What is a buying-signal agent? Definition and 2026 guide
2026-09-27 · 8 min read · Adam Axelsson, founder of Revexa
A buying-signal agent is an AI agent that starts from evidence that a company may buy now, not from a list. It watches for signals such as engagement with competitors' posts, hiring, funding or a system switch, judges each company against your ideal customer profile, finds the right person and drafts outreach that a human approves. An AI SDR sends to a list. A signal tool stops at the alert. The agent does both halves.
The category needed a new name. "AI SDR" promised a replacement for a salesperson and too often delivered volume, burned domains and churn. The buyers who stayed learned something: the wins came from reaching the right company at the right moment, with a person checking the message. That is what a buying-signal agent is built around.
Honest disclosure: Revexa builds Leo, a buying-signal agent. This article defines the category as we see it, links every figure to a source and says where the evidence is weak.
Buying-signal agent: a working definition
A tool qualifies as a buying-signal agent when it does all five of these, in this order:
- Detects a signal. An observable event that changes the timing: someone at the company engages with a competitor, the company hires for a relevant role, raises money or changes systems.
- Qualifies the company, not just the person. It checks the business behind the signal against your ICP: what they sell, to whom and how, not only the title of whoever clicked.
- Finds the right person. The engager is not always the buyer. The agent works out who at the company owns the problem.
- Drafts outreach tied to the evidence. The message builds on why this company, and why now, in your voice.
- Puts a human in the loop by default. You approve before anything is sent, and the agent keeps to human sending volumes.
Miss step 1 and you have an AI SDR working a list. Miss steps 2 to 4 and you have a signal feed. Miss step 5 and you have automation you cannot see into.
How is it different from an AI SDR?
The difference is where the work starts and who stays in control.
| AI SDR | Buying-signal agent | |
|---|---|---|
| Starting point | A list or database segment | An event at a specific company |
| Why now? | Usually none, cadence decides | The signal is the reason |
| Default mode | Autonomous sending | Human approves |
| Volume | High by design | Limited by signal and pacing |
| Typical price | $250 to $3,750+ per month, often annual | $49 to $199 per month, often monthly |
| Main failure | Generic copy at scale | Noisy signals without qualification |
Prices: AiSDR Solo $250 per month and 11x from $3,750 per month on an annual contract, per each vendor's site in September 2026. More in AI SDR pricing. Signal agents: Gojiberry $99, Valley from $149 billed quarterly, per their own pricing pages.
Why the word "AI SDR" is damaged:
- The poster child stumbled. In March 2025, TechCrunch reported that 11x listed ZoomInfo and Airtable as customers when they were not. Employees alleged that only about $3 million of a claimed $14 million ARR had passed the three-month trial, and some described losing 70 to 80 percent of customers.
- Even 11x's CEO says it. Prabhav Jain opened a May 2026 interview with "AI SDRs don't work", then clarified that the problem is how teams deploy them, not the technology itself.
- The evidence is thin. Sushidata's State of the AI SDR 2026 found practically no public, controlled studies comparing AI with humans all the way to revenue.
None of this means AI cannot help with outbound. It means the unit that works is smaller: one qualified company, one reason, one message a human stands behind.
How is it different from signal tools and automation tools?
| Category | Examples | What it does | What it leaves to you |
|---|---|---|---|
| Signal tool | traxy, SignalRaven, Trigify (closing 22 Oct 2026) | Surfaces engagers and events | Qualifying, finding the person, writing, sending |
| Intent data platform | 6sense, Demandbase, Bombora | Account-level research intent | Everything at person level |
| Automation tool | lemlist, Waalaxy, Expandi, HeyReach | Sends sequences, some import signals | Deciding who is worth contacting and what to say |
| Workshop | Clay | Lets you build any flow | Building and maintaining it |
| Buying-signal agent | Leo, Gojiberry, Valley | Signal to approved message | Approving and taking the meeting |
Signal tools are being absorbed. HubSpot bought Warmly in June 2026 and the Trigify team in September, Zoom agreed to buy Common Room in July. When the signal alone becomes a feature of a CRM, the value moves to what happens after it: judging the fit and turning it into a conversation. Our Trigify alternatives guide covers the fallout.
Which buying signals are strongest?
