Autopilot, Semi-pilot and Copilot are the three operating modes of an AI SDR on LinkedIn. Each mode defines who makes the decision to send a message: the AI alone, the AI with human approval before sending, or the AI that pauses and waits once a reply arrives. The right choice depends on your ICP, your average deal size, and how much control you want over each conversation.
What this post covers:
- What each mode does in practice — and what the AI decides independently in each one
- When Autopilot makes sense and when it can cost you a warm lead
- Why Semi-pilot exists and what problem it solves that Copilot does not
- How to choose the right mode for your B2B LinkedIn sales cycle
What Are Autopilot, Semi-pilot and Copilot in an AI SDR?
The three modes describe the degree of AI autonomy at each stage of a LinkedIn conversation. In Autopilot, the AI sends every message in the sequence without human approval. In Semi-pilot, the AI sends until the lead replies — after that, a human approves each response before it goes out. In Copilot, the AI drafts every message, but a human approves before sending, throughout the entire conversation.
These modes are not abstract product categories. They are decisions about where the human enters the process. And that decision has direct consequences for reply rates, conversation quality, and the volume you can realistically operate.
Precise definitions:
- Autopilot: the AI reads the lead's profile, writes a contextualised message and sends it. No human review. The sequence runs from the first connection invite to the scheduling link without interruption.
- Semi-pilot: the AI operates as it does in Autopilot until the lead replies. When a reply arrives, the conversation enters a human review queue. The human reads it, edits if needed, and approves the outgoing response.
- Copilot: the AI drafts each message before it is sent, regardless of whether the lead has replied. The human approves before every action is taken.
For a broader view of how AI SDR tools compare on these dimensions, see 7 Best AI SDR Tools for LinkedIn in 2026: Features, Safety & Pricing Compared.
What the AI Does Automatically Across All Three Modes
Regardless of the mode selected, several actions are automatic in all three: sending connection invites, running sequence campaigns, and withdrawing pending invites after 15 days without acceptance. The scheduling link is also included in the conversation — in Autopilot, the AI sends it; in Semi-pilot and Copilot, it goes out with the user's approval.
What changes between modes is exclusively control over the text messages the AI composes in response to conversation context. In Autopilot, the AI decides and sends. In Semi-pilot, the AI decides after a reply arrives, but a human approves. In Copilot, the human approves everything, every time.
Default operating caps per LinkedIn account (fixed): 25 invites per day, 40 conversations and 200 actions per day, 200 invites per week, and 80 profile visits per day. With Sales Navigator, the limits shift: 40 invites, 60 messages, and 100 visits per day. Sending happens only during business hours, and new accounts go through a 15-day warm-up period.
When to Use Autopilot — and When It Becomes a Liability
Autopilot is the right mode when volume is high, the ICP is tightly defined, and the early messages in the sequence do not require personalisation beyond what the AI already extracts from the prospect's profile. It allows a single operator to run hundreds of simultaneous conversations without manually reviewing each one.
Situations where Autopilot performs well:
- Broad outreach to a well-defined ICP where the first two or three messages are largely templated
- Campaigns with validated copy where reply rate is already established and the main goal is throughput
- Teams where the SDR's time is better spent on discovery calls than on message review
- Founders running lean operations who want prospecting to run without daily manual input
Autopilot's limitation is what happens after a lead replies with something specific. When a prospect asks a detailed product question, raises a technical objection, or provides context the AI cannot map to a standard response, an automated reply risks sounding generic at precisely the moment the conversation had become valuable.
HubSpot's State of Marketing research consistently shows that personalisation at the response stage — not just the initial outreach — is what separates conversations that convert from those that go cold. Autopilot handles the top of that funnel well. It is less suited to the middle, where nuance matters.
A practical rule: if your average deal size is high enough that losing one warm lead is meaningful, Autopilot alone is not your mode for the full sequence.
Semi-pilot: The Mode That Solves the Handoff Problem
Semi-pilot is the mode most teams eventually land on once they have run both Autopilot and Copilot and identified where each breaks down. The logic is straightforward: the AI is good at generating and maintaining outreach at scale; humans are better at reading intent once a reply arrives.
Semi-pilot splits the work accordingly. The AI handles the full outbound sequence — connection invite, follow-up messages, timing — until the moment a lead responds. At that point, the conversation pauses and enters a review queue. The human reads the reply, sees the AI's suggested response, edits if needed, and approves.
This matters for two reasons that Copilot does not address:
-
Volume before the reply: in Copilot, every message — including the first connection request and each follow-up in a cold sequence — requires human approval. For teams running outreach at scale, this creates a bottleneck that eliminates most of the efficiency gain from using AI at all.
