The outbound vocabulary, agent era

LinkedIn's distance labels: 1st = connected to you, 2nd = connected to your connections, 3rd = beyond that. Each degree changes what outreach is possible and how warm it lands.

Software marketed as a replacement for a human sales development rep — autonomous prospecting, writing, and sending. The 2025 generation's flagship vendors became the category's cautionary tale.

Low-effort AI-generated content and outreach — template mail with a first-name merge, generic flattery, obviously machine-written notes. LinkedIn now lets users flag it with a dedicated report option.

The share of your LinkedIn connection requests that get accepted. It is the leading indicator of account restriction — reply rate is not.

Any limitation LinkedIn places on an account — from a temporary invite block to a full suspension requiring ID verification. The spectrum matters: most restrictions are recoverable warnings; repeat offenses aren't.

A team of specialized AI agents that hand outreach work to each other — sourcing, qualifying, researching, writing, pacing, replying — with a human holding the only send trigger.

Software designed to be operated by an AI agent through typed tools, rather than by a human through a dashboard — the interface is the protocol, not the UI.

A hard checkpoint in an automated flow where execution stops until a human decides — distinct from a notification, which informs without stopping anything.

A holding area where every agent-drafted message waits for a human yes before it is sent. Nothing auto-fires; rejected drafts are rewritten, and the rejections teach the system.

The measured share of managed LinkedIn accounts that receive a restriction over a defined window — properly stated as restrictions per managed seat per rolling 90 days, with the denominator disclosed.

LinkedIn's post-2024 enforcement approach: scoring how an account behaves — session origin, timing patterns, acceptance trends, browser-extension script injection — rather than just counting its actions.

An observable event suggesting a company or person may be in-market: hiring for relevant roles, raising funding, adopting adjacent tooling, or engaging with the problem space publicly.

The rhythm of an outreach sequence: how many touches, over how many days, on which channels. Cold LinkedIn cadences typically run 3–5 touches across two weeks.

The two architectures of LinkedIn automation: tools that run in managed cloud sessions versus tools that inject into your own browser. In 2026's enforcement climate, the distinction is the first safety question.

Contacting people who don't know you yet — on LinkedIn, by invite, message, or InMail. Still alive in 2026, but only the researched kind survives inbox saturation.

The multi-seat safety rule that two of your LinkedIn accounts must never approach the same company in the same week — coordinated touches from multiple seats are a visible, reportable pattern.

Your accumulated LinkedIn network as an asset: the connections, their connections, and the warm paths between you and anyone you'd want to reach.

The optional message (up to 300 characters) attached to a LinkedIn invite. The strongest notes reference something the person actually did; the weakest recite their job title back at them.

Running each LinkedIn account through its own stable cloud session and IP address, rather than shared pools — so the account's 'location' looks consistent and unshared.

The explicit registry of people and companies your outreach must never touch: customers, partners, competitors, investors, people who said stop, and anyone who'd make a screenshot expensive.

Using your own product to run your own operation — in outbound tooling, selling the tool with campaigns the tool itself sends.

Filling in the facts about a person or company that the original list didn't carry — role, seniority, funding, headcount growth, verified email — from data providers or research.

A consistent structured reply shape from every tool call — success flag, data, a one-line summary, notable signals, and durable learnings — so an agent always knows what happened and what deserves attention.

The planned messages after the first touch — classically an invite, an acceptance-gated opener, then two or three spaced follow-ups over 10–14 days.

A product's first committed users, recruited with preferential terms — early access, locked pricing — in exchange for feedback, tolerance, and honest usage data.

The role that treats go-to-market as an engineering problem: building sourcing pipelines, enrichment flows, and outbound systems from tools and agents instead of running plays by hand.

Open roles as market intelligence: a company hiring SDRs is investing in outbound; one hiring RevOps is systematizing; one hiring for a pain your product removes is a warm door.

System design where a person approves consequential actions before they execute — in outreach, a human yes before any message reaches a real inbox.

The written definition of who you sell to — role, company size, industry, geography, and the exclusions — precise enough that a list can be filtered against it mechanically.

