From Shubhi K | Product & Market Analysis
Artisan AI Review: Ava's Price, Its Results, and the 'Stop Hiring Humans' Bet
On this page
Artisan sells Ava as an AI business development representative that finds leads, writes outbound email and books meetings. At its April 2025 Series A, Artisan said it had about $5 million in annual recurring revenue and 250 customers. Artisan AI publishes no price, and the best procurement data puts the median contract near $30,000 a year. Whether that buys pipeline comes down to one number the company itself put at about 1%.
Key takeaways
- Artisan does not publish a price for Ava. Vendr's buyer data shows a median Artisan contract of $30,000 a year, ranging from $18,300 to $80,371, normally paid annually in advance.
- Artisan's traction figures come from its CEO and are not audited. About $5 million of ARR across 250 customers was reported to TechCrunch at the April 2025 Series A, which works out to roughly $20,000 per customer.
- The planning number for Ava is a 1% response rate. Artisan's CEO called about 1% the sweet spot, so 12,000 leads on a $30,000 contract buys roughly 120 responses at $250 each, before a single meeting is booked.
- Artisan retired "Stop Hiring Humans" in 2026 and hired a human BDR. Its CEO told PitchBook that most customers do not deploy Ava to replace headcount, which is the more useful frame for a buying decision.
The short answer
Artisan AI sells Ava, an AI BDR that prospects from a contact database Artisan sizes at 250 million records, sends email from your own domains, handles replies and books meetings. There is no public price. Vendr's data shows a $30,000 median annual contract. Buy it only if a response rate near 1% pays for itself at your deal size.
What Artisan's Ava actually does
Artisan's homepage describes Ava as an AI BDR that "finds and enriches B2B leads, writes and sends personalized outreach, handles replies, and books meetings". The product page lists the same steps as lead discovery, enrichment, campaign creation, autonomous replies and scheduling. A BDR, or business development representative, is the junior sales role that books first meetings for account executives.
That is the whole top of the outbound funnel in one product. It competes with a stack of separate data, sequencing and deliverability tools, and Artisan pitches the consolidation as a saving in itself.
The documented workflow
Ava draws on a contact database that Artisan describes as 250M+ verified contacts, filtered by title, industry, geography and company attributes. Campaigns go out by email from your own domains, in your reps' names. TechCrunch reported in 2025 that Artisan also claims a proprietary database of brick-and-mortar businesses, which it treats as a differentiator.
The writing process is more constrained than the marketing suggests. In the same 2025 report, customers entered company information through a form, and Ava then used a set of rigid prompts built with model provider Anthropic. Rigid prompts are the reason Artisan gives for lower error rates. They are also why the messages can read as templated if your inputs are thin.
Because Ava sends from your domains, the deliverability risk sits with you rather than with Artisan. That cost, and how volume damages sender reputation, is covered in the analysis of what AI SDR volume does to cold email deliverability.
The controls have moved toward humans
Artisan's current product copy describes "human-in-the-loop controls", listed as approvals, tone and CTA locks, and escalation rules. The Ava 2.0 release, published on 26 May 2026, describes an agent that runs outbound "with approval gates wherever you want them". The same page says Ava is deployed for "thousands of AEs and BDRs at large enterprises," without a number.
That wording matters more than any feature on the list. A product once sold as a replacement now describes itself as a tool with gates. We think that is the more honest pitch, and it is the one you should buy against.
LinkedIn is the channel gap. TechCrunch reported in January 2026 that LinkedIn restricted Artisan's company page and staff accounts from 19 December 2025, objecting to its use of LinkedIn's name and to data brokers alleged to have scraped the site. Artisan was reinstated after about two weeks. Several 2026 reviewers still complain that Ava no longer automates LinkedIn outreach, which Artisan disputes in its replies.
Artisan AI pricing: what Ava costs in practice
Artisan's pricing page describes "credit-based pricing" and plan tiers from startups to enterprise. Its own site says plans are scoped to lead volume, mailboxes and dialer seats. It shows no dollar figure.
There is no list price, and that is a cost
One third-party pricing tracker records that Artisan showed public prices in early 2024 and then removed them. It also records self-serve tiers published briefly around the Ava 2.0 launch and pulled again by July 2026. We could not open Artisan's own archived pages to confirm those figures, so we do not quote them here. The pattern is the useful part: the price has been visible twice and withdrawn twice.
