heddl

A genuine middle case, and the same shape as the other vendor in this index that sits here. Strip the models out and a working multichannel sequencer survives, covering LinkedIn, email and WhatsApp with a unified inbox, and that is a saleable product because a named competitor sells exactly it. So this is not the top band. But what survives is precisely what the vendor spends its entire comparison table arguing is insufficient.

Every element of the differentiation is model dependent: the classification of ingested posts into six signal types, the strategy dashboard built from parallel analysis jobs, the weekly content calendar, the personalised opening lines generated from a prospect's own activity, and the planned module that turns a plain English description into a live campaign. Remove the models and the vendor becomes the competitor it defines itself against.

Sequences run automatically once launched, the planned campaign builder turns a single plain English instruction into a seven step multichannel drip with generated copy for every contact in under thirty seconds, and the planned monitor is sold on requiring zero manual oversight.

Nothing published describes an approval gate before generated copy reaches a real person, a review step on the campaign the builder assembles, a confidence threshold on the scoring that decides who gets contacted, or a statement of what the system may not do. One genuine control exists and it keeps this off the floor: the professional network compliance queue writes an activity log that the customer's operations and compliance staff can audit from the dashboard.

That is an auditable record of what the system did, which several much larger vendors in this index do not offer. It is scoped to platform usage limits rather than to the content or the targeting decisions.

This is the most complete technical stack disclosure in the index and it comes from one of the smallest vendors in it. The vendor's own published answers name, in sequence, exactly how the product works: content is ingested from nine platforms using Apify actors and a third party unified interface layer for LinkedIn, video and audio are transcribed by two named speech models from two different providers, and a named general model at a stated version classifies every item across six signal types.

The roadmap answers go further, naming breadth first graph traversal and vector similarity search as the mechanism behind a planned relationship mapping module. Provider, model and version are all present, and the surrounding infrastructure a buyer would otherwise have to guess at is named too.

The comparison that makes this notable is inside this index: a publicly listed vendor with tens of thousands of customers, graded the same week, describes its models only as artificial intelligence and machine learning and names nothing. Held at the top band on disclosure of what runs; the gaps that remain and should be re verified are inference location, what is sent to each named provider, and any retention or training terms with them.

One case study carries real weight: a named cybersecurity brand agency, a named co founder with a photograph, and four figures covering lists built, sequences run, reply rate and tools replaced. Ten client logos are displayed. Against that, every headline number is unsupported and one of them contradicts itself.

The 7 percent reply rate is claimed against a stated 2 percent industry average with no method, sample, period or definition of what counts as a reply, and neither figure is sourced. A claim of fourfold return on every 30 dollars of credits has no basis at all. No presence on any independent review platform was located.

Most concretely, the figure of more than 800 ideal customer lists appears once as a company wide statistic covering all clients and once as a result achieved by the single named case study client, and those two readings cannot both be true.

Recorded as observed and material to how the whole claim set reads: the vendor's own status badges mark two modules live and three as launching soon or dated to 2026 and 2027, while the published answers describe all of them in the present tense and the headline states the platform covers all four stages in one workflow today.

One channel is governed with a published number and the other two are not addressed at all. The professional network side names a specific daily ceiling, enforces it through a compliance queue, staggers sends and logs activity for audit, which is more operational detail than most vendors in this index publish about any channel. Cold email carries no named regulation: no commercial email statute, no consent standard, no unsubscribe mechanism and no suppression process.

The WhatsApp channel is the sharper gap, because business messaging on that platform is governed by its own policy requiring recipient opt in for most outbound categories, and the product markets it as a cold outreach channel for markets where it outperforms email with no consent position stated anywhere. Reaching a stranger on a personal messaging application is the most intrusive contact this platform enables and it is the one with the least published about it.

No privacy policy was located on the site or through a targeted search. The site footer carries three labelled columns covering product, competitor comparisons and company, and none of them is a legal section, so this is an absence rather than a page that failed to load. Set that against what the platform actually does with personal data.

It ingests posts, videos, reviews and conversations about and by named individuals from nine platforms including personal social networks and video services, transcribes their speech, classifies them by buying intent and stated pain points, builds contact lists, and holds mailbox, professional network and messaging application access on behalf of its customers.

