Meet Alfred

An established multichannel automation platform with a model layer added on top rather than a model native product. Founded 2017 as one of the first LinkedIn automation tools; artificial intelligence personalisation and Customizable AI Settings both carry a New badge in the current navigation.

Remove the model layer and a complete product remains: LinkedIn, email and X sequencing, a relationship database, a smart inbox, a 600 plus template library, social post scheduling, team roles and seven export functions. The price card confirms the removal test independently, applying the convention refined on Linkyfy: the three tiers are denominated in campaign counts, channel access and Lead Finder result volume, and not one tier prices a model capability, so the ladder is telling you the same thing the removal test is.

Account level oversight is enumerated concretely in the roughly eighty row feature comparison: assign roles and permissions, create and assign groups, advanced team settings, manage multiple connected accounts, pause, resume, archive and delete campaigns, return a lead to a campaign, and automatic withdrawal of pending invitations with its own analytics.

A dedicated Customizable AI Settings page frames control over the model layer, and the multichannel page claims artificial intelligence that knows when to hand a conversation back to a human. The gap holding this at C is specific and sits beside a feature the vendor ships: the comparison table carries an Impersonate Accounts row allowing an administrator to operate a team member's connected account, and the same table carries no audit log row anywhere. Review of generated copy before it enters a live sequence is undescribed.

Four model surfaces are marketed: message personalisation, Customizable AI Settings, a Lead Finder described as using artificial intelligence to surface prospects, and an assistant said to judge when to return a conversation to a person. Across every product page, legal document and pricing surface read for this build, no model, provider, version, hosting region or processing arrangement is named. The infrastructure page names Amazon Web Services, Cloudflare and Stripe openly, so this is a vendor willing to name suppliers where it chooses to. The model supply chain is the one it does not.

The strongest single figure carries a stated basis, which is rare in this block: the agency page claims 3.1 times more clients managed per account executive, described as an average across agency customers. Beside it a testimonial gives a role and a named employer, Director of Business Development at MIM Agency, with a quantified move from four clients per executive to over twelve. Scale is published at 89,000 users across 87 countries.

Against that, the page metadata claims the platform is trusted by over 40,000 businesses including Coca-Cola, WeWork, Airbus, Barclays and Amazon, and those five names appear in a meta description with no case study, customer story or quote behind any of them.

Reply rates two to three times higher than single channel outreach, 63 percent of agency prospects requiring five or more touches, four to six hours per client per week, three hours saved every day and an average support response under three minutes all appear without a source. Independent review aggregation places the vendor at 3.3 of 5, among the lower scores for an established tool in this category. A reviews page exists and was left unread.

The acceptable use policy carries a genuine consent position, and a demanding one for a cold outreach product: every messaging list run on the platform must be confirmed opt in, and verification of the address or telephone number owner's express permission must remain available for the lifespan of that list.

The same clause prohibits lists or databases purchased from third parties for unconfirmed messaging purposes, and the policy is enforceable by suspension or termination without notice, with no service credit for the resulting interruption. The European regulation and the California act are both addressed on the security page and each has its own document. Two things hold this off the top band.

The vendor sells a Lead Finder returning up to 10,000 prospect records a period, which supplies exactly the unconfirmed contacts its own acceptable use policy restricts, so the compliance position and the product design point in opposite directions. And the policy states it was last reviewed on 1 November 2021, nearly five years before the model personalisation, Lead Finder and X channel it now governs. An unsubscribe mechanism, suppression list, complaint threshold and recipient removal route are all absent, and no electronic marketing statute is named by title.

The published set is the most complete in this alphabetical block: a privacy notice, a cookie policy, a standalone data processing agreement at its own address, a dedicated page for the European regulation, an acceptable use policy, a fair billing policy, a vulnerability policy and a security and compliance page, all linked from every page footer.

The notice names its lawful bases explicitly, consent, contract and legitimate interests, with the right to withdraw and the right to object stated, and commits that personal data is not shared with third parties for their direct marketing.

Handling of the connected mailbox is unusually specific: the two scopes used are enumerated, human access to message content is ruled out except on affirmative agreement, for abuse investigation, for legal compliance or for internal operations on aggregated and anonymised data, and adherence to the mailbox provider's limited use requirements is stated. The gap is the one recorded at LinkSprig and it is larger here: every data subject in the document is the customer.

