FAQ
The questions you should be asking any AI vendor
Straight answers, including the ones that don't flatter us. If yours isn't here, email [email protected]. A person replies within one business day.
Is my data safe? Where does it actually go?
It goes nowhere. That's the whole architecture. The models run on a machine inside your building, and your drawings, orders, quotes, and financials never cross your firewall. There's no cloud account holding your data, no third-party API reading your documents, no terms-of-service clause about 'improving our services' with your files. Unplug the internet and the system still runs. We can't lose, sell, or train on your data, because we never have it.
How is this different from just using ChatGPT?
Three ways. Location: ChatGPT runs in someone else's cloud, and everything you paste goes with it; our models run on your hardware. Integration: ChatGPT doesn't know your ERP exists; we build systems wired into your ERP and MES that read your actual POs and post drafts into your actual workflows. Ownership: with a chatbot subscription you own nothing and pay forever; with us, you own the code, the setup, and the hardware. No per-token bills, no monthly fee to keep your own tools running.
What hardware do we need?
Usually a single server-class machine with a capable GPU, sitting in your server closet or under a desk. A one-time purchase, not a fleet. The exact spec depends on what we're building, so we size it during the first conversations, help you buy it (you own it, so it goes on your books, not ours), and set it up. You don't need a data center. You need one good box.
What does it cost?
It's a scoped consulting engagement, not a subscription: a project price for the work, plus the hardware you buy directly. No license fees on what we build, no per-token charges, no per-seat pricing, and nothing that keeps billing you after the work is done. Ongoing support is optional and separate. We're pre-launch and pricing engagements individually, so we won't publish a number that pretends to be one-size-fits-all. You'll have a real quote after we've scoped your project, and the meter never runs on your own tools.
We already have an ERP. Do we have to replace it?
No. And if a vendor tells you to rip out your ERP, hang up. We build around what you run. Your ERP stays the system of record; the AI reads documents, drafts entries and quotes, and posts results into it for your team to approve. ERP and MES integration is our specialty, including the old, crusty, heavily customized systems that actually run most shops.
We don't have an IT team. Is that a problem?
It's normal, and we plan for it. We handle the setup, build for low maintenance, and document everything in plain language: what the system does, how to restart it, who to call. Day to day it needs about as much attention as your file server. If something's beyond whoever watches your computers now, call us, or hand the docs and source code to any local IT firm. Nothing we build requires us to operate.
What happens if Mikimo disappears?
You keep everything, and everything keeps working. You own the source code, the documentation, and the hardware; the models run locally with no dependency on our servers, our accounts, or our existence. Any competent developer can maintain or extend what we built, because there's no black box and no proprietary lock. We plan to be around, but you shouldn't have to bet your operation on any vendor's longevity, including ours. That's why nothing we build depends on us existing.
How long does an engagement take?
We won't quote a timeline before we've seen your process and your data. Anyone who does is guessing. What we can tell you is the shape: pilots are scoped small, to one process, so you see results on your own documents before committing to more. After the first conversation you'll have a written plan with real dates for your project. We'd rather give you an honest schedule than a fast one we'll miss.
Will this replace people?
It replaces retyping, not people. The systems we build draft things: a quote, an ERP entry, a report. A person on your team approves before anything counts. Every shop we talk to has more work than hands; the point is getting your estimator estimating and your ops manager managing instead of keying in part numbers. We don't build fully autonomous systems, on principle. In manufacturing, one unreviewed bad quote costs more than the typing it saved.
Why should we trust a company with no case studies?
You shouldn't trust anyone on logos alone, and we won't fake ours. We're pre-launch, onboarding our first businesses now, and every claim on this site is about how we work, not results we haven't earned: your data stays in your building, you own 100% of the code, a human approves every output, and we answer email within a business day. The pilot structure exists so you can verify us on your own data, at small scale, before betting anything big.
Still weighing it? Ask us the hard one directly.
Talk to a person