The demo built a campaign from one sentence. Yours will produce nonsense. Not because the technology is bad — because an agent that has never been told your brand voice, your approved claims or your audience logic can only guess at them. Salesforce now calls the whole marketing suite Agentforce Marketing, and the agents are genuinely useful. But they are useful the way a talented new hire is useful: after onboarding, not before.
That onboarding — grounding, templates, guardrails, enablement — is the actual implementation work. It is what this service does.
What the marketing agents actually do (once grounded)
- Campaign Creation agent — turns a brief into a proposed audience, content set and campaign structure. Quality tracks the quality of the brief, which is why brief templates matter more than the agent itself.
- Content Builder agent — generates email, landing-page and WhatsApp content inside the editors. On-brand only if your tone, claims and taboos are written down where it can use them.
- Segment creation — builds audiences from natural-language descriptions via Einstein. Only as good as the data model underneath.
- Optimization & performance — answers performance questions and suggests adjustments; since mid-2026 also from Slack. The newest of the four, and the one to trust last.
Where to start, by situation
| If this is you | Start here | From |
|---|---|---|
| Not sure your org is ready for AI at all | AI Readiness Score — 10 questions, in your browser | Free |
| Want the readiness answer in writing, with a business case | MCN Readiness & business case — fee credited against any follow-on package within 90 days | €3,200 |
| Agents are on, but the output is generic | Agentforce Marketing Enablement | €6,100 |
| Team needs prompts that actually work, in German and English | Agentforce Campaign Prompt Library | €1,400 |
| One agentic use case, live and guardrailed | Agentforce Marketing Enablement + scoped pilot | from €6,100 |
What "implementation" actually means
Five concrete workstreams. No platform re-architecture, no twelve-week discovery phase.
- Activation & permissions. Einstein generative AI and Agentforce enabled, agent instances created from templates, permission sets scoped so the right people — and only the right people — can use them.
- Brand-voice grounding. Your tone, terminology, forbidden phrases and approved claims, formalized into the context the agents actually read. If it only exists in your head today, this is where it gets written down.
- Brief & prompt template library. Campaign brief templates tuned to your objectives, audiences and offers, plus prompt templates per use case — launch, event, reactivation, upsell — so nobody starts from a blank box.
- Guardrails & approval gates. Which outputs ship without review, which require a human, what happens with claims the agent must never make. Written rules, configured where possible, trained where not.
- Adoption workshop. A half day with your marketers on the real system with the real templates — because an agent nobody trusts is an expensive toggle.
What you need before agents work
You do not need everything on this list — missing pieces are part of the project, not a disqualifier. But it is honest to show the list up front:
The realistic prerequisites
- Marketing Cloud Next (Growth or Advanced) or Account Engagement — on AE, the path runs through Account Engagement+, which adds the agentic layer without a migration.
- Einstein generative AI + Flex Credits — the licensing layer the agents run on. I will tell you what your edition includes before you buy anything.
- Brand guidelines that exist somewhere — a slide deck is fine; formalizing them is workstream 2.
- A named human reviewer — someone who approves agent output in the early weeks. Non-negotiable, and cheaper than the alternative.
Why there are no “5× faster” numbers on this page
Competitor pages promise 5× faster campaigns and +35% engagement from AI agents. Nobody has honest numbers for this yet — the platform is barely a year old and every org starts from a different baseline. What I can promise is narrower and checkable: agents that produce on-brand output, rules for when a human looks first, and a team that actually uses them. The speed follows from that; the made-up percentages do not. My position on shipping agents safely is in grounding & guardrails.
Read before you buy
- Agentforce marketing agents: what they actually do, and where they help
- What Agentforce actually costs: Flex Credits and the honest math
- Grounding and guardrails: making an agent safe enough to ship
- Free check: should an agent really do this job? — scores one specific use case
- Free check: AI Readiness Score — scores your org’s foundations
Questions teams actually ask
How this engagement works
Fixed scope, fixed price: Agentforce Marketing Enablement at €6,100 covers all five workstreams with a defined cap — brief templates, prompt templates, one brand voice. The Prompt Library at €1,400 attaches to it (or stands alone) with twelve months of updates, in German and English. No retainer required, no minimum term.
The person you meet on the first call is the person who configures the agents, writes the prompts and runs the workshop. No handoff, no juniors.
Sound like your org?
Thirty minutes, free, no slides. Bring your questions about scope, timing or whether this package is even the right one — you will get straight answers.