LinkedIn's paid direct message to people outside your network, included with Premium and Sales Navigator plans. InMails don't consume your weekly invitation quota.

Third-party behavioral data suggesting an account is researching a category — content consumption, review-site activity, search patterns — sold by providers as buying-readiness scores.

Cancelling a sent connection request. LinkedIn blocks re-inviting the same person for up to three weeks after withdrawal — so it's cleanup, not a retry button.

A saved, deduplicated audience of people or companies managed together — the unit outbound actually operates on, distinct from a search result you haven't reviewed yet.

Ranking prospects by fit and timing so effort flows to the likeliest buyers — anything from a P1–P4 bucket system to a weighted model over ICP fit and signals.

The data files outbound tools produce (and LinkedIn itself offers for your connections) — the portable memory of who you know and who you've touched. The migration currency of the category.

A server that exposes LinkedIn capabilities — searching people, sending invites, managing an inbox — as typed tools an AI agent like Claude can call through the Model Context Protocol. The agent is the brain; the MCP server is the hands.

Operator slang for a temporary restriction — the account works for browsing but can't send invites or messages, typically for days to weeks.

Placeholders like {firstName} and {company} that a sequencer fills per recipient. The technology behind 'Hi {firstName}, I saw {company} is growing!' — and behind most of what gets reported as slop.

The open standard (launched November 2024) that lets AI agents call external systems through typed tools — the USB port between models like Claude and the software that does things.

Running outreach across several LinkedIn accounts as one coordinated operation — shared suppression, collision rules, per-seat pacing, and a unified inbox.

Coordinating touches across LinkedIn, email, and sometimes phone in one sequence — meeting the prospect where they actually answer.

Staying usefully present with people who aren't ready yet — engaging their posts, sharing something relevant occasionally — without pitching on a timer.

A Premium setting that lets anyone on LinkedIn message the person for free, without a connection or an InMail credit.

Connection requests you've sent that haven't been accepted or declined. A large pending pile is both a targeting report card and a quiet risk factor.

Making a message specific to its recipient. In 2026 the word covers two opposite things: merge-variable insertion (template personalization) and research-based writing (actual personalization).

The share of messages that get an interested response — not just any answer. The metric that actually correlates with pipeline, since 'no thanks' is also a reply.

The distinction dashboards blur: queued means the system plans to send; sent means it actually left. Reporting queued as sent is how tools fake momentum.

Varying the gaps between automated actions so activity doesn't tick like a metronome. Fixed intervals are the oldest automation signature there is.

The share of delivered messages that get any response. Useful for comparing copy angles; useless as a safety signal — acceptance rate is the number that predicts restrictions.

LinkedIn's paid prospecting product: advanced search filters, lead lists, alerts, and expanded network visibility, at roughly $99/month for Core.

One LinkedIn account under management — the operational and billing unit of outbound. A seat carries its own limits, ramp state, history, and risk.

Prospecting triggered by something that just happened — a job change, funding round, hiring spree, or a post the prospect wrote — instead of working a static list top to bottom.

LinkedIn's 0–100 score of how actively and credibly you use the platform, across four pillars: professional brand, finding the right people, engaging with insights, and building relationships.

The record of everyone any of your seats has ever contacted, used to prevent duplicate invites and cross-campaign collisions. Also called the contacted universe.

One place to read and answer LinkedIn conversations from every connected seat, with filters by seat, campaign, and tag — instead of logging in and out of accounts all day.

The one-sentence answer to 'why should this specific person care, now?' — the claim your outreach makes, which the rest of the message exists to make credible.

A route to a prospect through something you already share — a mutual connection, common employer or school, a customer in common, or a thread you both engaged in.

The practice of starting a new or dormant LinkedIn account at low outreach volume and increasing gradually — roughly 10 invites a day in week one, 15 in week two, 20+ from week three.

LinkedIn's rolling cap on connection requests — about 100 per week for most accounts, up to roughly 200 for older accounts with strong acceptance rates. In place since 2021.

Sending only during the target's local business hours, with human-shaped gaps — because a Mumbai-based account messaging New York prospects at 4 a.m. New York time is telling on itself.