-
Quality after the reply: in Autopilot, every reply is handled by the AI, including the ones where a human touch would materially improve the outcome. Semi-pilot preserves AI efficiency for cold outreach and human judgment for warm conversations.
For consultative B2B sales with long cycles — where a reply from a VP of Engineering or a CFO is not a commodity event — Semi-pilot is frequently the more defensible choice. The cost of a poor automated response to a high-intent reply is higher than the cost of a 10-minute review queue.
Copilot: Full Control Without Full Manual Operation
Copilot is the mode where human oversight is highest. Every message the AI drafts — whether it is a cold opener, a follow-up, or a reply to an inbound response — goes into a review queue before it is sent. The human reads, edits if necessary, and approves.
This is not the same as writing every message from scratch. The AI still reads the prospect's profile, pulls relevant context, and generates a draft. The human is reviewing and approving, not composing from a blank page. The time cost is lower than fully manual outreach, but higher than Semi-pilot for cold sequences.
Copilot is well suited to:
- High-ticket accounts where every touchpoint carries reputational weight and a poorly worded message can close a door permanently
- Account-based marketing approaches where the prospect list is small and each account requires a tailored narrative — see ABM on LinkedIn: How to Land Strategic Accounts in 2026 for how those sequences are typically structured
- New operators learning the tool, who want to see how the AI frames messages before allowing autonomous sending
- Sensitive verticals where compliance or regulatory context means every outbound message should be reviewed before it reaches a prospect
The tradeoff is throughput. Copilot is not the mode for campaigns targeting hundreds of new prospects per week with a single operator. The review queue becomes the bottleneck. Teams that need both coverage and control tend to use Copilot for a targeted list of named accounts while running Autopilot or Semi-pilot for broader outreach.
Choosing the Right Mode for Your Sales Cycle
The question is not which mode is objectively best. It is which mode matches your current operation. Three variables drive that decision:
1. ICP specificity and ticket size
A tightly defined ICP with a mid-market ticket and predictable objections is a good fit for Autopilot. A short list of enterprise accounts where each conversation is a significant revenue event points toward Copilot or Semi-pilot.
2. Sequence stage
Cold outreach — the first two or three messages before any reply — is where the AI adds the most value with the least risk. This is the case for Autopilot and the pre-reply phase of Semi-pilot. Post-reply conversations, where intent signals become meaningful, are where human review earns its cost.
3. Team capacity for review
Copilot requires someone to check the queue regularly. Semi-pilot requires availability when replies come in, which can be less predictable. Autopilot requires neither, which is its core operational advantage — and its core risk.
A common pattern: teams start with Copilot to understand how the AI behaves, move to Semi-pilot once they trust the cold sequence copy, and use Autopilot for high-volume campaigns targeting ICPs with well-tested messaging.
Mode Selection by Scenario
The following scenarios illustrate how the choice plays out in practice:
Scenario 1 — SaaS founder running outbound solo One operator, 200 to 300 prospects per week, well-defined ICP in a specific vertical. The goal is pipeline volume without hiring. Autopilot runs the full sequence. The founder reviews conversations only when a prospect replies and books time on the calendar.
Scenario 2 — Consulting firm targeting C-suite A boutique with 40 named target accounts per quarter. Each account has a specific entry point (a trigger event, a recent hire, a published report). Copilot lets the partner review every message before it goes out, ensuring the framing matches the account context. Volume is low; each message matters.
Scenario 3 — SDR team scaling outreach Three SDRs each managing multiple LinkedIn accounts. Cold sequences run on Autopilot. When a lead replies, the conversation shifts to Semi-pilot: the SDR reviews the AI's suggested response, edits for tone or specifics, and approves. The AI handles scale; the SDR handles intent.
Scenario 4 — New operator validating copy A team launching its first AI SDR campaign starts with Copilot to review how the AI constructs messages for their ICP. After two weeks and enough data to trust the patterns, they move the cold sequence to Semi-pilot and reserve Copilot for the highest-value accounts.
For more on how to structure cadences across these scenarios, LinkedIn B2B Prospecting Cadence 2026: 5 Touchpoints + Copy-Paste Templates covers the sequencing logic in detail.
Safety, Volume Limits and What Actually Protects Your Account
A common misconception is that one mode is inherently safer for a LinkedIn account than another. The mode does not determine risk — volume does. Autopilot, Semi-pilot and Copilot all operate within the same daily and weekly caps. Sending 200 manually approved messages in a single day carries the same account risk as 200 AI-approved messages.
