Campaign Agent workflow
How Campaign Agent turns campaign intent into SFMC execution specs.
Campaign Agent is Campaign Builder's planning workflow for moving from an idea, file, or workspace context toward a normalized brief, build package, and Bruno-ready execution handoff for Salesforce Marketing Cloud teams.
Planning workflow
From campaign idea to Bruno-ready build package.
Campaign Agent gives marketers a structured path from early campaign direction to a reviewable execution package. It keeps source context, missing fields, validation warnings, and downstream handoff metadata visible before any SFMC-facing workflow runs.
Capture campaign intent
Campaign Agent accepts typed campaign direction and supported working-file uploads, then uses session context and readable workspace knowledge to avoid planning from an empty prompt.
Normalize the top brief
Top brief mode fills required campaign fields, adds taxonomy and naming metadata, records assumptions, and returns safe grounding references with source title, section, snippet, rationale, and influenced fields.
Build the execution package
The build phase produces Campaign-owned drafts such as email plan details, audience direction, Data Extension plan, journey plan, validation warnings, execution readiness, and a Bruno-compatible campaignSpec.
Hand off to Bruno
Confirmed execution starts with dry-run planning, then Bruno durable execution can create draft artifacts through entitled downstream skills while Journey activation remains manual in Marketing Cloud.
Implementation map
What evaluators can trace to Campaign Agent routes.
The implemented route surface separates brief generation, working-file context, Knowledge saves, image suggestions, feedback, and Bruno campaign execution handoff.
Campaign Agent is mounted under /api/v1/agent/campaign/* and the /api/v1/campaigns/* alias behind authentication and the agent:campaign-agent entitlement.
POST /api/v1/agent/campaign/generate-brief supports ideate, brief, and build phases for candidate directions, required brief completion, and execution-package generation.
POST /api/v1/agent/campaign/context accepts PNG, JPG, GIF, PDF, DOC/DOCX, and XLS/XLSX working files up to 15 MB per file and stores summarized context with a six-hour TTL.
POST /api/v1/agent/campaign/context/:id/knowledge can save approved working context into scoped Knowledge after validation and actor resolution.
POST /api/v1/agent/campaign/execute-campaign passes normalized campaignSpec payloads to Bruno campaign execution, including dry-run-first planning and confirmed draft artifact creation.
POST /api/v1/agent/campaign/image-suggest and POST /api/v1/agent/campaign/feedback support campaign image direction and user feedback without changing the SFMC launch boundary.
Build output
What Campaign Agent hands to downstream SFMC workflows.
The buyer value is not just AI copy. Campaign Agent packages planning decisions in a shape that Email, Audience, Journey, Factory, and Bruno workflows can continue from after review and entitlement checks.
Normalized campaign brief fields for title, business goal, audience benefit, lifecycle stage, supported Email channel type, campaign audience, KPIs, copy, CTA, and email-series details when needed.
Build-package metadata with deterministic taxonomy, naming, assumptions, grounding references, validation warnings, and execution readiness.
Email draft definitions and emailSeriesDraft output for single-email, email-series, or journey-oriented campaign work.
Audience, Data Extension, and Journey plans that downstream Audience, Email, Journey, Factory, or Bruno workflows can continue from.
Capability package metadata that always advertises Email Agent handoff and adds Audience Agent and Journey Agent handoffs for single-email execution mode.
A Bruno-compatible campaignSpec for POST /api/v1/agent/bruno/execute-campaign rather than a hidden production send.
Execution modes
Campaign Agent changes the handoff based on the campaign shape.
The implementation distinguishes a single send from a multi-email series. That matters in a technical evaluation because the surfaced capability package should match the intended SFMC build path instead of advertising every downstream workflow for every campaign.
Single-email execution mode
When the brief resolves to single_email, Campaign Agent can advertise Audience Agent, Journey Agent, and Email Agent handoffs. That gives evaluators a practical path from audience plan to journey notes, email draft work, AMPscript support, and Bruno dry-run execution.
