AI SDRs in 2026: what buyers and teams report

What is an AI SDR?

An AI SDR (AI sales development representative) does the work of a junior outbound rep. TechCrunch describes these tools as using "large language models (LLMs) and voice technology to craft personalized outreach emails and place automated calls to potential customers." The leading products now handle replies too. Here is how three well-known vendors describe themselves, checked on October 8, 2026. Everything in the table is the vendor's own claim.

The pitch has moved. Artisan became famous for billboards that said "Stop Hiring Humans." In 2025 its founder told TechCrunch the line "was mostly just for attention," and in 2026 its homepage sells an agent that works "alongside your team."

How many sales teams use AI agents in 2026?

Most sales teams use some AI, and about half of sellers have tried agents. The largest recent survey comes from Salesforce, which sells its own SDR agent, so read it as vendor data. It covered 4,050 sales professionals in 22 countries, Canada and the US among them, in August and September 2025.

• 87% of sales organizations use some form of AI, for prospecting, forecasting, lead scoring or drafting emails.
• 54% of sellers have used AI agents, and nearly 9 in 10 plan to by 2027.
• 55% use AI for prospecting, and 48% say they lack the bandwidth for proper cold outreach.
• Sellers expect agents to cut prospect research time by 34% and email drafting by 36%. Those are their expectations for when agents are fully in place.

Gartner's survey of 210 sales leaders, run in January and February 2026, measured time saved: AI saves sellers an average of 4.8 hours a week. Yet 72% of sales organizations report low reinvestment of that time into high-value selling. The Bridge Group's 2026 study of 158 B2B companies found the same unevenness: leaders rated 10 of 11 AI uses in account executives' work as split between "game changer" and "hit or miss."

Do AI SDRs deliver results?

Some teams see a clear return, and many can't prove one yet. We found no independent benchmark of AI SDR reply or meeting rates. The evidence is surveys of sales leaders, analyst predictions and vendors' own numbers:

Gartner's Melissa Hilbert summed up the prediction: "AI agents are everywhere, but there's a value ceiling. Beyond a certain point, more AI does not mean more productivity." Gartner also warns about "agent washing", where chatbots and older automation get rebranded as agents, and estimates that only about 130 of the thousands of agentic AI vendors are real.

What happened with 11x and Artisan?

The two best-known AI SDR startups had public growing pains, and their stories show where the category was oversold.

• 11x. In March 2025 TechCrunch reported, from nearly two dozen sources, that 11x showed logos of companies that were not customers. ZoomInfo said that in a one-month pilot the product "performed significantly worse than our SDR employees." One employee said the company was "losing 70-80% of customers that came through the door." 11x blamed "human error" for the logos, said it reports contracted revenue and put its retention at 79%. Its founder stepped down as CEO in May 2025.
• Artisan. In April 2025 its founder told TechCrunch that first-generation AI SDRs "get a pretty low response rate" and have "relatively high churn," and that the product had "extremely bad hallucinations" at launch. He put the sweet spot at "about a 1% response rate" and said Artisan had "historically sold to a lot of the wrong customers."
• LinkedIn. In December 2025 LinkedIn restricted Artisan's company page and staff profiles. Artisan's CEO said LinkedIn objected to the use of its name and alleged that data brokers Artisan used had scraped the site; the company was reinstated about two weeks later. LinkedIn's User Agreement bans "bots or other unauthorized automated methods" used to add contacts or send messages.

Investors had doubts early. In late 2024 TechCrunch reported VCs questioning whether AI SDR adoption is "sticky." One investor relayed a public-company CRO's account of an AI SDR that generated a large volume of leads over nine months that "did not lead to actual sales."

These are single companies' stories, and both companies say their products have improved since. They are still a useful list of what to check before you buy.

How do buyers react to AI outreach?

Buyers use AI themselves, and they still want a person when the decision gets real. Gartner's buyer surveys are the most solid data here:

• 73% of B2B buyers "actively avoid suppliers who send irrelevant outreach" (632 buyers, surveyed in 2024).
• 67% prefer a buying experience without a sales rep, and 45% used AI during a recent purchase (646 buyers, 2025).
• 69% prefer to validate AI-generated insights with a sales rep (645 buyers, 2025).
• Compared with generative AI, buyers were 39 percentage points more likely to say a rep understood their needs, 32 points more likely to say a rep made them confident in the decision, and 28 points more likely to say a rep helped them reach the next step (same 645 buyers).
• 69% of decision makers say AI-written email bothers them "unless the output feels genuinely human" (Hunter, vendor data, sample size not published).

Gartner predicts that by 2030, 75% of B2B buyers will prefer sales experiences that prioritize human interaction over AI. One of its analysts described "a reversal" of the rep-free trend, "especially in complex or high-stakes transactions."

Where does AI help outbound, and where does it fall short?

Gartner draws the line clearly: "AI is well suited to activities, such as account research, personalized messaging, signal monitoring and next best actions, while sellers remain differentiated in empathy, judgment, contextual understanding and value framing." Here is how the evidence lines up by task. The middle column is our reading of it.

Several sources point to data as the limit. In Salesforce's survey, 51% of sales leaders using AI say disconnected systems slow their AI work. Salesforce's own SDR agent answered "I don't know" to 30% of questions about a lead at launch, and needed data cleanup to get that under 10%. Its EVP of Sales put it plainly: "Stand-alone agents without comprehensive customer context tend to fail."

How should a small team evaluate an AI SDR?

1. Get an early exit in writing. Both 11x and Artisan sold contracts with break clauses, typically at three months in 11x's case.
2. Measure meetings held and pipeline created per 1,000 contacts. Emails sent says little, and Gartner advises measuring capacity and outcomes beyond time saved.
3. Call references yourself. TechCrunch found logos on 11x's site from companies that said they were not customers.
4. Read a sample of real messages before anything runs on autopilot. Hallucinations were the top complaint about early AI SDRs.
5. Decide which replies go to a person, and when. Artisan's own agent lets you set escalation rules for this.
6. Ask where the contact data comes from. LinkedIn restricted Artisan over its name and scraped data, and LinkedIn's User Agreement bans unauthorized automation on members' accounts.
7. Plan what your reps do with the time saved. Gartner found 72% of sales organizations don't reinvest it in high-value selling.
8. Check it is an agent at all. Gartner's "agent washing" warning applies to sales tools too.

Where Duckrow fits

Duckrow is an outbound workspace, and we don't sell it as an AI SDR. It uses AI at the steps where the evidence above is strongest and leaves the conversation to you:

• Use AI columns research each lead in your table, with templates for lead fit, company summary and opening line. Advanced mode reads the lead's LinkedIn profile first.
• Draft with AI writes a first version of email, text and LinkedIn steps for you to edit.
• Email steps can wait in a review list, so you preview, send or skip each one.
• In the inbox, AI sorts replies (interested, not interested, out of office, unclear), summarizes a thread and drafts a reply that you send.
• A reply on any channel stops that lead's sequence.

Duckrow does not answer replies on its own or book meetings.