Advisor guidance
How Advisor turns account knowledge into campaign recommendations.
Advisor is Campaign Builder's account-grounded guidance layer. It helps teams review campaign strategy, retrieve relevant context, save useful recommendations, and decide next steps before execution moves through Bruno, Factory, or specialist SFMC workflows.
Guidance workflow
From marketing question to grounded next step.
Advisor works best when campaign teams need strategy review, source-aware recommendations, or reusable guidance before Bruno and Factory turn approved direction into SFMC draft artifacts.
Ground the question
Advisor chat embeds the user's prompt, searches the account-scoped knowledge corpus, applies readable tenant and workspace filters, and returns answers with snippet IDs when retrieved content is used.
Bring in account voice
The Advisor prompt can include account tone-of-voice settings, subject taxonomy, session history, uploaded context, and prior feedback signals so recommendations reflect the workspace instead of generic campaign advice.
Save useful guidance
Teams can save Advisor outputs as account-scoped advices, list recent saved guidance, and reload session histories when they need to continue a strategy discussion.
Handle long-running work
When async Advisor jobs are enabled, chat requests can be queued, polled by job ID, and returned with durable status instead of forcing every review into a single synchronous request.
Buyer use cases
Where Advisor fits before Salesforce Marketing Cloud execution.
Advisor is most valuable before production-bound work begins: it helps teams inspect strategy, retrieve account context, preserve useful recommendations, and decide which Campaign Builder workflow should take over next.
Review campaign strategy before production
Use Advisor when a team needs a source-aware second pass on audience fit, message hierarchy, channel direction, or next steps before Bruno turns approved direction into specialist execution work.
Answer questions from account knowledge
Advisor can retrieve readable workspace snippets and return snippet IDs, so evaluators can inspect which account context shaped a recommendation before relying on it in a campaign brief.
Work from temporary campaign notes
Uploaded or pasted context can support a specific strategy discussion with user-scoped records and expiry metadata, without making those notes permanent global knowledge or SFMC launch permission.
Preserve reusable guidance
Saved advices, session history, feedback, and async job status help teams keep useful recommendations available for later Campaign, Content, Email, Journey, localization, or Factory work.
Implementation map
What evaluators can trace to Advisor routes.
Advisor is implemented as an authenticated route family under Campaign Builder's agent API. It supports grounded chat and reusable advice, not direct Marketing Cloud launch automation.
POST /api/v1/agent/advisor/chat supports grounded chat, optional session history hydration, grounding limits, feedback hints, and async job enqueue when configured.
GET /api/v1/agent/advisor/snippet/:id fetches workspace-scoped source snippets for previews without exposing unrelated knowledge records.
GET /api/v1/agent/advisor/advices and GET /session/:sessionId return saved guidance and account-scoped conversation history.
POST /api/v1/agent/advisor/feedback records thumbs up or down feedback so later Advisor turns can adjust tone, format, or specificity.
GET, POST, and DELETE /api/v1/agent/advisor/context routes manage temporary account-scoped uploaded context for the authenticated Advisor user.
GET /api/v1/agent/advisor/jobs/:jobId returns queued, completed, failed, or dead Advisor job state for the authenticated account.
Uploaded context lifecycle
Temporary context helps Advisor review the real campaign in front of the team.
Advisor can work from uploaded or pasted campaign context without turning those notes into global knowledge. The context routes are authenticated, account-scoped, user-scoped, and separate from SFMC write execution.
Add extra campaign context
POST /api/v1/agent/advisor/context stores text context with filename, mimetype, modality, size, summary, sessionId, optional meta, createdAt, and expiresAt fields for the authenticated account and user.
Keep context user-scoped
GET /api/v1/agent/advisor/context returns the current user's non-expired Advisor context records only after account and user context are present.
Remove context cleanly
DELETE /api/v1/agent/advisor/context/:id deletes only matching Advisor context records for the same account and user, so uploaded strategy notes do not become shared launch permissions.
Advisor evaluation proof
What a useful Advisor pilot should prove before execution begins.
Advisor should make campaign decisions easier to inspect before production work starts. A credible pilot leaves evidence for grounding, saved recommendations, temporary context handling, feedback, and async status without implying direct SFMC changes.
Grounded answer evidence
Ask Advisor to answer from workspace knowledge and confirm the response includes snippet IDs, source previews from /snippet/:id, and a clear distinction between retrieved context and model recommendations.
Reusable recommendation evidence
Save a useful recommendation, reload the advices list, and reopen the session history to verify that guidance can be reused by the account without becoming an SFMC write action.
Campaign-specific context evidence
Add temporary context for a brief or campaign note, verify it appears only for the authenticated user, then delete it before moving the approved direction into Bruno, Factory, Email, Content, or Journey workflows.
Feedback and async evidence
Record thumbs up or down feedback and, where ADVISOR_ASYNC_ENABLED is configured, poll the Advisor job route so long-running strategy reviews return durable status instead of silent timeout behavior.
Technical evaluator FAQ
Where Advisor helps, and where execution still needs review.
The safest evaluation lens is simple: Advisor improves campaign decisions and preserves context, while SFMC write actions remain in confirmed specialist workflows.
Is Advisor a replacement for campaign approval?
No. Advisor provides account-grounded recommendations and next-step support. It does not approve sends, activate journeys, or bypass the reviewed SFMC write boundaries used by Bruno, Factory, Email, Content, and Image workflows.
What makes Advisor guidance account-grounded?
Advisor uses authenticated account context, workspace-readable knowledge filters, retrieved snippets, subject taxonomy, optional uploaded context, account tone-of-voice settings, and session history when building its response.
Can teams preserve useful recommendations?
Yes. Advisor routes support saved advices, session history lookup, snippet preview, and feedback capture so a recommendation can become reusable campaign guidance rather than a one-off chat answer.
How should technical evaluators verify Advisor safely?
Check that Advisor responses stay account-scoped, include snippet IDs when grounded content is cited, preserve request and session context, and do not claim direct SFMC writes or journey activation.
Enterprise guardrails
Guidance stays scoped before execution begins.
Advisor adds decision support to Campaign Builder's SFMC execution layer, but it keeps authentication, account scope, workspace-readable knowledge, and production write boundaries clear.
Authenticated account scope
Advisor routes require authenticated user and account context before chat, snippets, saved advice, sessions, feedback, context, or job status can be read or written.
Workspace-readable knowledge
Snippet lookup and grounding use tenant and workspace filters so Advisor recommendations draw from readable knowledge rather than every stored document.
Guidance, not production writes
Advisor is a support layer for campaign strategy, review, and next steps. Confirmed SFMC writes remain in the specialist execution paths that preview and scope production-bound changes.
Evaluation takeaway
Advisor helps teams decide what to execute next.
A buyer evaluating Advisor should expect account-grounded campaign recommendations, saved guidance, snippet-backed context, feedback capture, and async support where configured. They should not expect Advisor to activate journeys, send campaigns, or bypass confirmed SFMC write paths.