AI & Sales8 min read

How Agentic AI Is Changing the SDR Function — And What It Means for B2B Sales

Agentic AI doesn't just automate tasks — it changes the fundamental division of labour in sales. Here's what the SDR function looks like when AI agents handle the intelligence work.

Shane Kennedy

1 March 2025

The SDR model is under pressure

The traditional SDR model — hire junior talent, give them a list and a script, measure on activity — has been showing cracks for years. Response rates have declined steadily. Buyer tolerance for generic outreach is at an all-time low. And the skills required to succeed in the role have shifted from volume and persistence toward research, contextualisation, and genuine commercial insight.

At the same time, AI tools have proliferated. Every sales team has experimented with ChatGPT for email personalisation or Apollo for data enrichment. But most of these experiments haven't fundamentally changed the SDR operating model — they've just made the same processes marginally faster.

Agentic AI is different. And understanding why matters for any B2B organisation thinking about how to build or scale a sales development function.

What makes AI "agentic"?

The term gets used loosely, so it's worth being precise.

A standard AI tool responds to a prompt. You ask it to write an email; it writes an email. The human is still doing the work of identifying the account, finding the trigger, and deciding what to say.

An agentic AI system operates differently. It has:

A defined goal (e.g., surface in-market accounts for a given ICP)
Access to tools it can use autonomously (search, CRM APIs, enrichment databases, news feeds)
The ability to plan and execute multi-step tasks without being prompted at each step
Memory of previous actions and outputs to inform future decisions

In practice, an agentic AI system can be set to monitor a set of 2,000 target accounts, identify the ones showing buying signals today, enrich the contact records, draft personalised outreach for the top 20, and flag them for human review — all without a human initiating or managing each step.

That's not marginal efficiency. That's a structural change to how the intelligence and research work in a sales function gets done.

The new division of labour

Here's what changes when agentic AI handles the intelligence layer of SDR work:

Before agentic AI:

SDR spends 60–70% of time on research, list-building, and personalisation
20–30% on actual outreach execution
10% on follow-up and meeting prep
Output: 30–60 personalised touches per day if they're disciplined

After agentic AI:

AI handles account monitoring, signal surfacing, enrichment, and first-draft outreach generation
SDR spends 80% of time on outreach execution, follow-up, and relationship development
SDR reviews and refines AI-drafted messaging before sending
Output: 100–200 contextually relevant touches per day with consistent quality

But the more important change isn't volume — it's the quality floor. An SDR without AI is only as good as their research habits on a given day. An SDR with a well-built agentic system always starts with context, always has a reason to reach out, and always has a draft that reflects that context.

What the agents actually do

In a mature agentic GTM system, the individual agents have distinct, well-defined roles:

Signal Surfacing Agent Monitors intent data, news, job postings, social activity, technology signals, and ecosystem events across the target account universe. Flags accounts that cross a relevance threshold based on the defined ICP and current campaign priorities. Produces a daily ranked list with evidence for each flag.

Enrichment Agent When a new account is flagged or added to the CRM, the enrichment agent automatically fills gaps in the account record — firmographics, technology install base, key contacts, org structure, known pain points from public sources. Reduces the research burden on the human SDR to near zero.

Outreach Generation Agent Given an account, a signal, and a value proposition, the outreach agent drafts email, call scripts, and LinkedIn messages that are contextually grounded. It references the specific signal that triggered the reach-out, adapts tone based on the recipient's role, and matches the messaging to the relevant pain point.

Campaign Brief Agent When a new campaign needs to be built, the campaign brief agent synthesises ICP data, signal patterns, and competitive intelligence into a ready-to-execute brief: who to target, what message to lead with, what channel sequence to use, and what offer to make.

Meeting Prep Agent Before a booked meeting, generates a full brief for the account executive: stakeholder mapping, known pain points, relevant signals, competitive context, and suggested talking points. Eliminates the 45 minutes of prep that most AEs don't do anyway.

What the human SDR does better

Agentic AI creates enormous leverage — but it doesn't replace the human SDR. The things that agents can't do well are exactly the things that matter most in complex B2B sales:

Handling ambiguity in conversations. A live call requires reading tone, adjusting approach in real time, and knowing when to push and when to step back. That judgment is human.

Building genuine rapport. ERP sales runs on trust and relationships. The SDR who is known in the market, respected by gatekeepers, and trusted by prospects is irreplaceable. Agents can draft the first message. Humans build the relationship.

Making commercial judgments. Should we pursue this account even though the timing is off? Is this a multi-product opportunity or a narrow one? These calls require commercial intuition that LLMs don't have.

Representing the brand in live interactions. Every conversation an SDR has is a brand impression. The quality of that impression is determined by the human, not the AI.

The SDR role doesn't disappear — it evolves. The research, the list-building, the copy, the timing — agents handle those. The judgment, the conversation, the relationship — that's the human's job, and it's the part that actually closes deals.

Implications for GTM leadership

If you're building or managing a sales development function in 2025, the questions worth asking are:

What percentage of my SDRs' time is spent on tasks that an agent could do better? For most teams, it's more than 50%.
Am I hiring for the skills that will be valuable in an agent-augmented model? Communication, commercial judgment, and relationship quality matter more than raw capacity to research and personalise.
Is my data infrastructure ready? Agentic AI is only as good as the data it operates on. Teams with clean, enriched, segmented account data will extract far more value from AI agents than teams whose CRM is a mess.
What does my competitive set look like in 12 months? If the partners you compete with for the same deals are running agentic GTM systems, the volume and relevance advantage they'll have in outreach is significant.

The window to gain a lead on this is closing. But it's still open.

The bottom line for ERP partners

In the ERP partner market specifically, the combination of long sales cycles, complex buying committees, and relationship-driven cultures means that the quality of every interaction matters more than volume.

Agentic AI doesn't just help you do more outreach. It helps you do better outreach — grounded in context, delivered at the right moment, and freed from the low-value research work that burns SDR time and morale.

The model that wins in this market is not the one with the most SDRs. It's the one with the best-informed SDRs, operating with the most relevant market intelligence, at the right accounts, at the right time.

That's what agentic AI makes possible.

← All articles Talk to GTM Layer →