AI Scheduler

Describe your week in plain English and let Timely draft the schedule.

The AI Scheduler takes a plain-language description of what you need and returns a full week of shifts that respects your roster, time off, and rules. Useful when you're starting from a blank week or want a coverage proposal to edit down from.

Available on all plans during the trial. After the trial: 4 runs/month on Pro, unlimited on Business+.

Where to find it

Two surfaces:

  • Schedule grid (/dashboard/schedule) → click AI Scheduler in the top-right toolbar.
  • Onboarding wizard — step 4 of the first-time setup flow has an AI scheduler step that lets new users see a draft of their first week without manually placing shifts.
  • Mobile — same icon as a floating action button on the bottom-right.

What you tell it

A free-form text box. The simpler your prompt, the better the output. Examples that work:

  • "I need 3 closers Friday and Saturday evening. Maya should not work over 30 hours."
  • "Mike is the only one trained for opening. Schedule him every weekday morning. Rotate weekend openings between Sarah, Ravi, and Tara."
  • "Standard week: 4 people on lunch shift, 5 on dinner shift, 2 closers. No one over 35 hours."

Things that don't help:

  • Ultra-specific times like "9:00 AM exactly" — let the model pick reasonable defaults; you can edit afterward.
  • Asking for things outside the AI's scope ("set Sarah's wage to $20") — the AI only generates schedules, not employee data.

What it has access to

When you click Generate, the AI sees:

  • Your roster — names, employment types, tags, hours-per-week limits
  • Approved time off for the week — these days are excluded automatically
  • Your schedule rules — generated shifts respect both WARN and BLOCK rules
  • Your tags — used for coverage hints ("3 closers" maps to employees with the Closer tag)
  • Hours of operation at the location — shifts won't be generated outside business hours
  • The current week's existing shifts — by default the AI starts from a blank week, but you can ask it to "extend the current schedule" to fill gaps

It does not see:

  • Pay rates (so prompts about cost optimization won't work today)
  • Past weeks beyond a 1-week lookback (no historical pattern recognition)
  • Conversations from prior runs

What it gives back

A draft schedule populated in the grid. The AI returns its reasoning in a panel below the grid — which employees got which shifts and why. You can:

  • Accept — saves the draft as-is. Triggers a normal save (rules evaluate).
  • Tweak — click any generated shift to edit it directly. Edits don't lose the AI reasoning.
  • Regenerate — go back to the prompt, refine it, run again. Counts as a separate run against your monthly cap.

If a generated shift would BLOCK on your rules, the AI catches it and either rearranges or surfaces "I couldn't satisfy this constraint, here's the closest valid schedule."

When it works well

  • Stable rosters with clear roles — the more your tags (Closer, Opener, Lead) match your operational vocabulary, the better the output.
  • Clear constraints — "no one over 35 hours" + "Mike opens weekdays" gives the model concrete walls to plan around.
  • Starting from blank — generating a week from scratch is what it's tuned for.

When it works less well

  • Highly idiosyncratic preferences — "Tara prefers afternoons unless Mike's also working, then morning" — multi-conditional preferences are hard to express precisely.
  • Custom domain logic that isn't in tags — e.g., "only certified bartenders can pour after 9pm" — encode the certification as a tag and a rule first.

Cost

There's no per-run charge to your Timely subscription. AI Scheduler usage is tracked under your plan's monthly allowance — Business and Enterprise are unlimited; Pro is capped at 4 runs/month.

Privacy

Your prompt, roster, tags, and rules are sent to Timely's AI processor only for the duration of the run. Sensitive row-level data — pay rates, contact info, internal notes — is never included in the prompt.

Roadmap

  • Pattern recognition from past weeks — propose a schedule based on what actually worked last quarter
  • Conversation mode — refine a generated schedule with follow-up prompts ("move Sarah to closing instead")
  • Cost-aware scheduling — minimize labor cost given coverage constraints

These aren't shipped yet but are on the near-term backlog.

Found a typo or something missing? Let us know.