The Bionic SDR: AI-Assisted Outbound at a Sensible Volume

Why the old title went

This article used to promise "450 personalised emails daily at 12%". AfM has no published record that supports either number: no case study shows that volume, and none reports a 12% rate (the old title did not even say 12% of what). So the promise is gone. What replaces it is more useful: how to split an SDR's work with AI, and how to work out a daily volume that your market, your team and the mailbox providers will tolerate.

What a bionic SDR actually is

A bionic SDR is a person whose research and drafting are done largely by AI, so their hours go on judgement and conversations. It is not an unattended sending machine. The split that holds up looks like this:

Who does what in an AI-assisted SDR role

Task	AI does	The rep does
Account research	Summarises company news, hiring, tech stack clues and recent posts into a short brief	Decides whether the account fits and why now
Contact finding	Pulls candidate contacts and roles from a data provider	Picks the one or two people the message is really for
First draft	Writes a draft from the brief and your approved positioning	Edits for accuracy and tone, deletes anything invented
Sending	Schedules within the limits you set	Sets the limits and pauses when metrics move
Replies	Classifies replies and drafts responses	Answers anything with a question, objection or buying signal
Opt-outs	Applies suppression immediately	Checks that suppression is working weekly

The rule underneath the table: AI may prepare anything, but a person decides who is contacted and what is claimed. Personalisation that the rep has not checked is a liability, because a confident wrong fact about someone's company is worse than no personalisation at all.

Working out a sensible volume

Volume should come out of four constraints. Take the smallest.

1. How many good-fit accounts exist

Count the companies that match your written ideal customer profile, not the size of a database export. Illustrative example: a UK firm selling to manufacturers with 50 to 250 staff in two regions might find 2,000 companies that truly fit. At 450 new contacts a day, five days a week, that list is used up in under a week of sending if you contact one person per company, and in between four and five weeks even at five people per company. After that you are either repeating people or loosening the profile until it no longer means anything.

2. How long a real review takes

If a rep spends one minute checking each personalised draft, 450 drafts is 7.5 hours of checking a day before a single reply is answered. Illustrative example: at two minutes per draft and three hours a day set aside for review, the ceiling is 90 messages a day. That is the honest meaning of "personalised".

3. How many replies one person can handle

Replies are the point, and each one needs a considered answer within hours, not days. Illustrative example: if 3% of recipients reply, 450 a day produces around 13 replies a day, many needing research or a proposal. Decide how many thoughtful replies a rep can handle and work backwards.

4. What mailbox providers will tolerate

Google's sender guidelines ask all senders to keep their spam rate below 0.10% and never to reach 0.30%. Senders of more than 5,000 messages a day to Gmail accounts must also authenticate with SPF, DKIM and DMARC and offer one-click unsubscribe on marketing messages. Yahoo sets matching expectations for bulk senders, including a complaint rate below 0.3% and honouring unsubscribes within two days.

The arithmetic is unforgiving. At 450 emails a day, 0.3% is between one and two complaints a day. Cold email to people who have never heard of you draws complaints more readily than email to subscribers, so a high-volume programme can cross the line within days if targeting slips. The 5,000 a day bulk threshold is not a safe harbour below it either: the spam rate guidance applies to every sender.

For most B2B teams the four constraints land far below 450 a day per rep. That is not a failure of ambition. It is what a message that is genuinely worth reading costs to produce.

What UK law asks you to check

This is not legal advice; it is the list of questions to put to whoever advises you.

• • Who is the subscriber? Under PECR, email marketing to corporate subscribers (limited companies, LLPs and other bodies with separate legal status) does not need prior consent. Sole traders and some partnerships count as individual subscribers, so they need consent or the soft opt-in. Your list needs to know which is which.
• • Is there a lawful basis under UK GDPR? A named person's work email is personal data. The ICO says direct marketing can be a legitimate interest when it complies with the electronic marketing rules, which normally means doing and recording a legitimate interests assessment.
• • Can people stop it easily? Every message must give a valid way to opt out, and an objection to direct marketing must be honoured. The ICO describes that right as absolute.
• • Have the penalties changed? Legal commentary reports that the Data (Use and Access) Act 2025 raised maximum fines for PECR breaches to UK GDPR levels, so check current ICO guidance rather than older articles.

A daily rhythm that works

1. • Morning, 30 minutes: review yesterday's replies, complaints, bounces and unsubscribes. Pause sending if any of them moved sharply.
2. • Account selection, 45 minutes: approve today's accounts from the AI-prepared briefs and reject weak fits.
3. • Draft review, 90 minutes: edit and approve drafts; delete any claim you cannot verify.
4. • Replies, the rest of the day: conversations first, always.

What to measure

Track positive replies, meetings held and opportunities created per 100 accounts contacted, alongside complaint and bounce rates. A rising send count with a flat opportunity count means the machine is working and the programme is not. For the wider process, see automating B2B lead generation without spamming, and to price the result, the lead generation cost calculator.

FAQ

Is 450 personalised emails a day ever realistic?

Only where the addressable market is very large, drafts are checked quickly, a team shares reply handling, and complaint rates stay under the provider limits. For a single rep in a defined B2B niche, the four constraints in this article usually produce a much smaller number.

What was the 12% in the original title?

It was not defined and no AfM record supports it, so it has been removed. If you see a reply or meeting rate quoted without its denominator, ask what it is a percentage of and over what period.

Does the 5,000 a day threshold only apply to large senders?

The extra authentication and one-click unsubscribe requirements are tied to that bulk volume, but Google's spam rate guidance applies to all senders, and in practice authentication is worth setting up at any volume.

Should the AI send replies on its own?

Only for mechanical cases such as confirming an unsubscribe. Anything containing a question, an objection or interest deserves a person, because that is where a sale is won or a reputation is lost.