BixJet

AI language saturates the marketing, the page headline is AI powered outreach automation and the pitch is AI at every step, but the vendor's own packaging settles the removal test decisively. AI Agents are labelled Add-on in the product navigation, sitting outside the Core Platform group that holds Lead Finder, Outreach Automation, Lead Capture, Contacts and CRM, Social Media and API Integrations.

Strip the models out and a complete, saleable product remains: a 275M contact database, eight channel sequencing, server side LinkedIn automation, forms and landing pages, social scheduling, appointments and a CRM. This is the same class of evidence as Apollo's price list, where the entry tier ships with no AI at all, and it is the cleanest kind there is, because it is the vendor rather than the analyst deciding what is core.

The product is sold in two modes and the second is fully autonomous. A visual sequence builder puts the operator in control of the flow, and the interface surfaces a real, operator facing governor, a live invite counter reading 42 of 80 today with limits respected and zero warnings, which is more oversight instrumentation than most tools in this band expose.

But the alternative pitch is explicit, let AI create and run the whole sequence for you, and the published agent transcript shows the agent finding contacts, saving and tagging them to the CRM, then building and launching a live LinkedIn sequence in one turn. AI auto reply then handles inbound messages. Nothing documents an approval step, a withholding guardrail, an escalation path or an audit trail of agent actions, which is the bar Agentforce set on this axis.

No model provider, family or version is named anywhere on the retrievable surface, and nothing describes where inference runs, what is sent to it, or how the reply classifier behind AI auto reply decides what an inbound message means. The product writes outbound copy, writes public comments on prospects' posts, and answers inbound messages on the operator's behalf across eight channels, so the model is doing identity bearing work in three distinct modes, and none of it is attributed.

Better than the category floor on attribution, worse on substantiation. Genuine credit first: customer logos are named organisations rather than anonymous images, and testimonials carry full names, job titles and companies, which is more than several vendors already in this index manage.

Against that, every quantified claim is unsourced and several are implausible on their face: 10X more meetings booked, 90 percent less manual work, and a plus 320 percent revenue growth figure that appears inside an illustrative dashboard graphic alongside invented counts for connections, replies and meetings.

Recorded as observed rather than concluded: the same page states trusted by 300 plus businesses in the logo strip and join 1,000 plus companies in the closing call to action, so the vendor's own customer count is inconsistent within a single scroll. No case study, no named outcome with a measurement basis, no analyst coverage.

Bottom of band, and it now holds the record in this index for regulated surface count. Eight or more channels are sequenced from one builder: email through Gmail and Outlook, LinkedIn, SMS, WhatsApp, Facebook Messenger, Instagram DM, X DM and Slack.

That set carries CAN-SPAM and ePrivacy exposure on email, TCPA exposure on SMS where statutory damages run per message, and WhatsApp's own business messaging policy on unsolicited contact, and it is reached with no regulation named anywhere, no consent framework, no suppression or opt out concept and no do not contact handling described. Buzz previously held this record with five surfaces and the same silence.

The one control present is a Do Not Sell or Share My Info link, which governs the vendor's own website data rather than the outreach the product performs. Graded C rather than lower because the failure here is silence, not the affirmative evasion marketing that is graded separately under Platform Terms.

Recorded as observed rather than concluded: a Privacy Policy and a Cookie Policy are published and linked from both the header resources menu and the footer, and a Do Not Sell or Share My Info control is present, which is a real CCPA signal, but the policy pages returned server errors on repeated retrieval attempts at grading time and their contents could not be read.

This is a failure to retrieve documentation the vendor demonstrably publishes, not a finding that the documentation is absent or inadequate. A targeted second search for any BixJet privacy or security record returned nothing about this vendor at all. Re verify directly at the published URLs before this grade is quoted anywhere, because a readable policy could move it in either direction.

The vendor makes a provenance claim and then does not support it. The 275M plus contact database is marketed as search, enrich, verify, with no data vendor, which asserts that the inventory is built rather than licensed, but no sourcing route, supplier, collection method or refresh mechanism is named, no data broker registration is disclosed, and no Article 14 notice or self service removal path for the people in the database was located.