A product priced only through a sales call asks you to negotiate blind. That is normal in enterprise software. It is still a cost you carry, because the seller knows the distribution of deals and you do not.
What buyers actually signed
Vendr, which brokers software purchases, publishes the closest thing to a market price. Its Artisan page shows a median contract of $30,000 a year, with a range from $18,300 to $80,371. Payment terms are listed as annual in advance, with monthly options "may be available upon request."
Vendr's packaging data also describes the tiers by volume. The Accelerate tier covers 12,000 leads contacted a year, roughly 36,000 emails. Blitzscale covers 65,000 leads, roughly 195,000 emails. A Custom tier targets sales teams with 20 or more seats.
The 2025 figures give a rough cross-check. $5 million of ARR across 250 customers is about $20,000 per customer. That sits inside Vendr's range and below its median, which fits a customer base that has moved toward larger accounts since 2025.
| Source | What it says | Weight |
|---|---|---|
| Artisan pricing page | Credit-based plans scoped to lead volume, mailboxes and dialer seats. No prices. | Authoritative on structure, silent on cost. |
| Vendr buyer data | $30,000 median, $18,300 to $80,371 range, annual payment in advance | Best available, sample not disclosed. |
| Artisan CEO via TechCrunch, 2025 | $5M ARR across 250 customers, about $20,000 each | Company claim, dated. |
| Public reviews, 2024 to 2026 | One reviewer paying about $2,000 a month, another paying over £1,000 a month | Anecdotal, two customers. |
Credits are the second pricing question, and the harder one to check. One Trustpilot reviewer wrote in September 2026 that 40,000 credits were sold as enough for about 2,000 new prospects and covered about 1,200. That is roughly 33 credits per prospect against an expected 20. It is one customer, and it is exactly the gap you should test for, as set out in the guide to why credit and token billing confuses buyers.
The numbers Artisan has published, checked
Artisan has disclosed very little hard data, and almost all of it came from one interview. That is not unusual for a private company. It does mean every figure below rests on the CEO's word.
The April 2025 round was a $25 million Series A led by Glade Brook Capital, with Y Combinator, Day One Ventures and HubSpot Ventures participating. Artisan had raised about $12 million the previous September. At the time it employed 35 people and was hiring 22 more, including in sales.
The one-in-10,000 hallucination claim
Carmichael-Jack told TechCrunch that Ava "only hallucinates maybe one in 10,000 emails, if that". He was frank about the history: "We had extremely bad hallucinations when we first launched". A hallucination here means an email asserting something false about the prospect or the seller.
No test method was published, so the figure cannot be checked from outside. You can check it on your own account in an afternoon. Pull 200 sent emails at random and mark every factual claim. If you find one error, the rate is far worse than one in 10,000, and you have learned that before renewal rather than after.
| Claim | Who said it | Independent evidence | Our reading |
|---|---|---|---|
| $5M ARR and 250 customers | CEO to TechCrunch, April 2025 | None published | Plausible and dated. No update since. |
| Hallucinates about one in 10,000 emails | CEO to TechCrunch, April 2025 | None published | Untested. Audit a sample yourself. |
| About 1% response rate is the sweet spot | CEO to TechCrunch, April 2025 | Reviews range from 0.07% positive to over 1% | Use as a ceiling when you plan. |
| Output of a 20-person BDR team with no BDRs | CEO to PitchBook, August 2026 | None published | Describes Artisan's own use, not yours. |
| $2M spent on the billboard campaign | Artisan blog, via the Chronicle | Company figure | Credible, and against its own interest. |
Ava's 1% response rate and the cost per reply
The single most useful thing Artisan's CEO has said is the least promotional. He told TechCrunch that "the sweet spot" is about a 1% response rate, and that first-generation AI SDRs "get a pretty low response rate" with "relatively high churn". Take him at his word and build the business case from there.
From contract to reply
The arithmetic is simple and it is illustrative, because Vendr's median is not tied to a specific tier. At $30,000 for 12,000 leads contacted, each lead costs $2.50. A 1% response rate returns 120 responses a year, which is $250 per response.
Responses include "not interested" and "remove me". If you apply the 0.07% positive rate one 2026 reviewer reported, the same contract returns about 8 positive replies. That is close to $3,570 each, before anyone has taken a meeting.
| Leads a year | Response rate | Responses | Cost per response |
|---|---|---|---|
| 12,000 | 1%, the CEO's sweet spot | 120 | $250 |
| 12,000 | 0.5% | 60 | $500 |
| 12,000 | 0.07% positive, one reviewer | About 8 | About $3,570 |
| 65,000 | 1% | 650 | About $46, if the same $30,000 bought Blitzscale volume |
The last row concedes where volume wins. Higher tiers cost more than the median, so the real figure sits above $46, but the direction holds: Ava's economics improve sharply with list size, provided your domains can carry the volume.
