Managed AI for Columbus Manufacturers

Most of our people are on the floor. What are we actually paying for?

Seats are per person, and you buy them only for the people whose work happens on a screen. In a machine shop that is the office: estimating, purchasing, scheduling, quality, HR, customer service and the owner. A machinist running a job all day does not need a seat and should not show up on the invoice. The seat count comes off the office roster.

Billing is per seat plus usage. The allowance rides on the seat and renews monthly, and the ceiling on total spend is a number you set, so the invoice cannot run away from you between months. Nothing here is flat and nothing is unlimited.

Budget is allocated by team and department, never by individual person. Estimating gets a larger allowance because estimating does the heavy document work. HR gets a smaller one. If purchasing runs through its allowance in week three because a supplier follow-up workflow took hold, you move budget between departments. Nobody is metered as a person.

The published ranges are monthly totals for the whole account. Up to 10 seats runs $250 to $500 a month. Ten to 25 seats runs $500 to $1,175. Twenty-five to 100 seats runs $1,175 to $3,800. A $250 monthly minimum applies, and we do not publish a per-seat rate.

We run ITAR work. What has to stay out of this?

Export-controlled material is kept out through a scoping session and a written policy, not a technical filter, and that line gets drawn before the first seat is turned on. If you hold technical data under ITAR or EAR, or Controlled Unclassified Information that reached you under DFARS 252.204-7012, that material carries a defined handling boundary and a general AI workspace sits outside it. Drawings from a defense prime, process data tied to a controlled part, anything a customer marked and flowed down to you. Those are named in the policy before anyone logs in.

The practical version is a scoping session. Which mailboxes receive controlled material. Which shares hold it. Which estimator works those jobs and which one never does. The line gets written down, the workspace is configured around it, and the people who touch defense work are told in plain words what they may paste and what they may not.

What remains is most of the week’s office work. Commercial RFQs, purchasing, HR, safety documentation, customer service email, and the quality paperwork that is not tied to a controlled job. Anything sitting close to the line gets settled in that scoping session and written down before it is used.

Our sibling company TTS Cyber works from the same Columbus address and handles the managed IT and cybersecurity side. If your scoping question turns out to be a compliance question, we will say so and walk it across the hall rather than answer it here.

Why not just buy ChatGPT or Copilot and skip the middle?

You can, and the first few things you build will work. The cost shows up later. A workflow built straight onto one AI company’s model carries that company’s pricing and its release schedule with it. When the price moves or that model version is retired, the workflow gets rebuilt, and the rebuild lands on whoever wrote it, usually the one person in the office who already has a full job.

ScaleSight AI sits between your workspace and the models. When a model is repriced or retired, we re-point the workflow and it keeps running. The estimator who opens the RFQ intake template every morning does not find out anything changed.

The second half is routing. No single AI company leads at every task. One model reads a long customer quality manual better than the rest. One does live web research on a supplier you have never bought from. One drafts bulk text cheaply, which matters when a format change means redoing a stack of work instructions. Each job goes to whichever model handles it best, and the person using it never has to know which one answered.

What shows up in the workspace that we did not build ourselves?

Work that is the same in every business gets built once and given to every client. A purchase order follow-up is a purchase order follow-up whether the part is a machined bracket or a printed carton. That is the library. A shop that starts in March inherits everything built before March, so the workspace is not an empty screen on day one.

Anything built on your own documents, your own data or your specific process stays in your workspace and is never shared. Your quote history, your travelers, your customer’s quality manual, the pricing judgment your estimators have built up over the years. Those stay where they are and never enter the library for another client to receive.

New templates, agents and workflows ship into the workspace every month out of that shared library. The library grows as more general work is built into it, so the workspace holds more in month twelve than it did in month one, at the same allowance.

Work that is specific to your shop is where AI Build Projects come in: scoped work with a start date and an end date, delivered into your workspace. An intake workflow that reads the particular drawing formats your three biggest customers send is a build project. The generic supplier follow-up is a library item.

How do we know what people are actually doing with it?

You get reporting on what kinds of work AI is being used for, broken out by team. It does not include transcripts of conversations or productivity scores on individuals. The view is categorical: which departments are using it and for which jobs. That tells you whether estimating picked up the RFQ workflow or quietly went back to the old way.

Governance is that same reporting read a second way. If somebody starts drafting customer-facing commitments in a workspace set up for internal documentation, you hear it from a report before you hear it from the customer. A rule about what AI is used for holds only when someone can see whether it is being followed.

Collaboration is the other half. When your buyer works out a sharper way to phrase a late-delivery escalation, that phrasing gets built into a template and everyone else in your workspace has it. A method that lived in one head now sits in the workflow every buyer opens. It survives two weeks of vacation, and it survives a retirement.

The target for year one is plain. In a year, the best AI ideas in your company should come from your people, not from us. The ceiling is there so nobody is afraid to try something. The reporting is there so you can see what worked. The library is there so nobody rebuilds what already exists.