The prospects whose profiles populate the relationship database, whose contact details the Lead Finder returns and whose replies are stored appear nowhere in it. The data processing agreement and the European regulation page were left unread and are flagged.

Lead Finder is priced as a headline allowance at 1,000, 5,000 and 10,000 results a period, and across the pricing page, the feature listing and every legal document read, no supplier, database, record count, coverage figure, refresh cadence, matching method or licence is named for any of it.

Two further flows are equally unaddressed: profile data pulled from the connected account into the relationship database, and the contact information exposed through the View and Export Connections Contact Information rows in the comparison table.

The sharpest point is internal: the acceptable use policy prohibits customers from using third party purchased databases for unconfirmed messaging, while the vendor sells its own prospect database into the same campaigns and states nothing about where those records came from.

This is the only vendor in the LinkedIn block graded so far to state a conformance position outright, and its own documents contradict it. The lead generation page asks whether the product complies with the platform's terms of service and answers yes, citing cloud execution and built in safeguards such as smart delays and daily outreach limits that align with the platform's policies.

The privacy notice opens with a different characterisation of the same product: the service imitates human behaviour. Human behaviour simulation then appears as a paid feature row in the pricing comparison table. A conformance claim asserted in a marketing answer, with no partner status, programmatic access or cited policy behind it, and contradicted by the vendor's own legal document, is a marketing position rather than a conformance position.

Held at C rather than lower because the framing throughout is staying within the platform's connection and messaging limits rather than past them, and nothing here rents, rotates or substitutes an identity.

The reservation the customer grants is the broadest in this block, and the exception carved out of it is the finding. The notice reserves the right to use and disclose anonymised, aggregated or derivative information for any purposes including without limitation the vendor's internal use and research.

One channel is then fenced off: information reaching the vendor through a connected mailbox is excluded from that reservation even in anonymised, aggregate or derivative form, and the notice attributes the exclusion to the mailbox provider's restricted scope requirements. The fence therefore exists where a platform policy put it, not where the vendor chose to put it.

Everything outside it sits inside the reservation: connection request and message copy, the relationship database, prospect records returned by Lead Finder, replies captured outside the connected mailbox, and the X channel. A training statement of any kind is absent.

Two published facts decide this band and both are the vendor's own. The privacy notice's opening description of the product is that the service imitates human behaviour, and Human behaviour simulation is a paid feature row in the pricing comparison table. Evasion sold as a priced line item in the price card, and stated as the product's defining characteristic in a legal document, is the Alsona band, reached here by a far larger and otherwise well documented vendor.

The recipient facing half is worse rather than better: the platform ships four categories of automated personal greeting, congratulating a person on a work anniversary, wishing them a happy birthday, congratulating them on a new role and thanking them for an endorsement, all sent under the buyer's own name with the model personalisation layer writing them. A recipient reading a birthday message from a real person's real account has nothing indicating a machine composed and sent it.

The obligation on artificial authorship now in force in Europe is addressed nowhere. Identity substitution is absent, which is the one thing separating this from the rental vendors earlier in the block, and it is not enough.

Destinations are real and named: the professional network, all email providers, Facebook, X, Instagram, a workflow automation connector and webhooks, with import of leads from an external relationship system and seven distinct export functions across campaigns, leads, connections, replies, team analytics, team campaign data and team relationship data. A maintained help centre sits at its own subdomain.

The gap is material for a sales tool nine years old and independent reviewers name it consistently: a native connector to any customer relationship platform is absent entirely, so every sync runs through the general purpose automation connector. Public developer documentation or an endpoint reference was not located, and the programmatic integrations row is gated above the entry tier.

A plain single jurisdiction residency statement, disclosed clearly with no options in it, which this index rates above silence: the security page states the platform is hosted at Amazon data centres and that those data centres are located in the United States.

The named infrastructure suppliers around it, Amazon Web Services for compute and storage, Cloudflare for denial of service protection and Stripe for payment processing, give a buyer a legible picture of where the service actually runs.