A reply is not a meeting
Artisan has not published a meeting rate, a show rate or a pipeline figure. Those are the numbers that decide the purchase. We would not sign without a cost-per-meeting figure written into the trial terms, and we would not accept a response rate as a substitute.
Your comparison is a human SDR's loaded cost against the meetings they book. Use your own figures for both, because published SDR benchmarks vary widely by market. Data cost matters too, and the comparison of cost per contact across Clay, Apollo and ZoomInfo shows what the list alone costs if you build the stack yourself.
What Artisan customers say in public reviews
Artisan's Trustpilot profile showed a 4.1 rating from 42 reviews in October 2026, with 81% of them five-star. At least three of those five-star reviews describe different businesses called Artisan: a coffee shop, an apartment building and a plumber. Strip them out and the score is lower than it looks.
The negative reviews cluster on three things. Results come first: one 2024 reviewer reported "zero meetings booked through the platform," and a 2026 reviewer reported "only about a 0.07% positive response rate." Credits come second, as described above. Commitment comes third: a 2024 reviewer said Artisan would not allow cancellation and was "forcing me to pay $2K/month for a year."
The praise is mostly about people rather than the agent. Many positive reviews name individual customer success managers. One January 2026 reviewer reported "over 1% reply rate," which matches the CEO's stated sweet spot rather than beating it.
Our view is that the reviews describe a product that works when it is set up and supervised well, and fails quietly when it is not. That is the same pattern documented in the analysis of which vendor case studies show verified outcomes, and it argues for buying the service relationship as carefully as the software.
The 'Stop Hiring Humans' bet, and why Artisan retired it
Artisan's billboards ran in San Francisco and New York with lines such as "Stop Hiring Humans" and "Artisans Won't Complain About Work-Life Balance". The company's own blog said it spent $2 million on the campaign, according to the San Francisco Chronicle. As early as April 2025, Carmichael-Jack told TechCrunch the billboards were "mostly just for attention."
In August 2026 Artisan said it had retired the slogan, and that it was hiring its first human BDR. The Chronicle reported in September that the replacement line is "Be More Human," and quoted the CEO saying "Artisan is a pro-human company".
The reasons he gave PitchBook are the most useful buyer information Artisan has published. "The majority of the time, when we've seen companies roll out our product, they're not doing it in replacement of headcount," he said. He added that "AI can't legally cold call or go to conferences". He also claimed Artisan runs "the output of a BDR org of 20 people without having any BDRs on the team".
Read plainly, the vendor's own evidence says Ava is a volume tool for an existing team, not a substitute for one. We think the campaign worked as marketing and misled as a product description. The retirement corrects the description, and you should hold the contract to the corrected version.
Where this review is weakest
This review leans on thin evidence in two places, and one competing view deserves a fair hearing.
The cost math rests on a median nobody can audit
Vendr does not show its sample size or an update date on the Artisan page. Its tier descriptions use plan names that Artisan may no longer sell. The $30,000 median could be stale, skewed toward larger buyers, or both. Every cost-per-response figure above inherits that weakness, which is why each one is labelled illustrative.
Public reviews are a small, self-selected sample
Forty-two Trustpilot reviews, several of them about other companies, cannot represent a customer base that already numbered 250 in 2025. Unhappy customers and delighted ones both review more than the middle. We could not open Artisan's G2 or Gartner pages during this research, so those samples are missing too.
The strongest counter-case is about deal size. If your average contract is large, 120 replies a year at $250 each could be an excellent trade, and a 1% response rate is not a failure. Ava 2.0 is also recent. Results from 2024 and 2025 reviews may describe a product that has since changed, and a fair buyer should test the current version rather than the reputation.
Who Ava fits, and who should pass
Artisan has said who fails. Carmichael-Jack told TechCrunch that "some customers will just completely flop" and that Artisan had "historically sold to a lot of the wrong customers". He said the company now qualifies heavily and turns away some industries, naming offshore development agencies.
Ava fits a team with a high contract value, a tightly defined buyer profile, healthy sending domains and a person who owns every reply within hours. It fits poorly where deals are small, where LinkedIn is the main channel, or where nobody will review what the agent sends.
