Mise is built around your existing systems. You don't switch POS, you don't migrate documents, you don't re-write your SOPs. Here is exactly what happens, and exactly what we ask of your team, in the first fourteen days.
You connect your POS, your labor system, and your accounting. You upload your operations manuals and food-safety playbooks. You tell Mise which thresholds matter and who should hear about what. From day one, every store has a dashboard that updates with each new sale, shift punch, and audit. From day fourteen, your above-store leaders get a daily morning brief and ad-hoc answers to questions they used to spend an afternoon assembling.
You wire up the systems Mise should read from. Each one is BYOA — bring your own app — meaning you create a developer app inside the vendor's portal and paste a client ID and secret into Mise's settings. Mise gets read-only scopes; you can revoke access from the vendor side at any moment.
Typical day-1 set: Square or Toast (POS), 7shifts or Deputy (labor), QuickBooks (financials), Jolt (food safety). Most are 5–10 minutes of clicking; the longest is whichever one your IT team prefers to handle the OAuth dance for.
What we need from you: ~60 minutesDrop your brand-standard PDFs and DOCX files into Mise's documents page. Operations manuals, opening/closing checklists, food-safety playbooks, training guides, escalation procedures — anything your stores are supposed to be running by.
Mise chunks each document, embeds it, and indexes it for retrieval. From that moment forward, every answer that touches a procedure carries a citation pointing at the exact page of the source you uploaded. The model never makes up a "best practice" — it cites yours.
What we need from you: drag-and-drop, ~15 minutesFor every connection you've granted, Mise pulls the last 24 months of historical data and lands it in a provider-agnostic warehouse — daily sales, shift hours, inventory snapshots, monthly P&L. This gives the rule engine and the reasoning plane real history to ground against from day one, not just last Tuesday.
Backfill runs in the background. You don't wait on it; rules and answers progressively get sharper as more history lands.
Runs in background · 2–6 hours typicalMise ships with a default rule pack tuned for franchise operations — labor variance > 4%, food-safety task miss, sales drop vs. trend, audit grade dropping a letter, and so on. We sit with your team for an hour and tune the thresholds to your brand and your tolerance.
Rules are written in a tiny YAML grammar, versioned in your tenant, and editable from the settings page. Adding a new rule that's specific to your business — "alert if late-night drive-thru count drops 20% week-over-week at any 24-hour store" — is a single config change, not a release.
What we need from you: 1 hour of an ops leader's timeYou import or invite users, set GMs per store, and decide who gets which alerts. Above-store leaders by region, GMs for their store, ops directors for the rollups. Mise's alert engine routes the right thing to the right person — no "everyone gets every notification" channels.
Slack? We connect. Email? We send. Both? Sure. The user picks per category.
What we need from you: ~30 minutesEach morning at 5:30 local, Mise runs a full scan across every store: rules fire, alerts are graded by severity, and a brief is composed. Above-store leaders open Slack to a five-bullet morning brief; GMs see their store's status before they unlock the door.
Through the day, Mise watches each new POS posting, each new shift punch, each new audit submission, and updates the picture. New conditions throw new alerts. Resolved conditions auto-close. Stale ones get nudged.
Continuous · 5:30am scan + live updatesAnyone on your team can open chat and ask in plain English: "Which stores are running labor over plan this week and why?" Mise pulls live numbers from the warehouse, retrieves the relevant section of your labor playbook, and answers with inline citations.
Citations are filtered to only the chunks the model actually used in its answer — not a dragnet of every nearby chunk. If the data isn't there, Mise says so instead of guessing.
Available from day 1, gets sharper over the first 14 daysWhen an alert fires, Mise does not close the loop on its own. It surfaces, names a likely cause, points at the relevant SOP, and assigns it. A human snoozes, acknowledges, or marks resolved. That state is durable: the next morning's brief reflects what's still outstanding, not what's already been handled.
A snoozed alert reappears at the snooze deadline. An unresolved alert ages and escalates. Patterns compound — five "minor" labor variances in two weeks earn a different conversation than the first.
Always: humans decideThe before/after is concrete. Above-store leaders stop opening five tabs in the morning. GMs stop being the last to know that yesterday's labor over-ran. Audit findings stop sitting in a PDF nobody reads. The morning conversation between a multi-unit operator and a GM goes from "let me pull up the report" to "yeah, I saw — what are we doing about it?"
Sales tab open. Labor tab open. Audit portal open. Slack with three GMs. Spreadsheet that aggregates the four. Two hours before the first real decision.
Five bullets in your inbox. Each links to a store, an alert, and a cited source. The first decision happens before coffee.
An alert isn't just "labor over." It's "labor over by 6.4% — peak shift was understaffed Friday — your closing playbook says do X." Cited.
The fastest way to evaluate Mise is to put your own data in front of it. Book a 30-minute call — we'll spin up a sandbox connected to your sandbox POS and labor system, and walk through one full day of the loop. No pitch deck.