How to automate SDR workflows with AI (and what to keep human)

How do you automate SDR workflows with AI? Automate the work that happens before a message goes out: building the list, finding verified emails and mobiles, researching accounts, prepping calls and logging activity. Keep people on what gets sent and on the phone.

This guide maps the SDR workflow task by task, gives you six setup steps with copy-paste prompts, a weekly schedule, and an honest comparison of AI SDR platforms against building your own stack.

What does it mean to automate an SDR workflow with AI?

An SDR workflow is the repeatable set of tasks a sales development rep runs to turn a target market into booked meetings: build the list, find contact data, research, call and email, handle replies and log everything. Automating it with AI means software runs the repeatable steps, so the rep spends the freed hours on conversations.

The room to automate is large. The Bridge Group’s 2025 report puts the median SDR at 112 activities a day across phone, email and other channels. Salesforce’s 2026 State of Sales survey found the average seller spends 40% of their time selling. The other 60% goes to work like quotes, planning, manual data entry and training.

Which SDR tasks to automate, and which to keep human

Automate a task when it repeats and a mistake stays inside your team. Keep a person on it when a mistake reaches the buyer. That rule sorts the whole SDR workflow into four groups.

How to automate SDR workflows with AI in 6 steps

The six steps below go in order of return: the first ones remove the most manual hours with the least risk. Set them up one at a time; each builds on the one before.

1. Time the SDR week before you automate anything

Ask each SDR to log what they do for one week, in 30-minute blocks: list building, looking up contact data, research, writing, calling, CRM updates. You will automate the biggest non-selling blocks first.

This week of data is also your baseline. Keep meetings booked, conversations and reply rate per SDR, so you can show what changed. Our guide to outbound sales metrics lists the ones worth tracking.

2. Automate list building with a one-sentence search

Replace hand-built lists with one sentence that describes the buyer, run on a schedule. “Decision makers at US software companies with 51 to 500 employees” is enough for a first pass. Check what the tool understood and the count before you spend anything.

In Scalelist’s Lead Finder, the count is free and each lead found costs 1 credit. Run the same sentence every Monday, drop companies already in your CRM (the Monday prompt below does it from a CRM export), and each SDR starts the week with fresh accounts instead of an old export.

3. Automate contact data and verification

Find verified emails and mobile numbers for the new list in the same run. Route each contact by what came back: a mobile goes to the call list, a Valid email goes to the sequence, a Risky email goes to a secondary domain, and Not found means trying another role at the account.

Re-check any list older than a few weeks before it enters a sequence. ZeroBounce’s 2026 report found at least 23% of email lists degrade within a year. Scalelist’s Email Verifier labels each address before you send.

4. Automate account research and call prep

Give the assistant a standing research prompt that returns facts with links, then a call brief for each account on today’s list. Research at the account level, not per contact, so one pass can support two or three people.

Run it the evening before or first thing in the morning, so the brief is ready when the SDR opens the account. Anything without a source stays out. Our step-by-step guide to AI sales prospecting shows the research prompt and which claims are safe to send.

5. Let AI draft, and keep a person on the send

Have the assistant write first touches and follow-ups for review, as drafts or in a paused campaign, never as live sends. The SDR edits and approves each one. This keeps the speed of drafting without the risk of a batch of confidently wrong emails.

The reason is evidence, not caution. 6sense’s 2026 State of the BDR study found that content production, the most common use of AI, and automated outreach “are not reliably associated with increases in quota attainment.” Gartner found 73% of B2B buyers actively avoid suppliers who send irrelevant outreach.

6. Automate logging, reply sorting and the weekly review

After each call, have the assistant write the summary and outcome to the CRM. Sort replies into interested, not now, wrong person, unsubscribe and out of office, and draft answers only for the routine ones. Then pull the week’s numbers every Friday.

HubSpot, Salesforce (Enterprise Edition and above), Pipedrive, Attio and Close all publish MCP servers, so an assistant like Claude can write to the CRM under each user’s permissions. From the Scalelist app you can also push selected leads to HubSpot with field mapping.

