Skylead

The vendor's own packaging settles this: both model capabilities are add ons on every tier.

Data enrichment and the agent are each marked as an addition to the base plan rather than included in it, on all three published tiers. A vendor that sells its model layer separately is stating that the product is complete without it, and that is the correct reading here.

What the base plan contains is deterministic and substantial: conditional sequences branching on lead behaviour, network automation covering connection requests, messages, direct messages, profile views and follows, unlimited email accounts sending a hundred thousand messages monthly, warm up on both channels, address discovery and verification, text spinning and templating syntax, personalised images and animations, a suppression list, split testing and a consolidated inbox. A customer buying only that has a complete outreach platform.

The model layer is genuinely capable where it exists. The agent handles inbox conversations and works toward booked meetings, trained on the customer's product knowledge and conversation strategy. Enrichment adds context and writes opening lines, priced against a well known competitor on cost.

Both arrived in 2026, in April and July, so this is a deterministic platform that has recently bolted on inference rather than a product built around it.

Ask what the enrichment and agent add ons cost per seat.

The agent ships with two modes and the vendor frames the choice as how much control you want.

That is the right way to build it. Automatic mode lets the agent handle conversations and steer prospects toward meetings unattended. Semi automatic keeps a person in the path. Offering both, naming the trade off plainly in the product announcement, and letting the operator choose is better design than either extreme, and better than the several records this session that ship autonomy with no described checkpoint at all.

The agent is also trained on the customer's own material, their product knowledge, conversation strategy, goals and examples, so what it says is grounded in what the customer told it rather than in a general model's guess.

Around it sit real controls. Conditional sequences are built by the operator and inspectable before running. A global suppression list is included on every tier rather than gated upward. The agency arrangement adds granular client roles and permissions, and a performance overview with risk alerts.

What is unspecified: what constrains the agent in automatic mode, whether it can commit the customer to anything, what a risk alert actually detects, and whether any audit trail records what was sent.

Ask what the agent is prevented from saying in automatic mode.

An agent holds conversations with prospects and nothing is named behind it.

Two passes across the home page, the pricing page, the feature comparison tables and the product announcements located no model provider, no model name, no version, no statement of what customer or prospect content is transmitted for inference, no training or retention position and no evaluation or accuracy measure.

The agent is where this matters most. It reads inbound replies from real people, composes responses, and in automatic mode sends them without review while working toward a booked meeting. The customer trains it on their own product knowledge and conversation strategy, which means proprietary commercial material is being transmitted somewhere unstated and retained under terms nobody has published.

Enrichment carries the same gap on a lower stakes surface, generating opening lines from context gathered about a named individual.

The cost comparison sharpens the question rather than answering it. Positioning enrichment as ninety five percent cheaper than a named competitor is a claim about unit economics that depends entirely on what model is doing the work and how, and the vendor publishes the saving without publishing the mechanism.

Ask which provider runs the agent and whether training material is retained.

Four named customers with figures attached, and the figures are the kind a buyer can reason about.

A founder reports consistently generating between three and ten thousand dollars monthly in new recurring revenue through the platform. Another states forty five percent of his business came through it. An agency managing director reports generating a hundred thousand dollars for a client. A fourth went from booking twenty calls a week to eighty. Each carries a full name, a role and a named company, and each links to a dedicated story page rather than sitting as a floating quotation.

The named logo set includes a publicly listed email marketing company, which is a checkable reference rather than an unknown logo.

Scale is stated as more than ten thousand users across ninety three countries with the product localised into seven languages, and the language count is corroborated by the feature table rather than asserted alone.

What holds this below the top band: every figure is a vendor published case study with no independent verification, no measurement period and no baseline. The headline claims of three times more meetings and eleven hours saved weekly carry no methodology. And the vendor's own site displays no review count or rating from any independent platform.

Ask over what period the forty five percent and hundred thousand dollar figures were measured.

A suppression list on every tier, and nothing published about the rules governing either channel.

The suppression list is the genuine control and it is not gated, appearing in the base plan rather than reserved for larger buyers as it is elsewhere in this batch. Combined with address verification before sending and warm up on both channels, the operational hygiene is real.