For that activity nothing is published: no controller or processor position, no lawful basis, no retention period, no data subject request route, no contact address for a privacy enquiry, no international transfer position and no removal mechanism for the people whose content is being scored. The comparison inside this index is direct: the lowest band on this axis has previously been avoided by vendors that published at least a removal route, and none was found here. Re verify directly with the vendor before relying on this row, since a policy may exist behind the application login.

The collection mechanism is named more openly than most vendors manage, and the licensing position underneath it is absent. On the credit side, the vendor states that ingestion runs through a general purpose scraping service and a third party unified interface layer for the professional network, and names the nine platforms it draws from. That is a real answer to how the material is obtained and it is more than the usual reference to public sources.

What is missing is everything about the right to hold it: no lawful basis for processing the personal content of individuals who never contracted with anyone, no supplier or licence for the contact data behind the enrichment, no accuracy or match rate, no notification to the people whose posts and videos are transcribed and scored, and no removal route. The lead sourcing side is described only as mining eight sources in parallel, and those eight are nowhere enumerated. Two of the named platforms are review sites whose own terms govern reuse of their content, and no position is stated on either.

The vendor answers the compliance question directly and answers it with an unqualified yes, which is the finding. What is published in support is real and specific: a daily connection request ceiling of 100 per account enforced automatically through a compliance queue, sends staggered across correct day offsets, and an activity log the customer's own operations and compliance staff can audit from the dashboard. That audit trail is a control no other vendor in this category publishes.

What sits against it is the feature list on the same site. The product automates connection requests, direct messages, sponsored messages, profile views and post likes, pulls leads directly from the platform's own search product, and ingests that platform's content through a general purpose scraping service and a third party interface layer. None of that is permitted by the platform's user agreement at any rate, so a ceiling governs the volume of the exposure rather than removing it.

Rate limiting is not permission. Held at this band rather than lower because nothing here is marketed as evading detection and no account rotation is sold as a scaling mechanism, which are the framings that produce the bottom band elsewhere in this index.

The models are named, which is unusual and creditable, and nothing is said about what reaches them or what happens to it. Customer material, ingested third party social content and transcribed audio all pass to at least three external model providers, and no statement covers whether any of it trains anything, whether it is retained by those providers, or what terms govern the transfer.

The tenancy question is sharper here than at most vendors of this size because the product is explicitly sold to agencies running ten or more client accounts from one multi workspace platform, so a customer is asking not only whether their data reaches a model but whether it can surface in a competitor's workspace next door. The vendor states that each client workspace gets its own context and signal feeds, which is an architectural claim about separation but not a statement about the model layer sitting underneath it.

Two distinct problems sit on this axis and the second is new to the index. First, manufactured signals: the product automates profile views and post likes, and the entire communicative value of those acts is that a person chose to perform them, so an operator appears to have read and endorsed content nobody looked at. Opening lines are generated from the prospect's own recent activity, which is manufactured effort in its purest form.

Second, and broader than anything graded before: the listening layer ingests posts, videos, reviews and conversations from nine platforms including personal social networks and video services, transcribes the audio, and classifies named individuals by buying intent and stated pain points. A person discussing a problem on a public forum or in a video has no notice that their words are being transcribed and scored as a purchase signal.

A planned module extends this to mapping alumni connections, investor relationships and community overlap to find a route to a target. Cold contact reaches WhatsApp, a personal channel, and no consent position is published for it. No notice, no removal route and no position on the European transparency obligations of August 2026 were located. Held above the bottom band because nothing manufactures an identity: no synthetic voice, no invented persona, no assumed location, and no marketing of undetectability.

The dependency stack is named clearly, which counts for something: a scraping service, a unified interface layer for the professional network, two speech providers and a general model provider are all identified, so a buyer can see what the platform is built on. Beyond that the surface is thin and one absence stands out.

No connector to any system of record was located anywhere on the site, on a product that positions itself as a replacement for platforms whose own value rests substantially on writing back into the buyer's records. The unified inbox keeps replies inside the platform rather than syncing them out. No public interface documentation, no developer portal, no marketplace, no partner programme and no support for the emerging protocol that lets external agents query a vendor were found. A buyer replacing four tools with this one would be moving from a connected stack to an island.