Held off the top band because the customer has no region choice, no transfer mechanism is named on any page read, and the corporate seat is reported in the United Arab Emirates while all processing happens in the United States, a combination the read documents never connect. The data processing agreement was left unread and would most likely carry the transfer clauses.

A specific self described control set covering thirteen areas, with a separate vulnerability disclosure policy beside it, which is more than most vendors this size publish. Enumerated: cloud only infrastructure with no self hosted routers, load balancers or servers; transport encryption on all traffic; continuous patching through replaced images and continuous deployment; production access limited to named engineering staff with passwords expressly forbidden in favour of keys; database instances separated from application servers; denial of service protection; encryption at rest for all user data through the database provider; encrypted backups with restore testing; card processing outsourced to a certified level one provider so no payment data is held; and deletion of all account data on request.

Held off the top band on attestation rather than substance: a service organisation control report, an international standard, a penetration test summary, a sub processor list and a trust centre are all missing, so every control above is the vendor's own account of itself with nothing independent behind it.

One disclosure deserves attention rather than a penalty, in the same spirit as the LinkSprig finding: all user data including passwords is described as encrypted rather than hashed, which is what a service that must replay a credential into another platform actually requires, and saying so is more than silent competitors do.

Three tiers, three billing cadences, three currencies printed natively rather than switched by location, a roughly eighty row feature comparison, published Lead Finder allowances at each tier, a free plan, a free trial, and no quote only ceiling anywhere. A buyer can budget their own purchase, which is the test this axis applies, and almost nothing in this alphabetical block publishes as much. What keeps it off the top band is that the card cannot be read consistently.

The Team tier is cheaper than the Pro tier at every single cadence, 79 against 99 monthly, 59 against 79 quarterly and 39 against 49 annually, while being sold as Pro plan features plus more. The entry tier changes contents as well as price with the cadence: billed annually it includes Lead Finder, the advanced relationship database and data export, and billed monthly it includes none of the three.

Volume discounts sit under the Team tier on two cadences and under a ten user threshold on the third. Independent reviewers report a three user minimum on the Team tier and a real monthly cost several times the headline once prospect data, verification and mailbox warming are added, and the vendor's page states no seat minimum at all.

Export is a product capability here rather than a promise, enumerated row by row in the comparison table across seven data types: campaign data, relationship database data, leads, connections contact information, replies including bulk selection, team analytics and team campaign data.

Deletion is committed twice in two separate documents: the notice states that once an account is closed all data is deleted from the vendor's systems, and the security page states that all data associated with an account will be deleted on request. That is a better position than most of this index holds.

Held off the top band on three points: a deletion timeline and retrieval window are stated nowhere, no post termination right survives in any read document, and the export rows are tier gated, so the entry tier on monthly billing carries no export function at all while the same tier on annual billing does. The buyer who most needs their data out, the one on the cheapest monthly commitment, is the one who cannot get it.

The mechanisms on the professional network side are real and unusual for this block: vendor hosted execution with a unique dedicated address for every connected account rather than a shared pool, smart delays, stated daily outreach limits, human paced behaviour, and automatic withdrawal of pending invitations with its own analytics view, which addresses a genuine account health trigger almost nobody else publishes.

The email side is where a buyer should look closely, because the product sells email as a first class channel and the sending discipline did not come with it. Warming, inbox placement testing, bounce categorisation, complaint thresholds, suppression and mailbox authentication guidance were located on no page read, and independent reviewers state plainly that warmup and placement testing are undocumented.

The daily outreach limits the vendor relies on for its platform compliance claim carry no published number on any channel or any tier, so the single governor holding the whole safety argument together is unquantified.

Buyer segments are named individually and each has real surface behind it: individual founders and prospectors, sales teams, agencies with a dedicated page carrying its own economics, and recruiters. The tier structure encodes the boundaries numerically, with white label unlocking at five or more users and a dedicated account manager, migration, onboarding and priority support at ten or more, so a buyer can see which rung they land on.

Coverage is quantified at 89,000 users across 87 countries. Held off the top band because that coverage claim has no breakdown of any kind, by region, industry or company size; the enterprise names asserted alongside it sit in page metadata rather than on any customer page; and the vendor never states the boundary its own economics create, which is that per seat pricing on a product where one seat maps to one connected identity makes a multi account operator pay a seat for every identity they run.