What two passes could not locate: an unsubscribe insertion or propagation mechanism for email, a consent basis for contacting anyone, any reference to data protection regulation, any reference to the bulk sender requirements the major mailbox providers introduced in 2024, and any calling or contact frequency position.

The volume makes those absences material. A hundred thousand messages a month per seat places every customer inside the bulk sender thresholds, and the platform serves more than ten thousand users across ninety three countries, which spans every consent regime that exists.

The network channel is worse by construction, since it has no unsubscribe mechanism at all. The only protection available is not contacting people who did not ask to be contacted, and an automated sequence builder is designed to do the opposite at scale.

Terms and a privacy policy are published and were not opened.

Ask how unsubscribes are inserted and propagated across both channels.

Two documents published, and the custody position is heavier than most.

Terms and conditions and a privacy policy are linked in the footer of every page. Neither was opened in this pass. Two passes located no data processing agreement, no subprocessor list, no retention schedule, no deletion commitment, no residency statement, no rights request process and no named privacy contact.

What is held deserves stating. Cloud based network automation means the platform holds authenticated access to each user's professional network account and operates it server side, across more than ten thousand users. Connected email accounts sit alongside. Then a contact database, enriched records assembled about named individuals, discovered and verified addresses, and the consolidated inbox holding conversations from both channels.

The white label arrangement adds a layer the privacy documents would need to address and a buyer should check. Under it, a partner's clients are using this platform without knowing the company exists, which makes the vendor a subprocessor to an agency's clients twice removed.

The geographic spread compounds it, with ninety three countries served and no residency position published anywhere.

Ask for a processing agreement and how network account credentials are stored.

A contact database arrived in June and nothing has been published about where it comes from.

The announcement describes finding and importing leads and companies with filters, reviewing contacts, and saving results into lists without leaving the platform. Two passes located no record count, no supplier, no collection method, no consent basis, no accuracy claim, no verification standard and no refresh cadence for any of it. For a capability shipped two months ago that is an omission a buyer should press on rather than a mature gap.

Enrichment sits beside it, adding context to leads and companies from sources that are equally unstated, and the cost comparison against a competitor implies a different sourcing model without describing it.

Address discovery finds business addresses by an undescribed method.

The older half of the product is cleaner. Eight lead source options include uploaded spreadsheets, which are the customer's own, and network searches, which are extraction rather than supply and are graded on the platform terms axis.

What is genuinely absent, and worth noting, is any claim about the database that could be contradicted. The vendor publishes no record total and no accuracy percentage, so there is no figure to test.

Ask where the contact database is sourced and how often it refreshes.

The vendor prints the disclaimer on every page, which is candid and is also the admission.

The line reads that the company is not affiliated or connected with the professional network in any way. It sits in the footer sitewide. That is more honest than vendors in this index who imply a partnership, and it states plainly that no permission exists for anything the product does on that platform.

What it does is extensive. Connection requests, messages, paid and free direct messages, profile views and follows, executed from cloud infrastructure rather than a browser, with a dedicated network address per user, plus a warm up feature that ramps activity gradually and a capability described as reaching open profiles through that address. Lead sources include searches run on the network and its sales navigator product.

Every element of that is prohibited by the platform's user agreement, and the architecture is built specifically to reduce the chance of detection: server side operation, per user addresses so accounts do not share fingerprints, and graduated activity ramping.

The white label programme extends the same exposure to partner brands whose clients may not know which company operates the automation.

Email runs through the customer's own accounts, which is the sanctioned path and carries none of this.

Ask what happens to a customer's campaigns when their network account is restricted.

An agent that converses unattended, and no stewardship documentation at all.

The semi automatic mode is the mitigation and it is credited on the oversight axis. Automatic mode is the exposure: the agent reads replies from real prospects, composes responses and sends them without review, working toward a booked meeting. Two passes located no accuracy figure, no evaluation, no error rate, no statement of what it may not say, no limit on what it can commit the customer to, and no described stop condition when a conversation goes wrong.

The agency arrangement scales that, since a trainable agent deployed across an agency's whole client base means one misconfiguration reaches every client's prospects at once.

