Prestressed concrete with AI — the beam that leaves with a strand at 1,510 when it should have been 1,560 is invisible until it reaches the site.

The problem

Three realities that the bed, the lab and the yard never share in real time.

A sound prestressed unit is where three data points meet — and in most plants each one lives on its own island:

1. What tension each strand carries — the gauge reading. It usually sits on an old panel, sometimes written by hand on a bed sheet the quality manager sees the next day.
2. What concrete was batched — the mix design. In the ERP, in the batching plant control system, or in a spreadsheet kept by the lab manager that hardly ever reaches the bed at the moment of casting.
3. How it was cured — the temperature/humidity curve. Sometimes on a local recorder with no network, sometimes only in the bed supervisor's head.

The quality manager knows it, but cannot be in all three places at once. The unit leaves, goes into the yard, and two days later the yard flags a microcrack — or worse: it reaches the site and the client rejects it. The classic system works 99% of the time. That 1% is the one that reaches the customer — and every rejection means return haulage, an urgent replacement and a damaged relationship, not an internal incident.

How it fits the IRIS system

iLEAN does not add a fourth system — it seals the cracks between the three you already have.

The problem with prestressing is not a lack of information: it is information living on islands that, at the critical moment (tensioning, demolding, stacking in the yard), does not reach the decision-maker in time. iLEAN acts as the putty that fills those gaps, without asking you to change the old panel on the bed, the batching system, or the curing chamber.

Edge reads the gauge and the unit at demolding. Connect reads the mix design wherever it lives — batching plant control, MES or spreadsheet. The agent cross-references it with the curing curve and, if something does not add up, holds the unit. The person signs — never the other way round.

The three iLEAN pieces applied to prestressing control:

• Edge — a terminal with machine vision (CNN) over the bed. It reads the needle on the analog gauge, the display on the old unit and the surface crack pattern at demolding. It flags the unit in the gray zone and triggers an actuator (light stack, cart ejector). It works with no network. If the plant loses WiFi, Edge keeps reading and logging.
• Connect — captures the mix design whether it comes from the batching plant control system, from the MES or from a spreadsheet the lab updates every morning. And it also captures what arrives from outside (an admixture change from the supplier by email, the certificate for the new strand coil, a notification from the end customer with a change to the specification) at second zero.
• Agent — cross-references the actual tension strand by strand, the mix design, the curing curve and the crack pattern. If there is a deviation, it does not send an email at 10 p.m.: it holds the unit before shipping and alerts the bed supervisor on whichever channel they use. The person validates and signs; the cart does not leave on its own.

See the full IRIS architecture →

Before and after

Manual control vs. cross-referenced control with iLEAN

Aspect	Manual control + bed sheet	With iLEAN Edge + Connect + Agent
Tensioning force per strand	Written by hand on the bed sheet; reviewed the next day	Gauge reading at second zero, attached to the batch
Mix design available at the bed	The operator calls the lab if in doubt	The batch mix design is at the bed at the moment of casting
Curing curve	Local recorder — downloaded if the customer asks	Cross-referenced with tension and mix design, batch by batch, nothing to download
Microcrack detection	In the yard, or already on site	At demolding, with vision and raking light
Operation with no network	n/a	Edge keeps running on panel power alone
File for the client	Rebuilt by hand, days	Automatic per-unit dossier, with a photo at demolding

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Precast plant with 2-3 prestressing beds (beams, joists, half-slabs or girders), multi-mix and multi-customer.
• Edge pilot on one bed (gauge reading + vision at demolding + integration with the mix design and the curing chamber). First value expected within a few weeks.
• Reduction in units rejected on site of ≥ 30% against the record of recent years — a conservative estimate.
• Indicative payback between 4 and 9 months, depending on the frequency of documented rejections and the average cost of an emergency replacement unit (return haulage, express production, contractual penalty).
• The hard lever is a single beam rejection on site avoided: it pays for a good part of the pilot.

And the quality manager's reasonable doubt

“What if the AI gets it wrong and lets through a unit with a badly tensioned strand?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely recontextualizes a piece of data from one system into another (reading the gauge and comparing it against the specified set point), the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN holds the unit and the person signs. The three safety rings exist precisely for this.

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