The second half is sharper and specific to this product: the harvesting surface extends past search results to Sales Navigator lists, groups, and the people who liked a given post, each enriched with a verified email and a direct phone and exportable. Converting a public engagement signal on someone else's content into a contact record carrying a personal phone number is a provenance question the vendor does not acknowledge exists. Apollo holds the A on this axis for naming three sourcing routes, naming the extracting features, and publishing an actual Article 14 discharge; none of those three pieces is present here.

The sharpest exposure on this axis in the index, and every word of the evidence is the vendor's own marketing. Four claims compound. First, the architecture is server side: cloud sending from a dedicated cloud IP, described as API level, not a browser plugin, which for LinkedIn means automated action against endpoints for which no public partner API exists for connection requests or messaging at this scale.

Second, the outcome is sold as evasion: LinkedIn on autopilot without risking your account, with randomized delays, working hours windows and activity limits that mirror normal usage, plus cloud safe as a headline. Mirroring normal usage is a description of appearing to be something you are not.

Third, the automated action set is the broadest graded, adding automated commenting on prospect posts and AI auto reply to inbound messages on top of invites, messages and profile visits, and it is offered on free LinkedIn accounts with multiple accounts supported. Fourth, the same exposure is replicated across Facebook, Instagram, X and WhatsApp, whose terms likewise prohibit automated messaging, so a single sequence can put four separate platform accounts at risk at once. The buyer holds every one of those accounts. This is the direct architectural inverse of Bindago, graded in the same session and the same category, which never takes custody of the credential at all.

Recorded as observed rather than concluded: nothing on the retrievable surface addresses model training, tenant isolation, prompt retention or the boundary between one customer's data and another's, and the Data Security page linked in the navigation returned a server error at grading time.

The nearest statement is an ownership one, that the data stays yours and contacts are yours to keep and export even if you leave, which answers who owns the records and says nothing about what the models see. That distinction matters more here than usual, because the platform ingests connected mailboxes, LinkedIn inboxes, WhatsApp and SMS threads and CRM records into one system and then runs models over all of it. This is a failure to locate documentation, not a finding that none exists; re verify at the Data Security page.

Bottom of band, and this record adds a shape the index has not previously graded. The familiar surfaces are all present: AI written outbound personalisation, machine text inside live two way exchanges through AI auto reply, and no Article 50 position, no disclosure line and no acknowledgement that the obligation exists. The new one is automated commenting on prospects' posts.

Every other synthetic surface graded so far is private between the sender and one recipient; a generated comment is published under the operator's real name, on another person's content, where the prospect's whole network reads it and reasonably assumes a human chose to write it. That makes the audience for the deception larger than the target of it.

Held at C for consistency with Buzz, whose ringless voicemail was graded at the same level, but automated public commentary is the strongest candidate the index has yet produced for tightening this band.

A published API documentation page with developer endpoints and reference, plus a dedicated API integrations product page, is more developer surface than most vendors of this size expose and is worth the credit. But the integration strategy is deliberately closed, and the vendor says so in its own selling: replaces four subscriptions, not one; one record, no integration project; no Zapier, no field mapping.

A platform whose pitch is that you will not need to connect it to anything has, by design, a shallow connector surface, and no named third party integration, marketplace listing or app directory entry was located to verify depth independently, which is the standard Autotouch met for a B in this index. The trade is coherent for the buyer it targets and it is still a narrow ecosystem.

Recorded as observed rather than concluded: no hosting region, residency option, sub processor list or infrastructure provider was located on the retrievable surface. The only architectural detail published is that LinkedIn automation runs from a dedicated cloud IP, which describes the sending path rather than where customer records live.

The corporate entity is The Simplified Biz Inc and the site's own geographic metadata gives Dover, Delaware, which establishes the jurisdiction of the vendor but not of the data. For a platform ingesting mailboxes, message threads and a CRM for European and global buyers, the absence of any residency statement is material. This is a failure to locate, not a finding of absence; the Data Security page in the navigation is the obvious place to re verify.