On stewardship two passes located no encryption statement, no access control description, no logging, no retention schedule, no incident response or breach notification process, no named security contact and no vulnerability disclosure route.

The custody makes that consequential rather than procedural. Cloud based operation means the platform holds working access to more than ten thousand professional network accounts, each carrying that person's full network, private correspondence and professional identity, alongside connected email accounts and the consolidated inbox.

Ask what constrains the agent in automatic mode and how account access is secured.

Several features exist specifically so the recipient concludes something untrue.

Identity itself is genuine. Actions run under the customer's own network account and their own email addresses, with no persona, no rented profile and no synthetic sender.

What the product layers on top is the problem. Profile views and follows appear as sequence steps, and their entire function is to surface in the target's notifications so a named human appears to have taken an interest, before the connection request arrives. Text spinning varies message wording across recipients, which exists to prevent identical messages being recognised as bulk. Personalised images and animations insert a recipient's details into a graphic so it reads as individually prepared. And in automatic mode the agent conducts a conversation the prospect believes they are having with a person.

Two passes located no disclosure of automation on any channel, no indication that a message was machine composed, and no notice that the profile visit was scheduled rather than chosen.

The network channel offers the recipient no exit either, since it carries no unsubscribe.

Each feature is ordinary for the category. Together they describe a system engineered around a recipient's mistaken inference.

Ask whether prospects conversing with the agent are told it is automated.

A documented public interface, webhooks, and real time two way synchronisation with three named record systems.

That combination is rare anywhere in this index and unheard of in this category. The interface documentation is published openly as a linked collection rather than gated behind a sales conversation, so a developer can read the endpoints before buying. Webhooks push events outward in real time. And the record system synchronisation is named specifically rather than described generically, covering three of the most widely deployed platforms, and stated as real time and bidirectional rather than a nightly export.

That last point is what raises this to the top band. Most outreach tools in this sweep are islands, generating replies and outcomes that a customer must transcribe by hand into the system where their pipeline actually lives. Here the outcome lands where the revenue team already works, automatically, which is the difference between a tool and part of a stack.

Around it sits a named product partnership for video personalisation, an affiliate programme, an agency programme and a white label programme each with its own terms, and a dated product changelog published on the home page.

The education estate is substantial too: a masterclass platform, a community, a glossary, an ebook and a video channel.

Ask what the interface rate limits are and which objects synchronise bidirectionally.

The architecture is described precisely and the location is not stated at all.

Cloud based operation is named repeatedly and it is architecturally meaningful rather than marketing filler. Network automation runs on the vendor's infrastructure rather than through a browser extension on the user's machine, with a dedicated network address allocated per user, which the feature table refers to directly. That means the platform maintains persistent authenticated sessions and acts continuously on each customer's behalf, and it is the standard architecture for this category because it separates users from one another.

What two passes could not locate: a hosting provider, a region, a data centre, a residency commitment, a tenancy or isolation model beyond the per user address, an encryption statement, a backup position, a continuity plan, an uptime commitment or a status page.

The geographic reach makes residency a live question rather than a formality. More than ten thousand users across ninety three countries, with the product localised into seven languages, means substantial European usage, and no processing location is published anywhere.

A second consequence follows from the architecture: the address a customer's account operates from is chosen by the vendor, and a network account suddenly operating from an unexpected country is a well known trigger for platform security review.

Ask in which region each user's dedicated address is allocated.

Two legal documents, and nothing about security anywhere.

Terms and conditions and a privacy policy are published in the footer. Two passes located no certification of any kind, no audit report, no international standard, no penetration test, no trust portal, no dedicated security page, no encryption or access control statement, no subprocessor list, no vulnerability disclosure route and no named security contact.

The scale is what makes this the sharpest structural gap in an otherwise strong record. This platform holds working authenticated access to more than ten thousand professional network accounts across ninety three countries, operating them server side. Each of those accounts carries a person's entire professional network, their private message history and their identity. A compromise would expose all of it simultaneously, and there is no published position on how any of it is protected.

The agency and white label arrangements compound it in the usual way. An agency connecting fifty client seats has accepted custody terms on behalf of clients who never saw them, and a white label partner's clients do not know which company holds their credentials.

