Reactive mornings kill strategic work

Most knowledge workers open email and chat first. By mid-morning, someone else's urgency owns the calendar—while the release-critical task sits un touched on a board three clicks away.

AI auto-prioritize tasks does not mean letting a model reorder your entire backlog without context. It means aggregating authoritative queues from project systems and using AI to sort, explain, and draft—while humans start_work on the chosen item.

WKFGo exposes queue and search tools via MCP:

my_day — Personal daily lens: due, overdue, approvals
my_queue — Broader queue view for leads coordinating team work
smart_search — Natural-language discovery across tasks and wiki

Framework: four buckets every queue needs

Design prioritization around buckets humans already understand:

Hard dates — Due today, overdue, client commitments
Blocking others — Approvals and reviews waiting on you
Sprint goal alignment — Items tied to current sprint boundary
Quick clears — Small tasks that reduce cognitive load

AI assigns draft order within buckets; you confirm against sprint goal and stakeholder promises.

my_day vs my_queue

my_day fits individual contributors opening the day:

"Run my_day—list overdue and due-today."
"What approvals block my teammates?"
Pair with start_work when focus begins

my_queue fits people leaders and tech leads who need team-facing escalation without opening every assignee board:

Scan overload patterns before reassigning
Identify approval bottlenecks you can unblock
Feed standup with one query instead of fifteen DMs

Managers still workload_heatmap before pushing more work into someone's queue.

smart_search: prioritization starts with finding work

You cannot prioritize what you cannot find. smart_search collapses scavenger hunts:

"Blocked tasks assigned to me on Project Atlas"
"Open risks mentioned in wiki for payments release"
"Tasks waiting on legal approval"

Search results deep-link to action—comment, reassign, or update_status—inside the same product.

AI-assisted prioritization ritual

Morning (5 minutes)

MCP: my_day + overdue filter
Ask: "Given sprint goal X, top three tasks today?"
Verify IDs against dashboard while building trust
start_work on first item

Midday triage (2 minutes)

New urgent request? smart_search for duplicate work before creating tasks
list_pending_approvals if releases stall

End of day (3 minutes)

end_work or status updates
Draft tomorrow queue from remaining my_day items

Anti-patterns

Personal todo apps that diverge from team board truth
AI priority lists with no task IDs cited
Managers reprioritizing via chat without updating assignments
"Everything is P0" culture—fix with WIP limits and explicit sprint goals

When AI should not prioritize alone

Budget or legal commitments requiring executive sign-off
Cross-project staffing conflicts—use portfolio_overview and human negotiation
Security incidents—runbooks beat model sorting

Escalate with ask_human when requirements are ambiguous.

Prioritization for bilingual and distributed teams

Persian/English teams often lose context in mixed chat threads. smart_search across wiki and tasks in the language the artifact was written reduces duplicate work. AI can draft bilingual standup summaries from my_day output—verify names and dates before posting to Mattermost or team chat integrations.

Async-first teams benefit when priority is written on the board, not implied in a live meeting nobody recorded. Pair queue discipline with publish_meeting_wiki when priorities shift in steering calls. ## Metrics: did prioritization improve? Track proxy metrics monthly:

Overdue count per assignee (should fall or stabilize)
Time from "urgent ping" to task update on board
Standup duration (should shrink when queue is trusted)

AI helps narrate trends; humans interpret whether process—not tooling—is the bottleneck.

Combining my_queue with approvals

list_pending_approvals often hides work that blocks entire releases. Elevate approvals to bucket two in my_queue review—even when they are not "your" IC coding tasks. Unblocking approval is frequently the highest-leverage prioritization move a lead makes on Monday morning.

Prioritization breaks when every stakeholder marks work P0. Define a written rule: maximum three In Progress items per person, and anything new displaces something already on the board—not an invisible backlog. my_queue helps managers see escalation without opening every column. Pair AI narration with a weekly fifteen-minute reprioritization ritual so the queue stays trustworthy.

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.

FAQ — AI task prioritization

Is my_queue the same as my_day?
Related but different scope—individual daily lens vs broader team queue. Many managers use both.

Does smart_search replace the board?
No—it finds and filters; the board remains execution surface.

Can AI create tasks from priority discussion?
After human review, create_task—not autonomously on every chat suggestion.

How do permissions affect queues?
MCP returns only tasks and wiki your role may access—same as the web app.

Try it now

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