Live monitoring and AI recovery for Salesforce Marketing Cloud.
An automation fails at 2am. The send window was 6am. Nobody finds out until 9, and by then the answer to “what happened to yesterday's campaign” is a shrug and a screenshot.
Daedalus watches what's running across every business unit, auto-recovers what breaks, and escalates what it can't, so most routine failures fix themselves before anyone notices. No vendor sells this, the gap costs a send window, and one person can own it. That made it a straightforward build.
SFMC tells you what's defined. Not what's running.
Salesforce Marketing Cloud is the engine — but the operator view it ships with is a pile of definitions, not a live dashboard. Three things keep ops teams up at night:
Silent failures.
An automation fails at 3am and you find out when you log in the morning. SFMC's native UI shows you what's defined, not what's running.
No cross-BU view.
Fifty client business units, fifty separate logins. Agencies stitch health together by tab-switching all morning. Nothing tells you which BU is on fire.
Recovery is manual.
Reruns are a click in the UI, then a wait, then another click. If the fix is non-obvious, an engineer goes spelunking through error codes.
Three layers. No magic.
Connect to your SFMC environment, observe what's actually running, act on what breaks.
Connect
Daedalus installs as a Salesforce Marketing Cloud Installed Package plus a small CloudPage helper. No agents to deploy, no schemas to migrate, no automations to rewrite. It listens for what's actually running and only surfaces what's relevant.
Observe
Live activity for every automation that's run recently — across every business unit. Run history, durations, failure traces. The dashboard the platform never gave you.
Act
When an automation fails, Daedalus queues a rerun automatically. If it fails twice, an AI agent classifies the error, generates a targeted fix, and applies it to the retry — without touching the production automation. You review the diff and promote when you're ready. Or set it to autonomous and it promotes itself.
Real product. Maps to the gaps.
Each capability closes one of the three gaps above:
- Live automation activity across every business unit — runs, durations, statuses, failure traces(→ silent failures)
- Cross-tenant dashboard — every BU you own or manage in one view, sorted by what needs attention(→ no cross-BU view)
- AI recovery — auto-rerun on first failure, AI-classified fix applied on the second, full diff for human review(→ recovery is manual)
- Incident timeline + history per BU — what broke, when, what fixed it, how long it took
- BU-level health scoring — a single number for whether a BU is humming or on fire