In short: LinkedIn's terms forbid robots, extensions and scripts that send messages or add contacts on your behalf, and the platform says it looks for such tools. What works in 2026 comes from quality, not volume: short messages, reviewed by you, sent to a precise target. A tool saves time, it never makes you disappear.
What does "automating" mean on LinkedIn?
Three very different things hide behind the word. Preparing: building a list, drafting messages, tracking replies in a spreadsheet or a CRM. Executing: sending invitations and messages without clicking yourself. Simulating: liking, commenting or sharing automatically to suggest activity that is not there.
The three do not weigh the same. Preparation happens largely outside LinkedIn. Execution is covered by the site's terms. Simulation is what LinkedIn calls inauthentic engagement.
What exactly does LinkedIn say?
The User Agreement, version of 3 November 2025, forbids three things that concern tools. Using bots or other unauthorized automated methods to send messages, add contacts or post reactions. Using scripts, browser plug-ins or add-ons to scrape or copy the service. Bypassing usage limits, such as those on search results (LinkedIn, section 8.2, consulted on 7 October 2026).
The dedicated help page spells out the consequences: the account can be restricted or closed, and a forbidden tool may stop working without notice (LinkedIn Help, consulted on 7 October 2026). It does not list tools that are tolerated. No tool, ours included, can claim to be "authorized".
How does LinkedIn enforce these rules?
It restricts accounts whose activity looks like abuse. Among its reasons: many invitations in a short time, many invitations ignored, left pending or marked as spam, and suspected use of an automation tool. Most restrictions are lifted within a week (LinkedIn Help, consulted on 7 October 2026).
It also watches the browser. BleepingComputer reported on 3 April 2026 that a script on the site detects 6,236 extensions. LinkedIn says it looks for those that scrape data without members' consent or violate its terms (BleepingComputer, consulted on 7 October 2026). The details of that detection are not public.
What still works in 2026?
Messages addressed to one specific person. Expandi analyzed 13 million connection requests, from May 2025 to April 2026. It finds 28.5% acceptance and 10.4% replies to messages sent after acceptance, a rate that stayed flat (Expandi, consulted on 7 October 2026). Replies to connection notes, on the other hand, fell from 3.5% to 2.2%.
Expandi's later report adds three markers, on smaller samples of campaigns. Messages of 151 to 200 characters reply at 8.5%, against 6.8% on average. Three messages in total give 9.8% replies, against 5.0% from five onwards. And notes that cite an existing link, such as a follow or a shared group, far exceed the 30.2% average acceptance (Expandi, H2 2026, consulted on 7 October 2026).
These are figures from a tool vendor, on its own customers. They show a direction: short, targeted, few follow-ups.
What no longer works?
Volume without targeting, long sequences and generated text sent without review. The same source compares, within the same accounts, AI-written campaigns with hand-written templates: AI gets 12% less acceptance, with an identical reply rate.
The limit of these numbers: they describe one platform and one period. Your market may react differently, and only a test on your list will tell. Our article on acceptance and reply rates explains how to read your own results.
Which tasks should you keep under your control?
Four, whatever the tool. The target: who you contact, and who you rule out. The text: what you send, reviewed before it leaves. The replies: a refusal, a pricing question or a sensitive topic needs a person. The pace: you set the cap, and you step back at the first warning sign.
An example: someone answers "send me your prices". An automatic message that answers badly at that moment can lose the conversation. Sales Farmer's automatic reply is therefore optional and off by default, and the AI hands over on a refusal, a pricing question or a sensitive topic.
Where does Sales Farmer stand?
It is a Chrome extension that works in your browser, with your LinkedIn account. It falls under the rules above, like any tool of the same type. We promise neither invisibility nor the absence of restrictions.
What we offer: visible daily caps, time windows, delays between actions, and your approval. In Automatic mode, the AI proposes the people, the invitation goes out without a note, then the message cites a specific fact and asks a question; you approve. This mode is included in the Growth plan and in the 14-day trial. Manual mode, from the Starter plan, lets you write each step. Details: LinkedIn limits and limits and safety. For pacing rules, see our guide to automation limits.
FAQ
Is LinkedIn automation allowed?
No. The User Agreement forbids robots and extensions that send messages, add contacts or copy the service. The account can be restricted or closed.
Is a Chrome extension safer than a cloud tool?
Not by nature: the same rules apply to both. LinkedIn detects installed extensions, and a cloud tool uses your session from another location.
What can you automate outside LinkedIn?
Preparation: drafts, tracking replies in a spreadsheet or a CRM, reminders. These tasks do not touch the site. Anything that touches sending or extraction falls under the rules above.
Can AI write my messages?
It can propose a draft. Expandi's data shows that texts reviewed and rewritten by a human do better than those generated without review.
Keep control, save time
Try Sales Farmer for 14 days, no credit card: you set the caps, you approve the messages. Plans start at $15 a month.