Restaurant AI Readiness & Workflow Planner

Use a representative week

Enter the number of events and average manual minutes for each workflow. A phone question, phone order and reservation request should be counted separately. If a call includes an order, count it as an order event rather than counting the same handling time again as a question. For marketing, count a discrete draft or campaign task. For back-office work, count a document or review task with a clear boundary.

Use observed numbers when possible. An estimate is acceptable for planning, but mark that limitation in your exported notes. Zero is a valid answer if the restaurant does not perform the workflow. Do not add hypothetical demand because a product can handle it.

The included sample is fictional and exists only to demonstrate the tool. Loading it replaces current inputs. Reset removes those values. Nothing is uploaded, saved to a server or sent to Loman by this planner.

How the score is calculated

The opportunity score ranges from 0 to 100. It rewards frequency, pain, clear rules and usable data, and subtracts a consequence-of-error penalty. These weights are editorial choices for a first-pilot discussion. They are not trained on restaurant results or validated as a predictive model.

Weekly volume converts to a 0–5 frequency score: zero events = 0; 1–10 = 1; 11–50 = 2; 51–150 = 3; 151–400 = 4; above 400 = 5. Other ratings are entered directly on a 0–5 scale. Formula: (frequency × 5) + (pain × 6) + (rules × 4) + (data × 5) − (risk × 5), clamped between 0 and 100.

Frequency bins compare event counts, not equivalent effort. A long, low-frequency marketing or back-office task may deserve priority despite a lower score. Read manual hours alongside the score rather than treating the ranking as a precise cross-category comparison. Manual hours equal weekly events × minutes per event ÷ 60. This is a description of current workload, not time saved. Actual savings must subtract monitoring, exception handling, setup and correction time. The tool does not calculate revenue, ROI, staff reductions or a purchase budget.

Why the gates matter more than the score

A workflow with no events remains “No current volume.” A score below 40 is “Lower priority.” At 40 or above, the tool identifies preparation work unless the score reaches 65 and every readiness requirement passes. These thresholds are adjustable editorial conventions in the source methodology, not natural cutoffs.

To be a pilot candidate, a workflow needs rules and data scores of at least 3, a risk score of no more than 3, an accountable owner, a tested human fallback and a usable baseline. These are minimum discussion gates, not a complete safety assessment. Risk rated 4 or 5 requires redesign or additional human controls before this tool will label it a pilot candidate.

Candidates are ranked by score, with ties keeping the displayed workflow order. A restaurant may rationally choose a different sequence because of a lease, staffing transition, system migration or other constraint the planner does not know. Record the decision and its reason.

Turn the export into a working brief

The text export includes the input values, workload calculations, gates, score components, next actions and a four-stage pilot outline. The JSON export preserves the same information for later analysis. Keep the exported file wherever your team normally manages operations projects.

Choose one candidate, name the person who will review outcomes and replace the suggested measures with your location’s acceptance thresholds. Do not route live calls, publish a campaign or change a financial record just because a score is high. The planner has no connection to your phone, POS, guest database or other operating systems.