Recorded as observed rather than concluded, and the Bigtincan discipline was applied before settling this. The vendor publishes a Data Security page in its primary navigation, described as how we protect your data, so a security surface demonstrably exists and the vendor thought it worth a top level slot. That page returned a server error on repeated retrieval at grading time, and a targeted second search for a BixJet security or certification record surfaced nothing about this vendor.

No certification claim, audit report, trust centre, penetration test or vulnerability disclosure route is visible anywhere on the retrievable surface. This is a documented failure to reach published material rather than a finding that controls are absent, and it should be re verified directly before the grade is quoted. Note what is at stake if it stays unanswered: the platform holds authenticated sessions for the buyer's LinkedIn, mailbox, WhatsApp and social accounts simultaneously.

The numbers are published and the framing around them is not accurate. Three tiers exist: a free trial plan, Growth at 49 USD per user per month, and Professional at 79 USD per user per month billed annually or 99 billed monthly, with regional pricing on the pricing page. That is a legible ladder and the homepage FAQ states the headline figure plainly rather than burying it. What does not hold up is the marketing wrapped around it. The homepage says One platform.

One price, and repeats that every feature is in the Professional plan, which reads as a single tier product with no gates; the pricing page shows a cheaper Growth tier beneath it with a narrower feature set, so there is a gate and the buyer meets it after the pitch rather than during it. Held at B rather than lower because the prices themselves are real, published and per user, and rather than higher for two reasons.

The ActiveCampaign precedent applies directly, that an A is unavailable where a load bearing add on is unpriced, and AI Agents are labelled an add on in the vendor's own navigation while AI powered outreach is the product headline. And usage economics stay invisible: nothing published states how many of the 275M contacts a seat may reveal, enrich or export in a period, which is the number that decides the real bill on any database backed product.

The vendor answers the exit question directly, in the buyer's own words, in a homepage FAQ, which is rare enough to note: do I keep the lead data BixJet collects, answered yes, contacts you find, enrich or export including verified emails and direct phone numbers are yours to keep and export at any time, even if you leave.

Bulk export of connections, profiles, emails and phones is a marketed feature rather than a metered credit, and the vendor makes the point explicitly that the data stays yours. That is the direct inverse of the credit metered export pattern that holds Apollo at C, where a customer cannot leave with their own revealed records faster than their plan permits.

Not A because the commitment covers lead records only and stops there: nothing states a deletion timeline, offers a deletion confirmation artefact, describes export scope or format for CRM history, message threads and campaign data, or sets out post termination retention on the vendor side.

Split by channel, and the larger channel is the unprotected one. On LinkedIn the discipline is real and published: warm up ramping, randomized delays, working hours windows, daily invite and message caps, and a live limit counter in the interface.

On email, which is the highest volume surface the product sends on, nothing was located at all: no mailbox warmup, no domain health monitoring, no inbox placement testing, no bounce or complaint thresholds, no volume governance and no ramp guidance, on a platform that sequences cold email at scale from the buyer's own connected Gmail or Outlook account with open tracking enabled.

That is the configuration that produced the first D on this axis in the index at Apollo, and the buyer's sending domain carries the consequence. Buzz reached B on strictly less than this by including mailbox warmup; the absence of that one control is the difference.

The intended buyer is clear and the evidence supports it: small businesses and SMEs, a single flat per user price, self service signup, live in five minutes with no IT team and no onboarding, and a named customer set made up almost entirely of consultancies, agencies and micro businesses. What holds this to C is that the vendor's own positioning does not stay in that lane.

It markets itself as an alternative to ZoomInfo and HubSpot as well as to Expandi, Waalaxy and Dripify, which spans two very different weight classes, and the customer count is stated inconsistently as 300 plus businesses and 1,000 plus companies on the same page. No enterprise apparatus was located, no single sign on, role based access, administrative controls or procurement surface, which is consistent with the SMB read and inconsistent with the comparison set the vendor chooses.