Turning Certified Data Into AI-Powered Decisions

The turning point came when Alan shifted from dumping raw data into AI tools to feeding them filtered, certified datasets exported from Taiga. The distinction made everything work.

‍

You can't say, ‘Here’s a giant bucket of everything that’s happened forever — go figure it out.’ It’ll mix things up, find wrong answers, and ultimately you make bad decisions. The key is that Taiga’s certified datasets are what make AI actually usable.

— John Oakley, CIO & Co-Founder, Taiga

‍

Taiga connects to every system Mach 1 uses — pulling transactions in real time, reconciling discrepancies between systems, and cataloging data against NACS standards. By the time a dataset reaches a certified Taiga report — like the Product Scorecard, Item Summary, or Sales Per Labor Hour — it has been through multiple validation layers. That structured, filtered output is what Alan now feeds into AI.

‍

How Mach 1 Uses Taiga + AI

1. Food Service AI Workflow

Food service is where Mach 1 has built its most refined, repeatable AI process. Mach 1 runs Kate's Kitchen (proprietary grab-and-go) and four Little Caesars programs — with three metrics that matter for every shift: production, sales, and labor.

‍

Here is how the workflow runs today:

‍

• Alan opens Taiga and filters by food service SKU group and time period — one week, one month, or up to two years of history
• He exports the sales data (item, timestamp, revenue, location), the production report from Date Code Genie, and labor timesheets
• In the AI project, a structured instructions file tells the AI exactly what each dataset represents, what Mach 1 is trying to achieve, and who the company is
• A mapping file reconciles item names between Taiga and Dayco Genie — with standard retail prices and shelf lives — so the AI treats the same item consistently across both systems
• The AI generates a formatted daily summary: sales, COGS, waste %, and labor cost by location; 13-period sales calendar; waste by category and item; and labor efficiency by store and day of week
• That summary is distributed to the COO, district managers, and store managers every day or every other day

‍

It took us a series of hours to build and fine-tune. Now, to get a new set of data into the project takes five minutes. And once the AI has absorbed the structured data, I can go deeper — ask about one item, compare two stores, ask what the AI sees that I’m not asking about.

— Alan Meyer, CEO, Mach 1 Stores

‍

2. Labor Optimization

The food service AI workflow has had a direct, measurable impact on how Mach 1 manages labor. Before, reviewing labor data quarterly or semi-annually was already a heavy lift. Now the team reviews monthly and acts decisively on what they find.

‍

We just added 25 hours of labor at a store a couple of days ago — that’s over $20,000 a year in budget. We never would have acted that decisively before. But the AI clearly showed that our labor percentage was below our target range and we had room to invest. It turned what would have been a one-day analysis project into a 10-minute conversation.

— Alan Meyer, CEO, Mach 1 Stores

‍

Alan asks the AI to run labor efficiency for the last four weeks. It calculates labor cost as a percentage of revenue for each hour across the dataset and compares results against Mach 1's defined thresholds — the point to add labor, the point to cut it, and the target range. The output is a clear, store-by-store picture of where to act.

‍

Mach 1 also tracks self-checkout adoption by location as a labor KPI — transactions on self-checkout as a percentage of total transactions, trending over time — using the same process.

‍

3. Managing AI Like a New Employee

Both Alan and John describe the same mental model for working with AI: treat it like training a new employee, not running a Google search. The instructions file that opens each AI project sets the full context — who Mach 1 is, what each dataset contains, and what the AI is expected to produce. That structure is what makes the output reliable.

‍

• Every output is spot-checked: if the AI reports 400 cheeseburgers sold in a month, Alan pulls a specific store and day from Taiga to verify
• When the AI is wrong, Alan corrects it and asks it to remember the fix — building accuracy into the project over time
• John runs Scrum-style standups at the start of each AI session — confirming context is loaded, prior tasks are closed, and the agent is ready
• Error rate in output started at 15–20% in early months; it has been approximately one month since Alan has found a reporting error

‍

AI went from being severely underrated — people using it as a glorified Google search — to completely overrated, where people think it can magically do everything. The truth is somewhere in the middle, and structure is the key.

— Alan Meyer, CEO, Mach 1 Stores

‍

Result:

Alan's 2021 investment in Taiga — made long before AI was mainstream — is now the competitive advantage that makes these workflows possible. The results are concrete and compounding.