Example: a week for an AI-assisted SDR team

Here is the setup on one real market. Jordan Price manages three SDRs at Ridgeline Security, a fictional Denver company that sells security awareness training to US software companies with 51 to 500 employees. The people and companies are fictional; the count is real.

Jordan’s list sentence matches over 83,000 decision-maker profiles (director level and above) at US software companies of that size, as of September 2026. Each SDR takes 50 new accounts a week, one contact each.

At most, 150 fully enriched contacts a week cost 3,300 credits: 1 per lead, 1 per email, 20 per mobile. That is about 13,200 credits a month, which fits the Scaler 20K plan at 649 USD a month. Misses cost nothing, so the real spend is lower.

These are the three prompts the team runs in Claude with the Scalelist connector. They also work in ChatGPT.

On the morning of a call, the brief looks like this:

AI SDR platforms vs building your own stack

An AI SDR platform runs the whole loop for you: it picks leads from its own data, writes, sends and handles replies. Building your own stack means an assistant, a data connector and the sequencer you already use, with a person approving each send.

Pick a platform if you want one vendor and are comfortable with software sending in your name. Artisan offers controls to approve messages before they go out.

Build your own if you want the send, the data and the prompts to stay yours. Claude Pro costs 20 USD a month on monthly billing, and your sequencer stays the same. How far autonomous agents can go, and where they break, is covered in AI agents for lead generation.

Common challenges when you automate an SDR workflow

These are the problems to plan for before launch. Each one has a fix you can set up in advance.

AI calling that breaks the law

AI voices are not a shortcut for cold calls. The FCC confirmed in February 2024 that AI-generated voices count as “artificial” voices under the Telephone Consumer Protection Act, so calls to cell numbers need prior express consent. Artisan’s own site says “Ava cannot legally make calls.”

Fix: let AI prep the call and log it, and have a person dial. This is not legal advice; see our guide to B2B cold calling laws.

Research that sounds right and is not

An assistant will fill gaps with plausible guesses unless told not to. One invented funding round in an email tells the prospect you did not check. Fix: facts with dates and links only, and an SDR reads the brief before the call.

Stale contact data inside an automated loop

Automation runs on whatever list it was given. An old list keeps bouncing on schedule. Fix: rebuild the list weekly from the same sentence and verify emails right before they enter a sequence.

Personal AI tools nobody manages

6sense found only 27% of BDRs rely on company-provided AI tools alone, while 35% use personal tools only, such as ChatGPT or Claude. Fix: give the team one sanctioned setup and shared prompts, so results are comparable and data stays in company accounts. Our 30 Claude skills for sales are a starting library.

Saved time that turns into more sends

Gartner found AI saves sellers 4.8 hours a week, but 72% of sales organizations fail to reinvest that time in high-value activities. Fix: turn the saved hours into call blocks on the calendar, and hold send volume steady.

Best practices for AI SDR automation

The rules below keep automation from reaching the buyer before a person has checked it. Put them in writing before the first automated week.

• Start with one SDR and one segment for two weeks, then roll out.
• Keep prompts in one shared place and version them, so a change is visible.
• Review 20 drafts end to end before any new sequence goes live.
• Put call blocks on the calendar first, then let automation fill the rest of the week.

For the wider question of which sales tasks to automate across the whole team, read what to automate first in sales.

What to look for in AI SDR tools

The right tool fits into your SDR workflow without taking the send away from you. Check for:

• A readback and a free count before a list costs anything.
• Mobile numbers as well as emails, with a status on every email and no charge for a miss.
• An MCP connector for the assistant your team already uses.
• Direct sync to your sequencer, such as lemlist, Instantly or Smartlead. See sending leads to your sequencer.
• Drafts, not sends: any tool that writes should let a person approve before anything goes out.
• Published pricing you can model per SDR before a demo.

If you are comparing assistants for the research and prep steps, see our roundup of AI sales assistants for prospecting. A power dialer covers the calling side; our comparison of power dialers shows which fit a small team.