A vendor at this level of product maturity, with a public interface and named enterprise record system integrations, would ordinarily have been asked for an audit by now.

Ask whether any independent security assessment exists.

The agency pricing is published in a form an agency can actually model, and the two newest features have no price at all.

The good half first. A single seat is a hundred dollars monthly with a full inclusion list. The agency arrangement publishes two figures rather than hiding behind a conversation: nine hundred and ninety nine dollars for fifty seats, or one thousand nine hundred and ninety nine for unlimited seats. Naming both lets an agency calculate its own margin per client and work out where the unlimited tier becomes the better buy, which is precisely the calculation vendors usually force into a sales call. Annual is stated as paying ten months for twelve. A seven day trial is offered.

A full feature comparison runs across the three tiers, with a second table covering the white label arrangement separately.

The deduction concerns the two capabilities the vendor is currently promoting hardest. Data enrichment and the agent are both marked as add ons on every tier, and no add on price appears anywhere. Those launched in April and July of this year, they lead the product navigation, and a buyer reading the pricing page cannot establish what the advertised product actually costs.

White label perks are all recorded as per agreement, and the annual tier requires a conversation.

Ask what the enrichment and agent add ons cost per seat per month.

Real time synchronisation to systems the customer already owns means much of the value never lives only here.

That is the substantive point. Campaign outcomes, replies and lead status synchronise in real time and bidirectionally with three named record systems, so for a customer running any of them the record of what happened accumulates in a platform they control rather than in this one. A customer leaving does not need an export for the part that matters most, because it already left.

Beside that sits a documented public interface and webhooks, both giving programmatic routes for extracting or streaming out anything else. Very few records in this index offer either, and this one publishes the interface documentation openly.

What two passes could not locate: an export scope statement, a format, a deletion commitment, a retention position after cancellation or an account closure process.

Two things would not travel. Sequence designs, which are the accumulated operational knowledge of what works, and the network conversations, which live in the customer's own network account rather than here, so they survive by accident of storage rather than by commitment.

The white label case is different and worse, since a partner's clients would be exiting a platform they cannot name.

Ask what an export contains and what is deleted on cancellation.

Warm up on both channels, verification before sending, and a published ceiling.

The email side covers the recognised mechanics. Warm up runs unlimited across every connected account, delivered through a named third party service rather than built in house, which is worth noting because it means the warm up network is somebody else's and its behaviour is not this vendor's to describe. Address discovery is paired with verification so invalid addresses are caught before they bounce, which is the single most effective protection for sender standing. Unlimited connected accounts spread volume, scheduling controls timing, and a monthly ceiling of a hundred thousand messages is published rather than left open.

The network side has its own warm up, ramping activity gradually rather than starting at full rate, which is the equivalent protection for that channel.

A global suppression list, split testing and a published guide to warming a new domain sit alongside.

What is missing: sender authentication guidance, per account daily ceilings, bounce handling policy, placement testing, blacklist monitoring and any measured deliverability figure.

One inconsistency worth recording for a future reader: the warm up page address references one provider name while the page text names a different one, which suggests a supplier change that has not fully propagated.

Ask which warm up provider is current and what the per account daily ceiling is.

Four documented use cases, ninety three countries, seven languages and three partnership models.

The use cases are not a repositioned feature list. Sales, marketing, recruiting and agency each carry a dedicated page and each is paired with a named customer story from that segment, so a prospect can read how somebody in their own position used it. Recruiting is a genuine inclusion rather than an afterthought, since network automation serves candidate outreach with identical mechanics, and the vendor treats it as a first class motion.

The geographic claim is backed rather than asserted. More than ten thousand users across ninety three countries, with the product localised into seven languages, and the language count appears in the feature comparison table as a shipped capability rather than only in marketing copy. Actual product localisation is a material investment and very few vendors at this size make it.

The commercial ladder covers every shape of buyer: a single seat, an agency at fifty seats, an agency at unlimited seats, an annual arrangement for steady headcount, and a white label programme for partners who want to sell it as their own. Three separate partnership tracks operate alongside.

What is absent: no industry vertical and no company size band, so segmentation runs by motion and by seat count rather than by who the customer is.

Ask which seven languages are supported and which regions the contact database covers.