AI SDR Detector: Spot the Robot Pretending to Be a Salesperson

A new category of software has replaced human sales development reps: the AI SDR. These are fully autonomous agents — built on LLMs — that prospect, research, personalize, and send cold emails without any human per-send involvement. They're in your inbox right now, and most spam filters can't touch them.

What is an AI SDR?

An AI SDR (Sales Development Representative) is an autonomous AI agent that handles outbound sales outreach from start to finish. Unlike traditional sales sequences — where a human writes the playbook and an automation tool sends it — an AI SDR uses a large language model to research each target, generate a personalized message, and send it, all without a human reviewing or approving each email.

Tools like 11x.ai, Artisan, AiSDR, and Relevance AI sell exactly this: an "AI employee" that fills the BDR role. The AI researches your LinkedIn, reads your company website, finds a plausible hook, and sends you something that looks hand-written.

Why AI SDRs are different from spam

Traditional spam filters were built to catch bulk, impersonal email. AI SDR emails are the opposite: individually targeted, well-written, contextually aware, and sent from real Gmail accounts. They pass every spam heuristic. The only way to catch them is to look at the signals they can't hide.

• ESP platform fingerprints — AI SDR tools route email through infrastructure that leaves detectable headers (X-Mailer strings, MIME boundaries, DKIM patterns specific to certain platforms)
• Superhuman response speed — an AI SDR replies to your reply in seconds, regardless of timezone. Humans don't do that.
• Clockwork send cadence — follow-ups arrive at suspiciously regular intervals, stamped at precise offsets from the previous message
• LLM vocabulary patterns — phrases like "I hope this finds you well", "I'd be happy to", "please don't hesitate" appear at a frequency that humans don't naturally reach
• Structural templates — the exact 3-part formula: personalisation hook, value proposition, soft CTA. 80–250 words, every time

How AgentProof detects AI SDRs

AgentProof runs 23 signals against every incoming email in Gmail. The signals are organized into three false-positive tiers:

• Zero-FP signals (any one = flag): ESP platform fingerprint, prompt leakage artifacts, agent metadata in headers
• Low-FP signals (combine 2+): superhuman response speed, heartbeat cadence, API-only send path, ghost sender identity
• Medium-FP boosters: LLM vocabulary, low personalization ratio, structural template, sentence uniformity

The result is a 0–100 agent probability score displayed as a colored badge next to every sender name in your Gmail inbox. No configuration required. No rules to write. It just works.

The honeypot (Pro)

AgentProof Pro adds a feature that is provably unfakeable: an invisible instruction embedded in your sent emails. Any LLM that processes your email as a tool call will follow the instruction — and get caught immediately. A human reading the same email never sees it. Once a sender triggers your honeypot, AgentProof flags every future email from that agent with a 95+ score, regardless of how well-disguised it looks.

Which AI SDR platforms does AgentProof detect?

AgentProof's ESP fingerprinting database covers 25+ automation platforms, including Outreach, Salesloft, Apollo, HubSpot Sequences, Instantly, Smartlead, Woodpecker, Lemlist, Reply.io, Klenty, Mailshake, SalesLoft, and emerging AI-native SDR platforms including 11x, Artisan, AiSDR, and Relevance AI. The platform fingerprint database is updated continuously.

Try AgentProof free →