Negotiate the commercial terms as hard as the features. TechCrunch reported in 2025 that Artisan's contracts included break clauses allowing early exit, and that it was piloting pay-per-response pricing through the billing platform Paid.ai. Ask for both by name. The clause structure worth asking for is set out in the guide to outcome pricing contract clauses.
Frequently asked questions
How much does Artisan AI cost?
Artisan does not publish prices. Its pricing page describes credit-based plans scoped to lead volume, mailboxes and dialer seats, and every quote goes through sales. Vendr's buyer data shows a median Artisan contract of $30,000 a year, with a low of $18,300 and a high of $80,371, normally paid annually in advance. Treat those figures as directional, because Vendr does not disclose its sample size on the page.
What does Artisan Ava do?
Ava is Artisan's AI business development representative. It searches a database Artisan describes as 250M+ verified B2B contacts, enriches leads, writes and sends personalised outbound email from your own domains, handles replies and books meetings. Since the Ava 2.0 release in May 2026, Artisan also describes approval gates and a dialer, so humans can review messages before they send or take warm leads by phone.
Is Artisan AI worth it?
It depends on your deal size more than on Ava. Artisan's CEO has described a 1% response rate as the sweet spot. At 12,000 leads a year on a $30,000 contract, that is about $250 per response before any meeting is booked. For a product with a large contract value, that can pay back quickly. For a low-value product it usually will not.
What response rate does Artisan Ava get?
Artisan has not published an audited response rate. Its CEO told TechCrunch in April 2025 that the sweet spot is about a 1% response rate, and that first-generation AI SDRs had low response rates and high churn. Public reviews range from one customer reporting over 1% to another reporting a 0.07% positive response rate. Ask for positive reply and meeting rates in writing.
Can you cancel an Artisan AI contract?
Terms are negotiated, not published. TechCrunch reported in April 2025 that Artisan's contracts included break clauses allowing customers to end early, and that it was piloting pay-per-response pricing through Paid.ai. A 2024 Trustpilot reviewer said they were held to $2,000 a month for a year. Vendr lists annual payment in advance as standard, so ask for the break clause before you sign.
Why did Artisan stop using Stop Hiring Humans?
Artisan retired the slogan in 2026, after about two years, and said it was hiring its first human BDR. Its CEO told PitchBook the line was viral marketing, that most customers do not deploy Ava to replace headcount, and that AI cannot legally cold call or attend conferences. The San Francisco Chronicle reported the replacement line as Be More Human.
Where to start before you take the demo
Write three questions down and get the answers in writing. What is the positive reply rate and the meeting rate across customers in your industry, with the sample size? What does one prospect cost in credits for a full campaign? Which break clause or per-response option will Artisan put in the contract?
If you already run Ava, pull 200 sent emails this week and mark every factual claim and every reply. That one sample tests the hallucination claim and your real cost per response at once, and it gives you a number to bring to renewal.
References
- TechCrunch, Julie Bort, Artisan, the 'stop hiring humans' AI agent startup, raises $25M and is still hiring humans, 9 April 2025. Used for funding, ARR, customers, staff, hallucination, response rate, contract and fit statements.
- TechCrunch, Julie Bort, Yes, LinkedIn banned AI agent startup Artisan, but now it's back, 7 January 2026. Used for the LinkedIn restriction and its stated reasons.
- PitchBook News, Jacob Robbins, Artisan told companies to 'stop hiring humans' for sales. Now it's hiring one for just that job, 18 August 2026, via Yahoo Finance. Used for the CEO's statements on headcount and the human BDR.
- San Francisco Chronicle, Aidin Vaziri, San Francisco AI start-up apologizes for 'Stop Hiring Humans' billboards, 10 September 2026, via Yahoo. Used for the retirement, the replacement slogan and the $2M campaign spend.
- Vendr, Artisan AI pricing and plans, accessed October 2026. Used for contract values, payment terms and tier volumes.
- Trustpilot, Artisan reviews, accessed October 2026. Used for rating, review count and quoted customer experiences.
- Artisan, Ava, the AI sales agent product page and Ava 2.0 launch post, 26 May 2026. Used for feature descriptions, the contact database size and human-in-the-loop controls.
The weakest part of this source base is that Artisan's traction and quality figures all come from its own CEO. The only market price comes from one procurement platform that does not disclose its sample. Figures are current as of 11 October 2026.
Related reading