Red dashboards are lagging indicators

By the time a milestone tile turns red on a portfolio slide, teams already burned weekends, finance already missed a accrual window, and customer comms already promised a date you cannot defend. Early deviation alerts exist to interrupt that arc—while choices still change outcomes.

This is not another notification channel. WKFGo background monitors watch signals you already record—task aging, burn variance, approval queues, review depth—and when lines cross, they raise decision_inbox items consolidated into a single Executive Digest per cycle. Officers see material forks; teams are not pinged for every wobble.

Problem: alert systems nobody trusts

Common failures:

Threshold spam — fifty emails; humans mute everything
Metric without owner — "velocity down" with no suggested action
Implicit escalation — red tile in a deck, no logged decision
Siloed lenses — finance sees burn; engineering sees queue; nobody synthesizes

AI-assisted monitoring works when it prioritizes executive judgment, not when it pretends to auto-fix delivery.

Framework: signal → threshold → inbox → resolve

Define signals — What predicts slip in your domain? Aging WIP, CFD trend, budget line variance, dependency depth, desk SLA
Set thresholds — Material, not noisy; review quarterly
Raise inbox items — Auto via monitor or manual add_decision_item when humans spot drift early
Resolve with evidence — executive_brief, simulate_scenario, then resolve_decision_item and log_decision

Alerts are inputs to a queue, not conclusions.

Signals worth monitoring

Tasks in review or blocked state beyond team baseline (flow_aging)
Sprint carry-over rate rising two cycles
Finance burn vs plan crossing tolerance
Critical dependency chain with upstream overdue
Pending approvals aging past policy

Signals to handle at team level

Single task estimate miss with clear owner
Normal code review latency within WIP limits
Non-release experimentation spikes

Not every deviation is executive work—protect inbox WIP limits.

Solution: deviation alerts in WKFGo

Configure the monitor mindset

Treat monitors like on-call runbooks:

One digest per cycle—not per-officer duplicate pings
Each inbox item links tasks, finance lines, or report excerpts
Suggested owners and decision question stated plainly

When decision_inbox returns an item, read triggering signal before meeting. Async pre-read cuts live debate time.

Weekly executive triage (30 minutes)

Open decision_inbox — top three by priority
Pull role brief (executive_brief cfo/cto/cpo) for FACTS
For date or scope items, run simulate_scenario before choosing
resolve_decision_item with owner and review date

Close the loop

Resolved items should log_decision forecasts. Next alert cycle calibrates: did last week's cut_scope decision move CFD as expected? decision_log prompts keep monitors honest.

Pair with team visibility

Executives act on alerts; teams fix root causes on the board. Publish monitor thresholds in wiki so engineers understand why a fork escalated—transparency reduces "surprise leadership decisions."

Tuning thresholds without alert fatigue

Start conservative: one or two signals per domain (delivery aging, finance variance). Run a shadow week where monitors write draft inbox items visible only to PMO—calibrate before executives see noise. Compare alerts against what flow_aging and finance_summary already showed manually; merge duplicate root causes.

When a alert fires twice for the same unresolved issue, escalate severity or assign executive sponsor—do not create parallel items. add_decision_item manually when humans spot drift before math catches up; monitors are not the only on-ramp.

Publish threshold definitions in wiki so teams understand escalation is data-driven, not mood-driven. Transparency reduces "leadership panic" narratives that erode trust in otherwise sound monitors.

Anti-patterns

Alert on every overdue task — destroys signal
Auto-resolve without human — hides accountability
Duplicate items for same root cause — merge in inbox review
Ignoring data gaps in briefs when acting on alerts

Alert fatigue kills early warning systems. Prefer decision_inbox consolidation over per-task pings. Define thresholds per project—aging days, finance variance, approval queue depth—and review the digest once daily, not on every micro-threshold cross.

Run a monthly audit: do AI answers cite task IDs from MCP? If not, tighten prompts to require citations before any stakeholder-facing draft leaves the team.

Permission-aware tools mean the same question gets different grounded answers for PM vs contractor—that is a feature, not a bug. Train teams to expect scoped truth.

FAQ

Will I get flooded with notifications?
Executive Digest consolidates officer concerns into one notification per cycle when configured—designed to reduce spam.

Can teams subscribe to lighter alerts?
Use team channels and digests for operational noise; reserve inbox for constraint-changing choices.

Does AI set thresholds automatically?
Humans define material lines; AI summarizes context when items fire.

How is this different from Jira reminders?
Reminders are atomic; deviation alerts tie portfolio signals to decision queue workflow.

Catch drift while it is still a choice

Let monitors raise the fork early; humans resolve with scenarios and logged forecasts.

Try it now

Put these patterns on live project data—not slide decks.