Email-series execution mode
When the brief resolves to email_series, Campaign Agent focuses the capability package on the series itself. The build response carries emailSeriesDraft output and Email Agent capabilities instead of forcing journey structure that still needs marketer review.
Journey-oriented planning
Journey-like briefs can include Data Extension and Journey plans inside the build package, but the page should be evaluated as draft preparation: Journey Builder activation remains a manual Salesforce Marketing Cloud action.
Pilot handoff evidence
What to capture before Campaign Agent work becomes SFMC draft production.
A practical Campaign Agent evaluation should leave a compact handoff packet. That packet connects planning inputs to Bruno dry-run evidence and makes it clear which Factory, Email, Audience, Journey, or Data Extension step will run next.
Brief completeness
Keep the required-field result for title, business goal, audience benefit, lifecycle stage, Email channel, campaign audience, KPIs, copy or CTA, and email-series fields when the campaign needs multiple sends.
Grounded inputs
Save the uploaded-context summary, six-hour working-file reference, readable workspace Knowledge snippets, assumptions, taxonomy, naming metadata, and grounding references that influenced the brief.
Execution package
Review the buildPackage and campaignSpec for email definitions, audience direction, Data Extension plan, journey plan, validation warnings, execution readiness, and advertised downstream handoff targets.
Dry-run handoff
Run the campaignSpec through Bruno with dryRun enabled first, then compare proposed stages, missing inputs, pending write previews, Factory prerequisites, and manual Journey activation warnings before confirming draft creation.
Technical evaluator FAQ
Where Campaign Agent helps, and where SFMC review remains.
Campaign Agent is the planning layer for campaign production. It prepares briefs and execution specs; it does not replace Marketing Cloud approval, send, or Journey activation decisions.
What does Campaign Agent create from a campaign idea?
It can create candidate campaign directions, a normalized top brief, build-package metadata, email draft definitions, audience direction, Data Extension and Journey plans, validation warnings, grounding references, and a Bruno-compatible campaignSpec.
Can uploaded documents shape the brief?
Yes. Campaign Agent can summarize supported image, PDF, Word, and spreadsheet uploads as temporary working context, then combine that context with session details and readable workspace knowledge during planning.
Does Campaign Agent create live Marketing Cloud sends?
No. It prepares reviewed planning and execution handoff artifacts. Confirmed execution moves through Bruno and entitled downstream workflows, while sends and Journey activation remain manual Marketing Cloud steps.
How does Campaign Agent connect to the rest of Campaign Builder?
Build outputs include capability packages and a campaignSpec that downstream Email, Audience, Journey, Factory, and Bruno workflows can use after entitlement checks and review.
Enterprise guardrails
Campaign planning stays grounded, scoped, and draft-oriented.
Campaign Agent connects early planning to execution transformation without claiming automatic production launch. Auth, entitlement checks, safe source references, dry runs, and manual Journey activation boundaries remain visible.
Entitlement-gated planning
Campaign Agent routes require an authenticated user and the campaign-agent entitlement before they can read campaign context, workspace knowledge, or generate build packages.
Safe grounding references
Grounding references expose reviewable metadata such as title, section, document ID, short snippet, rationale, and influenced fields instead of dumping unrestricted source material.
Draft execution boundary
Campaign Agent can prepare Bruno execution specs and draft artifact creation, but Journey Builder activation and final send decisions stay manual inside Salesforce Marketing Cloud.
Evaluation takeaway
Campaign Agent shortens the path from idea to reviewed SFMC build work.
A buyer evaluating Campaign Agent should expect grounded campaign ideation, required-field completion, uploaded context summaries, Knowledge-aware brief generation, build packages, validation warnings, capability handoffs, and Bruno campaignSpec output. They should not expect automatic sends, Journey activation, or hidden Marketing Cloud writes.