What limits risk is the volume ceiling, not who approves the content. Chattie's default caps — 25 invites per day, 200 actions per day, 200 invites per week — are fixed by design, not by mode. The platform also simulates human-paced behaviour in sending rhythm and runs a 15-day account warm-up for new accounts.
LinkedIn's terms of service prohibit third-party automation tools as a category (see LinkedIn Help, article a1341387). Volume controls and human-paced sending reduce the likelihood of detection and restriction, but do not eliminate it. Any tool that claims zero risk of account restriction is misrepresenting how LinkedIn enforcement works.
For a detailed breakdown of what automation tools can and cannot safely do, AI for B2B LinkedIn Prospecting: 5 Ways to Use It Without Losing Your Account (2026) covers the enforcement landscape.
How Chattie Implements the Three Modes
Chattie — an AI SDR for LinkedIn — gives users the ability to select and switch between Autopilot, Semi-pilot and Copilot per campaign or per conversation. The mode can be changed mid-conversation without restarting the sequence, which matters when a cold campaign transitions into a warm dialogue.
In all three modes, the platform handles: connection invite sending, campaign sequencing, 15-day pending invite withdrawal, and scheduling link delivery (in Semi-pilot and Copilot, with user approval). The AI reads the lead's LinkedIn profile and recent activity to generate context-aware messages regardless of mode.
Chattie's current plans:
- Solo — US$75/month: 1 LinkedIn account included (up to 2 with extra seats), 3,000 AI credits and 2,500 leads per month (2,000 standard + 500 Deep Leads)
- Growth — US$190/month: 3 LinkedIn accounts included (up to 10 with extra seats), 10,000 AI credits and 7,500 leads per month (6,000 + 1,500 Deep Leads)
- Scale — US$285/month: 5 LinkedIn accounts included (up to 10 with extra seats), 18,000 AI credits and 12,500 leads per month (10,000 + 2,500 Deep Leads)
A quarterly billing cycle is available: Solo US$67.50/month equivalent (billed US$202.50 every 3 months), Growth US$171/month equivalent (billed US$513 every 3 months), Scale US$256.50/month equivalent (billed US$769.50 every 3 months).
AI credits are consumed at 1 credit per AI-generated message and 1 per ICP Filter verdict. Manual messages, invites, profile visits and withdrawals do not consume credits. Credits do not carry over between months.
Across all active Chattie campaigns in H1 2026, the platform recorded a 34% average reply rate (aggregated, anonymized data across all active campaigns).
FAQ
What is Autopilot mode in an AI SDR for LinkedIn?
Autopilot is the mode in which the AI sends every message in the sequence without human approval. The AI reads the lead's profile, writes a contextualised message and sends it within the configured volume limits. It is the highest-throughput mode, suited to operations with a well-defined ICP and validated copy.
Which AI SDR mode carries the lowest account risk on LinkedIn?
The mode does not determine account risk — volume does. Autopilot, Semi-pilot and Copilot all operate within the same daily and weekly caps. A high volume of manually approved messages carries the same risk as the same volume of AI-approved messages. What protects the account is the volume ceiling and human-paced sending behaviour, not who approves the content.
When is Autopilot the wrong choice for B2B prospecting?
Autopilot loses effectiveness when a lead replies with complex technical objections, specific product questions, or context the AI cannot map well without account history. For high-ticket ICPs or long consultative cycles, Semi-pilot or Copilot prevent a generic automated reply from burning a warm lead at the worst possible moment.
Can the mode be changed mid-campaign or mid-conversation?
Yes. In Chattie, the mode can be switched per campaign or per individual conversation without restarting the sequence. A common pattern is to run cold outreach on Autopilot and switch a specific conversation to Semi-pilot or Copilot when a high-value reply comes in.
What is the difference between Semi-pilot and Copilot?
The core difference is when human approval is required. In Copilot, every message — including cold openers and follow-ups — requires approval before sending. In Semi-pilot, the AI sends the full cold sequence autonomously and only pauses for human approval once the lead replies. Semi-pilot preserves AI efficiency for cold outreach and human judgment for warm conversations. Copilot applies human oversight to the entire sequence.
Does Chattie have a free trial?
No. Chattie does not offer a free trial, a free plan, or a no-card-required period. Paid plans start at US$75/month on the Solo plan.
The choice between Autopilot, Semi-pilot and Copilot is a decision about where your time and your AI's time are best spent. Cold sequences at scale benefit from automation. High-intent replies from accounts worth closing benefit from human judgment. The mode is the dial between those two realities — and in Chattie, it is adjustable per conversation, not locked at account level.