Where the platform runs and where customer and prospect data are stored is not stated anywhere on the public surface. No region, data centre, cloud provider, hosting partner or residency commitment was located, and no residency option is offered.

The question has more weight here than the vendor's size would suggest, because the platform holds mailbox access, professional network sessions and messaging application access on behalf of its customers, and because the listening layer accumulates transcribed personal content about individuals who are not customers. The named third party providers in the stack each process that material in their own locations, and no statement addresses where any of them sit. The corporate entity behind the brand is named in the site footer and no address, jurisdiction or registration detail accompanies it.

The platform holds the widest set of customer credentials of any vendor graded in this index, and no security posture supporting them was located. Between them the product takes sending access to the customer's mailboxes, session access to their professional network account, access to a messaging application account, and custody of prospect and campaign data, and it does so on behalf of agencies operating ten or more client workspaces at once.

Against that: no certification, no audit report, no trust centre, no security page, no penetration test statement, no encryption description, no access control model, no status page and no vulnerability disclosure route were found on the site or through a targeted search, and the footer contains no legal or trust section of any kind.

Some of this is stage appropriate for a young company and the note records that honestly, but credential custody at this breadth is the one thing that does not wait for maturity, and a buyer handing over three account types has nothing published to evaluate. Re verify directly with the vendor.

Both purchase paths carry real numbers and the buyer is told plainly what governs the bill. Pay as you go credits start at 30 dollars with no monthly commitment, the managed tier is 500 dollars a month plus credits with a dedicated engineer, both tiers enumerate their contents, and the vendor states there are no locked in annual contracts. A credit calculator is offered so a buyer can size their own consumption.

Unusually, the vendor also publishes its competitors' prices alongside its own, naming four rival products with their monthly ranges and stating a typical combined stack cost, which is genuinely useful procurement information and something almost nobody in this index does.

Held off the top band on the piece that decides the real bill: the credit consumption rate per action is not published anywhere on the page, so the pay as you go path can only be costed through the vendor's own tool, and the managed tier is quoted as a fixed fee plus an open ended credit line. A buyer can see the entry price and cannot compute the running one.

Nothing governing the end of the relationship was located, and no governing terms exist to contain it. No terms of service, subscription agreement or customer contract was found on the site or through a targeted search, and the footer has no legal section.

Consequently there is no post termination retention period, no deletion timeline, no deletion confirmation, no statement of what happens to contact lists, sequence history, unified inbox conversations or the accumulated signal corpus once an account closes, and no stated export path for any of it. The standing distinction in this index puts an export feature claim without governing terms in the band above and neither in the bottom band, and neither is present here.

Two genuine positives are recorded and neither reaches the data question: the vendor states plainly that there are no locked in annual contracts, and the pay as you go model means a customer can simply stop buying credits. Walking away is easy; taking anything with you is undescribed.

Named controls exist across both sending surfaces, which is more than most vendors of this size manage. On email: multi mailbox rotation, inbox warmup, deliverability monitoring and spam score checking are each listed as built in rather than as add ons. On the professional network side a compliance queue enforces a stated daily ceiling per account and staggers sends across day offsets rather than firing them together, which is a real pacing mechanism described in operational terms.

Held off the top band because no standard is named behind any of it: no sender authentication protocols, no blocklist monitoring, no complaint or bounce threshold, no placement testing and no volume ramp schedule. Recorded rather than penalised, because context decides it: mailbox rotation appears here alongside warmup and spam scoring as sending infrastructure rather than being marketed as a way to exceed a provider's limits, which is the framing that has taken other vendors in this index down on this axis.

The segmentation is specific, falsifiable and written as four buyer situations rather than as a list of industries. Each names a role, quotes the problem in that role's own words, and states a claimed benefit: a founding seller or solo operator stitching four tools together before every campaign, a sales development squad lead whose team has no time for content, an agency running outreach for ten or more clients from separate stacks, and a founder before product market fit who cannot tell which market signal to act on.

A separate published answer restates the same three primary users. The agency case is backed by a real product decision rather than a claim, since multi workspace architecture with per client context and signal feeds is a shipped capability. Held off the top band because no headcount band, revenue band, geography or ceiling is published, so a buyer outside those four sketches has nothing to test themselves against, and the vendor never says where the fit ends.