How Scalelist helps automate the SDR workflow

Scalelist is an AI Lead Finder: you describe your buyer in one sentence and get matching people with verified work emails and mobile numbers. It covers steps 2 and 3 of this workflow, and it runs inside the assistant your SDRs already use.

• Lead Finder shows what it understood and the count for free. Then 1 credit per lead found, 1 per email found (Valid or Risky), 20 per mobile found, 0 for a miss.
• The MCP connector (https://mcp.scalelist.com/mcp) runs the same search and enrichment inside Claude, ChatGPT or Cursor, at the same credits. The Scalelist MCP server page has 10 copy-paste prompts and 30 downloadable Claude skills.
• Where the list goes: export CSV, push selected leads to HubSpot, or send them to lemlist, Instantly, Smartlead, Woodpecker or Saleshandy.

Plans start at 29 USD a month for 500 credits, with no per-seat pricing, so adding an SDR adds no seat fee; see Scalelist pricing. New accounts get 20 free credits, no credit card required. For the full picture of where this fits, start from our AI lead generation hub.

FAQ

How do you automate SDR workflows with AI?

Automate the repeatable work before the send. Run a one-sentence list search every week, enrich it with verified emails and mobiles, have an assistant research each account with sources and prepare call briefs, and log calls to the CRM automatically. Keep a person approving every email and making every call, and measure meetings per SDR hour.

What is an AI SDR?

An AI SDR is software that runs outbound prospecting end to end: it finds leads in its own data, writes and sends sequences, handles replies and books meetings. Examples include 11x’s Alice, Artisan’s Ava and RegieGO. All three can send email on your behalf, but AI voice calls to US cell numbers need prior express consent, and none of the three uses AI to cold call.

Will AI replace SDRs?

No. AI is taking over the list building, data lookup and research share of the SDR job, which frees hours for conversations. The conversation, the judgment about which accounts matter and handling a real objection stay human. Gartner predicts AI agents will outnumber sellers 10 to 1 by 2028, yet fewer than 40% of sellers will say agents improved their productivity.

Which parts of the SDR workflow should not be automated?

Live calls, answers to real replies and the final approval of anything sent to a prospect. These are where a mistake reaches the buyer. AI can prepare all three: a call brief, a draft reply, a first-touch draft. A person should decide what actually goes out and hold the conversation.

How much does it cost to automate an SDR workflow with AI?

It depends on volume and on mobile numbers. In a build-your-own setup, the data is the main cost: in Scalelist, a lead found costs 1 credit, an email 1 and a mobile 20, with nothing charged for a miss. Three SDRs working 50 fully enriched accounts a week each need at most about 13,200 credits a month, which fits the 649 USD plan.

Can I automate SDR work inside ChatGPT or Claude?

Yes, for the preparation steps. Connect a data source over MCP, such as Scalelist’s connector, and the assistant can build a list from one sentence, find verified emails and mobiles, research accounts and write call briefs. CRM vendors including HubSpot and Salesforce publish MCP servers too, so the same chat can log the results.

Sources

• The Bridge Group: 2025 SDR metrics and compensation report
• 6sense: State of the BDR 2026 deep dive
• Salesforce: State of Sales report, 7th edition (February 2026)
• Gartner: AI saves sellers nearly 5 hours per week, yet 72% of sales organizations fail to reinvest time (May 19, 2026)
• Gartner: 61% of B2B buyers prefer a rep-free buying experience (June 25, 2025)
• Gartner: AI agents will outnumber sellers 10 to 1 by 2028 (July 28, 2026)
• FCC Declaratory Ruling FCC 24-17 on AI-generated voices (February 2024)
• ZeroBounce: Email list decay report 2026
• 11x: demo and pricing request page and consented outbound calling
• Artisan: pricing and Artisan vs 11x
• Regie.ai: RegieGO pricing
• Anthropic: Claude pricing
• Scalelist people count, 27 September 2026.