Asbestos enclosure verified before work starts

The problem

An eyeballed sign-off is the weakest point of the whole asbestos work plan.

A sealing flaw missed in time is a possible exposure for personnel and puts the company's asbestos-work authorization at risk. Today, the "ready to start" validation depends on an eyeballed sign-off, with no objective visual evidence at each critical point.

• Before touching asbestos-containing material, the work plan requires a sealed enclosure, signage, an airlock, negative pressure and the specific PPE for the crew going in.
• All of that is checked today by a supervisor who walks around it and says "ready". There is no objective visual evidence of any single critical point, only a signature on a form.
• A sealing flaw missed at that moment is a possible exposure for the crew going inside, and it is exactly the kind of finding that puts the company's asbestos-work authorization at risk if an inspector sees it first.
• And the crew changes between sites: the person signing the enclosure off this morning is often not the one who built it yesterday afternoon, and has nothing but the enclosure itself to go on.

How it fits the IRIS system

Edge with JIDOKA AI — removal does not start until every point confirms its state.

Edge + JIDOKA AI: fixed cameras with CNNs trained to recognize "sealed enclosure vs. visible flaw" at each critical point. The system blocks the start of removal work until every camera confirms the correct state and the safety officer signs off based on that evidence.

The camera does not decide alone and it does not replace the safety officer. It checks each critical point against its reference state and puts that evidence in front of the person who signs, so the signature rests on something other than a walk-around and a good memory of how the enclosure looked yesterday. What changes is not who decides, but what they get to decide on.

See the full IRIS architecture →

Before and after

Today's sign-off versus the verified enclosure

Aspect	Today	With iLEAN Edge
What "ready" rests on	A walk-around and a signature	Evidence at every critical point
Sealing of the enclosure	Checked by eye	Compared to the reference state
Airlock and signage	Assumed in place	Confirmed before start
PPE of the crew going in	Checked if there is time	Checked every entry
Start of removal	Whenever the supervisor says	Blocked until the state checks out
Record of the check	The signed form	The form plus its images

Before: eyeballed sign-off with risk of a flawed enclosure. After: sign-off grounded in objective visual evidence at every critical point.

Impact estimate

Impact estimate — this one is defended as protected authorization, not as payback.

The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.

• The brief gives this case no payback range, and it is not invented here: what it protects does not reduce to a monthly figure.
• It protects the asbestos-work authorization, which is the permit the whole business line depends on.
• It reduces the exposure risk of the crew who go inside the enclosure, which is the part no estimate should ever be put on.
• It is a case to size with your safety officer, against the last enclosure that was signed off by eye.

Protects the asbestos-work authorization and reduces personnel exposure risk. Estimate to validate with the safety officer.

And the fair question from the production manager

"What if the camera blocks a removal for no reason?" — the block is what makes the case worth having, so the threshold matters. Conformity is checked against the reference state of that specific enclosure, and recognizing a sealed panel versus a visible flaw is an anchored task, where the best models drop below 1.5% error [1]. In case of doubt it does not block silently: it escalates to the safety officer, who decides on the evidence.

^[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.