For example, if you type in Claude, 'Help me write a LinkedIn post; the topic is this week's customer retrospective,' LinkPost will generate 25 versions from different angles in the background, score each one, and push the highest-scoring one back into the chat box (LinkPost MCP[1]). The one you adopt does not stop in the chat: it goes into the calendar, is sent through the official LinkedIn API, and after it is posted, you come back to check the data.
The MCP page describes exactly this workflow. MCP is a protocol that lets AI assistants call external tools; users add LinkPost once as an MCP server in the assistant settings, after which they do not need to open the web app, and it supports five assistants: Claude, ChatGPT, Perplexity, Gemini, and Grok (same as above[1]).
'The AI will write posts' space is already crowded. What is worth looking at with LinkPost is the extra step it takes after generation: generating 25 variants, scoring them one by one, and returning only the highest-scoring one. Scoring and ranking are the selling points it puts up front: in the feature list on the pricing page, 'predict viral score' comes right after 'unlimited AI generation' (pricing page[2]).
Image: LinkPost official display[3], accessed 2026-11-10, original interface preserved.
The score does not bet on a viral hit; it measures against the relative average
The pricing page FAQ says that the V2 score analyzes more than 300 factors, 27 of which are personal factors learned from the user's own writing style; behind it are multiple models (regressor, ranker, calibration), several sets of dedicated algorithms, and a self-developed scoring engine, trained and validated on 50,000+ posts and 3500 creators (pricing page[2]).
The site also gives a set of experimental numbers: on 49,908 posts, it defines a 'breakthrough' as likes reaching 1.5 times the person's average or more, and a 'failure' as below 0.5 times that average; after selection by score, the site claims a 71% increase in breakthrough posts and a 75% decrease in failure posts (same as above[2]).
Both definitions hang on the user's own average. Under this experimental definition, the score is not aimed at 'going viral'; it is aimed at not falling too far below one's own average. For people who need to put out a fixed few posts every week, the former is like a lottery; the latter is what they face every day.
This score is about what happens before publishing. The MCP page says what the assistant gets is scoped access, but it does not spell out read/write permissions, whether data is stored locally or in the cloud, or how to revoke the connection; the page also does not say who retries a failed post or who maintains LinkedIn authorization when it expires. These are not in the selling points, yet they determine how cleanly it can be wrapped up on the day something goes wrong.
Before buying, there is another condition: the full product has no free trial; the free entry point offered on the website is LinkedIn Post Analyzer, for trying out the scoring logic (LinkPost MCP[1]).
He lets the algorithm pick first, and the person decides whether to use it
LinkPost is made by Yannis Haismann; he and Matteo Kocken built the Link* series together: LinkPost, LinkHub, LinkMagnet, LinkEarn (founder page[4]). In the founder page's self-introduction, he describes LinkHub as helping you leave the right comments for the right people at the right time, rather than automating actions for you.
That self-imposed constraint belongs to LinkHub and cannot be taken directly as LinkPost's product boundary. What the LinkPost MCP page can confirm is another matter: AI generates 25 variants from different angles in the background and scores them one by one, returning the highest-scoring one to the chat; the person decides whether to adopt this version, and only after adoption does it enter the calendar, get published through the official LinkedIn API, and have its analytics checked. Whether it is posted is up to the person, and posting is still completed inside LinkPost.
He is also in the habit of laying out the accounts. His personal site lists 13 SaaS products over five years and cumulative 196,009€, saying this money mainly came from two of the products; the remaining failed and low-income projects are also listed, including five projects with extremely low revenue or that never launched—among them, SvgAnimator was sold only once, for 9€ (yannishaismann.com[5]). The same page says that for many years before that he had no real success, at one point burned out and moved back to his mother's home, and 12 months later reached 10,000€ MRR. These are all statements he put on public pages himself; they show he is willing to put failed projects out in the open too, but they do not explain the scoring algorithm.
There is a public case for the customer acquisition step, but the definition needs to be read carefully. Lumail's customer case study says he made email the main marketing channel for LinkHub and LinkPost: since early 2026, he has been running four products in one Lumail organization, segmented with language, product, purchase, and milestone tags; the app automatically creates contacts through the subscribers API, starts onboarding by product tag, and paying customers have a separate post-purchase flow. The case gives an average tracked open rate across all campaigns of 40%, and notes that tracked opens include automatic opens from Apple Mail Privacy Protection, so they can only be viewed as a trend, not as actual human reads (Lumail case study[6]). These data cover four products and contain no open, click, and paid attribution for LinkPost alone, so they can show how this company talks to users, but cannot show how many orders LinkPost sold through email.
90 subscriptions and one listing
Prices need to be looked at separately because LinkPost currently carries two sets of prices. The Black Friday page has Standard at about $44.30 (€39.50)/month and the Signature Pack at about $274.74 (€245)/year; the latter, in addition to all features, includes a 1-hour strategy call, video course, early access to new features, and 24-hour priority support. The page says the discounted price is locked in for life as long as the subscription remains active; the Standard tier is marked 'cancel anytime', and the promotional price is held until Sunday midnight Paris time (pricing page[2]). TrustMRR's Pricing field, by contrast, lists Standard at about $88.59 (EUR79)/month or about $549.49 (EUR490)/year, and also lists USD79/month or USD490/year, noting that legacy and promotional prices will differ (TrustMRR[7]). One is a limited-time promotion, the other is a list price; reporting them mixed together distorts the picture.
Using MCP as a common condition makes it possible to find a comparable product. Postiz's pricing page says Standard is $29/month for 5 channels, ranging to Ultimate at $99/month for 100, and all plans include MCP server, CLI, and API, with a 7-day free trial and support for multiple social networks (Postiz Pricing[8]). The difference is not 'whether you can post from chat' but what the two companies put in front of the product: Postiz puts cross-platform scheduling first, while LinkPost puts LinkedIn multi-version scoring first; the latter only does LinkedIn and also has no trial for the full product.
On the money side, the Stripe metrics on the TrustMRR page cover only LinkPost: MRR is $4,877, with 90 active subscriptions and historical total revenue of $181,587, and data updated to 2026-11-14 06:00 (TrustMRR[7]). MRR is subscription revenue, not profit; full costs and P&L require signing an NDA, and the refund rate and failed-charge rate are also not disclosed.
There is also an item on the same page: LinkEmpire is for sale as a whole, asking $79,000, including three LinkedIn SaaS products (LinkPost, LinkHub, LinkMagnet), the LinkEarn course, and shared distribution assets; combined revenue between November 2025 and October 2026 was $152,912, and the founder's note says offers are being reviewed until November 14 (same as above[7]). In the upper left corner of the page, an original price of $99,000 is crossed out, and it is marked '1.3x revenue'; the page does not explain the denominator of this multiple. The full accounts are behind an NDA, and outsiders cannot tell whether this price is high or low.
What LinkPost is really selling is ranking: the algorithm picks the highest-scoring one from 25 variants in the background, and what the person keeps is the last move: whether to use it. As for whether this ranking is accurate, what this article can find is the vendor's own experimental definition and results; independent verification has not yet been seen.
[1] LinkPost MCP: write LinkedIn posts from ChatGPT
[2] LinkPost
[4] Yannis Haismann — Founder of LinkHub
[5] Je ship des SaaS devant tes yeux.Et je te partage tout, chaque semaine.
[6] How Yannis Haismann made email the main channel for LinkHub and LinkPost