Griffith Consulting Group

working with ai

I learn your business before I bring in AI

I start where your work happens. I watch how jobs move, where they wait and who does them by hand. Once I understand your system, I can tell you where AI fits, and I build it in there.

the slow stepAI
fig. 1 · find the slow step, then add AI

System first, AI second

I start with your process. I can't improve a system I haven't watched run.

  • I learn the system first. Gemba means going to the place where the work happens. I sit with you and your team, follow the work and find the slow step.
  • AI goes only where it fits. Some steps need a person. Some need a checklist or a reminder. I use AI on the steps a machine handles well, such as sorting messages or drafting follow-ups.
  • I tell you what AI can't fix. I'm always of the mindset of under promise rather than over promise. If the problem is a broken handoff or a staffing gap, I'll say so up front.

How a project runs

  1. Walk the process

    I come to where the work happens and follow one job from start to finish. I note each step, who does it and where it waits.

  2. Measure the slow step

    I put numbers on the process: how long each step takes and how often it repeats. You see the same data, so we agree where to start.

  3. Build the change

    I build the fix for the slow step. Sometimes that's AI. Sometimes it's an automation or a cleaner spreadsheet. I keep it small enough to test.

  4. Test and hand over

    We run the change on real work next to your current way. Once it holds up, I hand it over with notes your team can follow.

Where I learned this

I've built AI with real clients, written code inside an airline and trained as an industrial engineer.

  • I built AI with real clients at Actual Intelligence Labs. As CTO, I developed my AI engineering skills there. I looked at businesses through an AI lens and found what to optimize.
  • I brought Python to a Delta Air Lines department. Our tool cut the bid shell build from 5 to 8 minutes to 5 to 10 seconds. I sorted complaint data into 20+ categories with machine learning.
  • I'm an industrial engineer trained in Lean. I earned a B.S. in Industrial Engineering from Georgia Tech in spring 2026 and my Lean Green Belt in June 2025.
Portrait of the founder of Griffith Consulting Group

Questions owners ask about AI

Have another one? Ask it on a call.

Do I need AI at all?

Maybe not. I look at your process first. If a reminder, a checklist or a cleaner spreadsheet fixes the slow step, I'll tell you and build that instead. I bring in AI only where it fits.

Will AI replace my staff?

I build AI to take repetitive work off your team, such as sorting messages or sending follow-ups. Your people keep the work that needs their judgment. You decide who does what.

What does AI see of my data?

Before I build anything, I show you which data the tool reads and where that data goes. I use only what the job needs, and nothing moves without your OK.

Where do we start?

We start with a call and a walk through your process. I find the slow step, show you the data, and tell you up front what I can fix and what I can't.