Calendar tools, a mail provider, a telephony provider and the Meta business interface are each named. The automation plan adds single sign on and webhooks, which are the two integration primitives an operations team actually needs, and putting them on the mid priced plan rather than reserving them for enterprise is the right call.

A commerce platform assistant is separately productised.

The gaps are documentation and coherence. Two passes located no public interface reference, no rate limits, no developer documentation and no marketplace listing on any of the named platforms.

The structural point is more interesting. Customers log in through two entirely separate portals, one for sales automation and one for chat and voice agents, on different subdomains. That indicates two products sold under one brand rather than one platform, and nothing describes whether data moves between them, whether an enriched lead from the automation side can reach a voice agent, or whether a conversation on one appears on the other.

Ask whether the two portals share data, and for the interface documentation.

European operation is genuinely evidenced, which is more than most records in this index manage, and it is evidenced by artefacts rather than by statement.

The operating entity is Spanish and therefore under European supervision. Pricing is denominated in euros only. The site publishes in five European languages with a switcher. And both the checkout flow and the agent client portal carry an explicit European region path in their addresses, which indicates regional routing is actually implemented rather than promised.

What is missing is the vendor saying any of it. Two passes located no hosting provider named, no country, no region stated in words, no residency option, no single tenant deployment, no customer managed keys and no subprocessor list. A buyer has to infer residency from a checkout address, which is not a control.

The third party infrastructure domain serving both checkout and the agent portal is the unaddressed complication. It indicates the conversation layer runs on another company's platform, and where that company processes and stores voice recordings and chat transcripts is unstated.

For the healthcare and recruitment deployments the vendor markets, processing location is the question a customer's reviewer will open with.

Ask in which country conversations and recordings are stored, who operates the agent platform, and whether any residency commitment is contractual.

Two dedicated passes located no security page, no trust centre, no attestation, no certification held by this vendor, no auditor, no assessment period, no penetration test summary, no completed questionnaire, no subprocessor list, no status page and no responsible disclosure route.

The single security statement appears in a questions section and reads that the vendor uses servers compliant with a service organisation control standard. That claims the credential for whoever operates the servers rather than for the vendor, names neither the provider nor the report, and is the same borrowed credential pattern seen repeatedly in this index. Under the credential test there is no verb attaching the certification to this company and no scope boundary at all.

A claim of strict data protection protocols sits beside it with no supporting document.

The absence weighs more here than for a lighter product for two reasons. The vendor markets voice agents into healthcare clinics, where a customer will face sectoral review and have nothing to submit. And an undisclosed third party appears to operate the conversation platform, so even the borrowed claim points at an unnamed party.

The company publishes extensively across product, vertical, blog, events and template pages, so this is a choice about what to publish rather than a lack of capacity.

Ask what this entity itself holds, who operates the platform, and for that provider's report.

Strong on structure and undermined by three internal contradictions.

What works is substantial. Four plans publish monthly and annual rates in euros with the annual saving stated, a full side by side comparison table maps every feature across the agent tiers, and the metered units are defined precisely: roughly ten credits per chat exchange and twenty nine per voice minute, with worked monthly examples per tier and an interactive slider translating a credit allowance into messages and call minutes. A metered product whose consumption rates are published is forecastable, and most competitors leave the buyer to discover them.

The cancellation terms are unusually favourable and unusually specific. No contracts, cancel at any time, monthly plans on thirty days notice with no fee, and annual plans cancellable with a refund on unused months. A pro rata refund on an annual commitment is rare in this category and is disclosed rather than buried.

Trials are stated per plan at fourteen days on the agent tiers and seven on automation, with no card required.

The contradictions are on the same pages. Language support appears as more than fifty in the plan features and more than ninety five in the questions section. The support answer describes a Plus package with a dedicated success manager that does not exist among the four published plans. And the automation plan lists both the monthly and the annual lead allowance as though both applied to one subscription.

Ask which language figure is correct and whether a Plus plan exists.

The commercial half is genuinely good and published; the data half is undocumented.

Cancellation is stated plainly and favourably: no long term contracts, cancel at any time, monthly plans on thirty days notice without fees, and annual plans cancellable with a refund on unused months. Term lock is therefore minimal and a customer who wants to leave mid year recovers the balance, which is better than most annual arrangements in this index and better than several vendors several times the size.

Single sign on and webhooks on the automation plan give a partial technical route out, since webhooks can stream records to a system the customer controls while the account is live.

What two passes could not locate is any export route, file format, statement of what an export contains, or retention and deletion position after termination.

The accumulated material is substantial and split across two platforms. Conversation transcripts and voice recordings sit on the agent side, while sequence structures, campaign history and enriched lead records sit on the automation side, and nothing describes retrieval from either. The knowledge base a customer builds to ground its agents is a further asset with no described export.

The undisclosed third party operating the agent platform adds a question about who actually holds the conversation data at termination.

Ask what an export contains from each portal and what is deleted on cancellation.

Some real provision on the email side and nothing measurable behind it.

Account warm up is a named feature of the automation plan, which is the correct primary control and its inclusion rather than omission is credited. Unlimited email sending, a conditional sequence builder and a shared mail inbox complete the sending workflow.

What is absent is everything that would let a buyer judge whether it works. Two passes located no authentication guidance for the three sender records, no domain health diagnostics, no per domain throttling, no bounce handling, no inbox placement testing and no published delivery or placement figure from the vendor. The one number available sits inside a testimonial claiming deliverability moved from sixty one to ninety seven percent, which is a customer's claim rather than a vendor metric.

The enrichment product creates the specific risk. The vendor supplies email addresses with the plan and nothing states whether they are verified before a sequence sends to them, which is the standard route to a bounce spike on a warmed domain.

On the voice side there is no equivalent discipline at all: no caller identity reputation monitoring, no spam labelling remediation and no number rotation, despite outbound calling being a headline capability.

Ask whether supplied addresses are verified before sending, and what happens when a calling number is flagged.

Vertical selection is coherent and the coherence is the signal. Five dedicated pages address real estate, training academies, hotels, recruitment agencies and healthcare clinics. All five are appointment driven businesses with high inbound enquiry volume outside office hours, which is exactly the shape a voice and chat agent product fits. That is a considered market position rather than a scattered list.

European reach is evidenced rather than claimed. The site publishes in five languages with a switcher, pricing is euro denominated, the entity is Spanish, and testimonials span thirteen countries across three continents.

Seat structure indicates the buyer: five seats at entry, ten above, and the automation plan sold as one user with two accounts, which is a small team or an operator rather than a sales floor. Agency use is addressed directly in the testimonial set and by a customer relationship integration aimed at agencies.

The reservation is that the two product lines serve different buyers. Inbound agents for a clinic or hotel and outbound business to business automation for an agency are distinct motions with distinct evaluators, and they run on separate portals. A buyer arriving for one may find the other irrelevant.

Gaps: no supported country list for voice numbers, no data coverage statement by geography, no company size guidance.

Ask which countries voice numbers are available in and what enrichment coverage looks like in your market.