Not all signals are equal, and in Europe not all are available. Here is our reading of the evidence.
| Signal | Why it matters | Evidence | Usable in the EU? |
|---|---|---|---|
| Product usage (PLG) | They already use you | Unify reports 9.1% positive replies, its best signal type, on its own data | Yes, first-party |
| Champion changes job | Someone who bought before, in a new seat | Widely described as top converting, few public numbers | Yes |
| New sales or revenue leader, hiring for your area | Budget and a mandate to change | Consistent across vendors, no neutral benchmark | Yes, public |
| Funding round | Money and pressure to grow | Common trigger, weaker on its own | Yes, public |
| Engagement with competitors or category voices | Shows interest in the problem at person level | Vendor claims only, no published method | Yes, public LinkedIn activity, with GDPR care |
| Person-level website visits | Strong intent | RB2B states its person-level identity is "US-only" | No |
For scale: Instantly's 2026 benchmark, based on 2025 data, puts the average cold email reply rate at 3.43%, the top 25% of campaigns at 5.5% or more and the top 10% at 10.7% or more. The report does not tie those levels to signals, and we have found no neutral study that measures how much signals move reply rates.
The European gap matters. Person-level website identification works in the US only, per RB2B's own site. In the EU, public engagement on LinkedIn is one of few signals that point to a person rather than a company. That is why most buying-signal agents lean on it, and why we wrote a separate guide to LinkedIn engagement as a buying signal.
What we have learned building Leo: the signal is rarely the bottleneck. The ICP is. A signal filtered against a vague ICP is noise with a timestamp.
The honest risks
LinkedIn restrictions. LinkedIn's user agreement prohibits scraping and automation. It sued Proxycurl, which shut down in July 2025, and in March 2026 it deleted HeyReach's company page and restricted its founders' profiles, per HeyReach. Any agent that sends from your LinkedIn account carries account risk. Pacing, caps and working hours reduce it. Nobody can promise zero risk.
Engagement is interest, not purchase intent. A like on a competitor's post can mean evaluation, loyalty or boredom. Without qualification, engagement signals are mostly noise.
Privacy law. Names and work emails are personal data under GDPR. France's CNIL fined Kaspr €240,000 in 2024 for collecting LinkedIn contact details users had chosen to hide. Legitimate interest can work for B2B, but it needs a documented balancing test and a clear opt-out. See our GDPR guide for B2B email.
Country rules. Germany requires prior consent for advertising by email and, per the courts, by LinkedIn message, even B2B. Read cold email to Germany under UWG §7 before you add DACH to your ICP.
AI disclosure. Article 50 of the EU AI Act applies from 2 August 2026 and covers AI systems that interact directly with people. A model where a person reads and approves every message sent in their own name is, in our reading, the more defensible design. This is not legal advice.
How to evaluate a buying-signal agent
Run a seven-day test with your own judgment as the benchmark:
- Give it your website only. See what ICP it builds without hand-holding.
- Ask for 25 leads. Grade each one yourself: would you want a meeting with this company?
- Count agreement. Your verdicts against the agent's. Low agreement means you will spend your week correcting it.
- Open five leads at random. Can you see the signal, the evidence and why the company fits?
- Read ten drafts. Would you send them unedited, in your name?
- Ask about sending. Daily caps, working hours, what happens if LinkedIn restricts your account.
- Check the exit. Monthly terms, data export, no annual lock-in.
For reference, in an internal measurement the ICP Leo built from a website matched 19 of 25 of our founder's own lead verdicts on day one. Hold every vendor, us included, to a number like that from your own test, not from a case study.
Where Leo fits
Leo is Revexa's buying-signal agent. You enter your website and Leo builds the ICP, judging business model rather than job titles. He watches engagement on competitors' and industry voices' LinkedIn posts, plus hiring, funding and system switches, finds the right person and drafts in your voice for LinkedIn and email. Every lead shows the signal and the reasons it fits, with quotes from the company's own site. You approve everything by default, and you can steer Leo from iMessage. $99, €89 or 999 SEK per month for 2 LinkedIn profiles and 1 mailbox, 7 days free, no commitment. Start Leo.
Frequently asked questions
What is a buying-signal agent?
An AI agent that detects events showing a company may buy now, qualifies the company against your ICP, finds the right person and drafts outreach that a human approves. It combines a signal tool and an outreach tool, with judgment in between.
Is a buying-signal agent the same as an AI SDR?
No. An AI SDR starts from a list and usually sends autonomously at volume. A buying-signal agent starts from a specific event at a specific company, qualifies it and keeps a human approving by default.
What are the best buying signals in B2B?
First-party signals such as product usage and repeat visits are strongest, followed by champion job changes, new leaders and hiring in your area. Engagement with competitors on LinkedIn sits in the middle: weaker alone, useful once qualified against a sharp ICP.
Are buying-signal agents safe for my LinkedIn account?
Not risk-free. LinkedIn restricts automation and has acted against vendors. Choose an agent with daily caps, human pacing and approval before sending, and never run two tools on the same account.
How much does a buying-signal agent cost?
Most sit between $49 and $199 per month: Gojiberry $99, Valley from $149 quarterly, Leo $99, €89 or 999 SEK. Autonomous AI SDRs typically cost $250 to $3,750 or more per month.