1. The code stopped being the expensive part
For many manufacturers, custom software used to be out of reach. You bought an ERP built for thousands of companies and bent your work to fit its screens. The median ERP project in Panorama Consulting's 2025 report cost $450,000 and took nine months (median means the middle value in the group).
AI now writes much of the code, and a person reviews what it builds. The tool that does the writing is a coding agent, an AI that writes and tests code on its own for long stretches. In 2026, these agents work in sessions that run for hours, and Anthropic bills its hosted agents by the session-hour. People still guide the agent and decide what is good enough to use.
A digital receiving form is now a small build, and so is a review screen that fits the way your quality lead works. We think neither needs a project the size of an ERP rollout anymore.
2. The hard part is what lives in people's heads
When the code is cheap, the cost moves to the question the code has to answer: what exactly is the job?
At many companies that have run for decades, nobody has written the answer down. People do the job the way they learned it, and they get it done under pressure. Managers often think they know the process until they sit with the person who does it and find steps nobody wrote down. The spreadsheet has forty columns, and nobody can say why half of them exist. Some are there because somebody left and nobody cleaned up after them.
We call the work of fixing this the clarity bill: the time and attention it takes to get a job out of heads and off paper into steps a new person, or a piece of software, could follow. Software got cheap. The clarity bill did not, and nobody can pay it for you.
3. What clarity looks like: a decision map
Clarity sounds vague until you write it down. What you write, one job at a time, is a decision map. It lists each step in order and marks where software may act and where a person must check the work.
Take a customer email asking for a certificate of conformance. A decision map for it answers three kinds of questions:
- What are the steps? Read the request, find the order, pull the part, revision, quantity, and heat from the job record, fill in the template, and put the draft in the review queue.
- Where does a person verify? The quality lead checks every heat number against the mill cert and signs before anything goes to the customer.
- What happens when the answer is unclear? If the heat on the mill cert does not match the receiving log, or the email names a PO the system cannot find, the work stops and the quality lead gets a flag with the reason.
If your team cannot answer all three, the job is not ready for software or AI.
4. Plain software solves the first half, then AI sits on top
Once your team has the map, much of the fix is plain software. Take a company that runs its shop floor on paper travelers, receiving logs, inspection sheets, and cert requests, with some of those forms on paper and some in spreadsheets. The first tools you build for that company have no AI in them. Each form becomes a digital form with fixed fields, so a heat number goes into a box marked Heat Number and nowhere else.
Digital forms and one record for each job show where each job stands, who has touched it, and what still waits. They give you traceability too, which means you can follow a part from the mill heat to the shipment. Clear fields do that work on their own. What they give your team is structured data, where the same fact goes in the same labeled field each time.
You write the map first, build plain software next, and add AI last. When the AI joins the work, it uses the record and the review queue your team already uses. It looks up the heat in the record instead of guessing it from a scan, and it puts each draft in the review queue, where a person decides what goes out. When the map says stop, it stops and sends a flag to the person the map names.
Those rules keep the AI inside the lines your team drew. Your team agreed on each step, each check, and the owner for each one, and wrote them down before anyone wrote code.
What to do this week
- Pick one request your office handles every day, such as a cert request or an order confirmation. Write its steps on one page.
- Mark every point where a person checks something. Write that person's name next to it.
- For each step, write what should happen when the answer is unclear. The blank lines are your clarity bill.
- Ask if a digital form or one shared record would fix most of it before you add AI.
For the full argument, read Manufacturing's ChatGPT Moment Hasn't Happened Yet: Why AI Fits Your Paperwork and How to Prepare, and for how an AI employee uses a decision map on the job, read AI Is a New Hire: What an AI Employee Needs on Day One.
Questions people ask
If AI can write software, why does an AI project still take work?
Code is the cheap part now. Your team still has to get clear on the job by naming each step and each check and saying what happens when an answer is unclear. Without that, cheap software can speed up a process nobody has agreed on.
Do we need AI, or just better software?
We think better software often comes first. Digital forms and one record for each job fix many problems on their own. They also give AI a solid record to work from later.
Terms in this piece
- Clarity bill
- the time and attention it takes to get a job out of heads and off paper into steps a new person or software could follow.
- Coding agent
- an AI that writes and tests software on its own for long stretches, with a person reviewing the result.
- Decision map
- a written, step-by-step map of a job that marks where the AI may act and where a person must verify.
- Structured data
- information stored in labeled fields, the same way every time, like a heat number in a box marked Heat Number.
- Traceability
- being able to track every part from the mill heat to the customer shipment.
- Flag
- a mark the software puts on a value it is unsure about or that does not match another document, so a reviewer checks it first.



