Agentic marketing: from automation to autonomy

In short: Marketing automation follows the rules you write. Agentic marketing pursues the goals you set. AI agents build audiences, draft content, launch journeys and optimise spend, all within guardrails your team defines. Here is what changes, what stays human and how to start.

What is agentic marketing?

Agentic marketing means using AI agents that can plan and carry out marketing work on their own, not just generate text when asked. Salesforce describes it as marketers setting the strategy (for example, increasing loyalty among at-risk customers) while agents handle the execution: building audiences, personalising content, optimising performance, answering customer questions and handing work over between teams and channels.

The difference from classic marketing automation is the starting point:

• Automation is rule-based. A marketer designs every step: if a contact opens the email, wait two days, then send the next one.
• Agentic marketing is goal-based. You define the objective, the audience boundaries, the brand rules and the budget. The agent proposes the next best action, takes it where allowed, and learns from the results along the way.

Why now?

The pieces have finally come together: language models that can reason, unified customer data that agents can act on, and platforms that put guardrails around both. Adoption is still early, which is exactly why it is worth paying attention now:

• 75% of marketing organisations already use AI in at least one form, but only 13% of AI-using marketers tap into self-directed agents (Salesforce, State of Marketing, 10th edition, 2026).
• Marketers expect AI agents to save them up to eight hours a week. That is a full working day.
• Gartner predicts that by 2028, 60% of brands will use agentic AI to facilitate streamlined one-to-one interactions.

What agents actually do in marketing

Campaign creation

Describe a goal in plain language and the agent drafts a campaign brief, recommends audiences, writes email and SMS variants and sets up the flow for your review. In Salesforce, this is the job of Agentforce Campaign Creation in Marketing Cloud Next.

Segmentation without SQL

“Customers who bought twice last year but haven’t opened an email in 90 days” becomes a ready-to-use segment on unified data, with no query writing and no ticket to the data team.

Personalisation and journeys

Agents adapt content, product recommendations and the next step in a journey based on real-time signals from email, web and service interactions, always within the strategy and guardrails you set.

Paid media and performance

Agents compare ad performance against industry or custom benchmarks, flag underperforming ads and, in supported channels, pause them once a marketer approves. In Salesforce, this is the Agentforce Paid Media Optimization Agent in Marketing Intelligence.

Two-way conversations

Instead of one-way broadcasts, agents can respond to replies, qualify leads and hand warm prospects over to sales. According to the same Salesforce report, 83% of marketers agree that customers want two-way interactions.

What stays human

Agents are fast executors, not strategists. Your team still owns the positioning, the creative idea, the brand voice and the decision about what is worth saying at all. The best setups make the split explicit: humans set direction and approve what goes out, agents execute and report back. Transparency matters too. In a Gartner consumer survey, 78% of respondents rated clear labelling of AI-generated content as very important for trust.

Data is the real prerequisite

An agent can only personalise what it can see. Yet just 26% of marketers say they are fully satisfied with how their data is unified. If customer data sits in separate email, CRM, e-commerce and service systems, agents will make confident decisions based on half the picture. Unifying profiles in Data 360 (formerly Data Cloud), or at least cleaning up the CRM you already have, has to come before any agent rollout. If you are still choosing between Salesforce marketing tools, start with Salesforce Marketing Automation: Pick the Right Tool.

How to start: a 30-day pilot

1. Pick one measurable goal, for example reactivating lapsed customers or speeding up lead follow-up.
2. Check the data. Make sure consent, contact data and the behavioural signals that matter for that goal are in one place.
3. Write down the guardrails: brand voice, claims you never make, budget and frequency limits, and which steps need human approval.
4. Run it side by side. Let the agent work next to your current process and compare the results.
5. Scale what works. Move on to the next journey only when the first one beats the baseline.

Where Claudeforce fits

On Monday we looked at Claudeforce, the Salesforce and Anthropic partnership that brings Salesforce into Claude. Sales comes first, but Salesforce has already announced marketing capabilities for campaign automation and personalisation. The direction is clear: marketers will increasingly brief agents in plain language and review their work, rather than building every step by hand.

How Cumultec can help

Cumultec implements Salesforce Marketing Cloud, Data 360 and Agentforce for companies across the Baltics, the Nordics and wider Europe. We can help you find the first agentic use case worth piloting, get your data ready and set guardrails your brand and legal teams are comfortable with. Get in touch to plan your first agent-led campaign.