Your Manager Catches It. But What If He Didn’t Have To?

Most labs run quality control the way it should be run. The designer checks the case. The ceramist checks the case. The manager does the final check before it ships. Multiple people, multiple stages, each one catching what the last one missed.

The manager’s final check is the most valuable step in that chain. It’s not just a pass/fail. It’s years of accumulated knowledge about which tolerances matter most for which case type, how different dentists want their contacts, what “right” looks like for a specific material under specific conditions. That kind of judgment is built over thousands of cases. It’s the most valuable asset in the QC process.

Here’s the thing about that knowledge: it mostly lives in the manager’s head. He applies it at the final check, case by case. When a case gets sent back, the tech learns the outcome. This one didn’t pass. But the standard behind the decision, the reasoning, the full picture of what the manager weighed, that part doesn’t always transfer. The tech knows what was wrong with this case. They don’t always walk away knowing what “right” looks like across all the variables the manager considers.

That’s not a flaw in the process. It’s the nature of expertise. The manager’s judgment involves critical thinking that can’t be fully reduced to a checklist. But some of what he checks, the repeatable, pattern-based standards, can be made available earlier in production, so techs catch the straightforward issues before the case ever reaches his desk.

---

How We Built a System Around This

We built a QC system using Anthropic’s Claude that captures the repeatable parts of the manager’s process and makes them available at every production step.

The manager doesn’t change anything about how he works. He does final QC exactly as he always has. His checks, his sequence, his judgment. But now, every decision he makes feeds the system. When he sends a case back, the system records what he flagged, what type of case it was, and what standard it didn’t meet. When he passes a case, that’s recorded too. Over hundreds of cases, those decisions form a detailed picture of how the manager evaluates quality across different case types, materials, and dentists.

Claude is what makes this practical. The manager doesn’t fill out structured forms after every check. He gives feedback the way he always has, in plain language. Claude interprets that feedback, maps it to the relevant criteria, and adds it to the model. The manager’s natural way of communicating becomes the data the system learns from.

Once the system has enough of the manager’s decisions, it starts surfacing relevant checks earlier in production. When a tech finishes a stage, Claude looks at the case context: the restoration type, the material, the dentist. It pulls up the checks the manager consistently applies to that combination. The tech sees the same criteria the manager would use at final QC, applied at their stage, before the case moves forward.

The manager’s final check stays exactly where it is. Nothing ships without him. But by the time a case reaches his desk, the repeatable issues have already been caught. His time shifts from the straightforward catches to the judgment calls that actually need his expertise. The edge cases. The decisions that require years of experience, not a pattern match.

No system replicates the way a good manager thinks through a case. What this does is make sure the repeatable parts of his standard don’t wait until the final check to show up.

---

How You Can Do This Right Now, For Free

You can set up a version of this yourself using a free tool. It requires your manager to sit down and articulate his standards, but once it’s set up, your techs can run cases through it before the manager ever sees them.

Here’s how it works. You set up a QC assistant that knows your manager’s standards. When a tech finishes a case, they run it through the assistant before handing it to the manager. The assistant walks them through the manager’s QC sequence. Same checks, same order, same criteria. It confirms whether each one meets standard.

What you need:

• A free Claude.ai account (claude.ai)
• 20 to 30 minutes of your manager’s time for initial setup
• A phone or computer with a web browser

Step 1: Set up the workspace

Go to claude.ai. Create an account if you don’t have one. Click “Projects” in the left sidebar, then “New Project.” Name it something your team will recognize. “[Your Lab Name] QC” works fine.

In the project settings, click “Set project instructions.” This is where the system learns how to behave. We’ve published a ready-made system prompt. Copy it from the link below and paste it into the project instructions field.

Step 2: Train it with your manager’s process

Open a new conversation in the project and type “Train.” The system will start asking your manager questions about how he runs QC:

• What do you check first? Second? Third?
• What are the pass/fail criteria for each check?
• Does the sequence change by case type?
• Do any dentists have specific preferences?
• What mistakes come up most often?
• What’s an automatic reject?

Your manager answers in plain language. No forms, no technical setup. Just a conversation. The system takes his answers and generates a structured QC Protocol. His standards, documented and organized.

When the conversation is done, copy the QC Protocol document it produces. Go back to the project settings, click “Add knowledge,” and upload it.

That’s it. The project is trained.

Step 3: Techs run QC checks

When a tech finishes a case, they open the project on claude.ai and start a new conversation:

• “QC check, single crown, posterior”
• “QC check, 3-unit bridge, anterior”
• “Implant case QC”

The system reads the QC Protocol, identifies the case type, and walks the tech through the manager’s checks one at a time. The tech performs each physical check and reports what they see. The system compares against the manager’s standard and moves to the next one. At the end, it summarizes what passed and what needs attention.

The tech addresses what they can and hands the case to the manager. Over time, techs start internalizing the standards they keep getting flagged on. The same way they would working alongside the manager, except the manager doesn’t have to be there for every case.

Updating it:

When standards change or a new dentist comes on, the manager opens a new conversation and tells the system what changed. It generates an updated protocol. Replace the old knowledge file with the new one.

---

The manager’s expertise is the most valuable thing in your QC process. The way he thinks through a case, the judgment calls, the instinct built over years. That’s not something a checklist or a system replicates. What this does is take the repeatable parts of his standard and make them available earlier, so his time at the final check goes toward the decisions that actually need him.