
AI for Glass Shops: Measurements, Quotes and Tempered Glass
AI for glass shops: collect measurements, glass type and install site before quoting, with a technical review before any quote reaches the client.
Want to apply this in your company?
Book a call with Juan →AI for glass shops, in practical terms, means using AI to organize everything the business needs to know before quoting: measurements, glass type, purpose and installation site. The system collects, checks and prepares a draft, but the quote only goes to the client after a technical review by a responsible person. In this guide, the term is used in an operational sense, as a set of agents and workflows that support the team, not as a promise that a machine will decide safety questions on its own.
Key takeaways
- A measurement without a reference becomes rework. Record where it was taken, with a photo and a sketch, before any price is discussed.
- The glass type depends on the application. Shower enclosures, balustrades, windows and facades have different requirements, and that must appear in the request.
- The installation site changes the specification: height, exposure, access and fixing method all count.
- AI organizes the intake and flags inconsistencies, but it does not approve technical specifications or take responsibility for the quote.
- Start with one type of request, log every step and agree on a success metric before expanding.
What AI does and does not do in a glass shop
A glass shop receives requests by WhatsApp, website, phone and architect referrals. In each channel, the information arrives incomplete: a photo without a scale reference, an approximate measurement, a request for tempered glass without knowing its use. A well-designed AI agent asks the questions in the right order, checks whether the data is plausible and builds a standard summary for the team. It can also flag the cases that require a site visit.
What it should not do is estimate on its own whether a pane can bear a certain load, whether a cutout is feasible at the requested size, or whether the existing fixing will hold the piece. Those decisions depend on standards, the supplier and a technical lead. Measuring from a photo alone also has limits: without a scale reference, the margin of error grows. That is why the design of the workflow starts by separating what is reliable from what needs a person.
Which data to collect before quoting
The list below is a starting point for the intake script. Adjust it to your catalog and product lines.
- Width and height of each piece, with the measurement reference: opening, frame, wall face or profile edge.
- Quantity and identification of each piece, with room name and position, so a door is not swapped for another in production.
- Purpose of the glass: shower enclosure, window, door, mirror, balustrade, facade, shelf or partition.
- Intended glass type, or a note that the technical team will define the specification.
- Installation site: floor, height, exterior or interior side, sun and wind exposure, foot traffic area.
- Fixing method and existing structure: profiles, hardware, wall type, window frame or floor.
- Access conditions: stairs, elevator, permitted hours in the building and protection needed along the route.
- Photos with a scale reference and a simple sketch with dimensions.
- Desired deadline and whether the client accepts a site visit before closing.
None of this requires sensitive personal data. If the workflow asks only for the address needed for the visit and the information essential to the quote, the intake is easier to justify and easier to store securely, in line with Brazil's LGPD.
Tempered glass: what to decide before quoting
Tempered glass is where intake most often fails. Once tempered, the glass should not be cut, drilled or reworked at the edges, so measurements, holes and cutouts must be defined in the specification before production. A change in size after that usually means a new piece.
In addition, the maximum size and the finishing options vary by supplier and equipment. For that reason, the agent should not promise a dimension without confirmation from the supplier. It is also necessary to verify whether the intended use calls for tempered, laminated or another type of glass, according to ABNT standards for glass in construction and the guidance of the technical lead. AI can ask and record the purpose, but the technical team defines the specification.
Six-step flow: from the message to the reviewed quote
A simple flow, which can be designed before any tool is chosen, follows this order:
- First, receive the request and identify the purpose (shower enclosure, window, facade, other) and the urgency.
- Second, apply the intake script, with fixed questions and the option to send a photo and sketch.
- Third, check consistency: measurement within a plausible range, glass type compatible with the purpose, installation site provided.
- Fourth, decide the path: quote draft, request for more information or site visit.
- Fifth, send the draft for technical review, with the summary, photos and the list of open points.
- Sixth, only after approval, send the quote to the client, with conditions, deadline and validity written out.
This design separates organization, which AI performs in a repeatable way, from judgment, which belongs to the technical team. It also creates a record: each quote has the collected version, the reviewed version and the name of the person who approved it.
Technical review before anything reaches the client
The review is the step that makes the process safe. The reviewer does not need to redo everything, but should check specific points before releasing any price:
- Do the measurements match the photo, the sketch and the measurement reference?
- Is the glass type compatible with the purpose and the installation site?
- Are there holes, cutouts or edge finishes in the specification, and were they defined before production?
- Can the existing fixing support the piece, or is a site visit needed?
- Does the deadline fit the supplier's capacity and the team's schedule?
- Does the quote text state what is included, what is not, and the validity period?
At the end, the reviewer makes one of three decisions: release the quote, ask the client for specific information or schedule a site visit. That decision should be recorded. Over time, the reasons for sending quotes back show where the intake script needs to improve.
Illustrative scenario: bathroom shower enclosure with a swing door
Illustrative scenario, not linked to any real client: a person requests a bathroom shower enclosure by WhatsApp and sends a photo with an approximate measurement. The agent asks whether the door should swing open, whether the wall is tile or drywall, whether there is a profile or the fixing will go directly into the wall, and whether the bathroom can be accessed for a visit. The answer indicates hinges and a handle, which means holes and edge finishes must be defined before production. The agent does not close a price. It generates the summary, marks the case as a site visit request and sends the team the photo, the sketch and the open questions. During the visit, the technician checks the wall, defines the specification, and only then does the quote go to the client.
The value of this example is not the speed of the reply. It is that the questions about holes and fixing were asked before the price, not after the piece was already in production.
How to apply Juan Carlo's method
The recommended sequence for a project like this starts by mapping the current process: who answers requests, how quotes are prepared, where information gets lost and how many quotes come back for correction. That map becomes the baseline. Next, data and actions are limited: the agent collects only what is needed, does not decide the specification and does not send a price without review. Then a real request type is piloted, with a log of each step and a success metric agreed in advance, such as the share of quotes returned for correction. Only then does the company decide whether to expand, adjust or stop. The method is the same for glass shops, survey firms or custom furniture makers, changing only the intake script. To compare how other sectors handle measurement and quoting, see AI for Custom Furniture Makers: From Quote to Installation and AI for Surveying Firms: Intake, Quotes and Reviewed Deliveries.
Before serving real clients, the flow can be tested with fictitious data. The process is described in Mock Tools for Testing AI Agents: Safe Sandbox Testing Guide. If inconsistent data is the bottleneck, the first step is Data Ready for Superintelligence: What to Standardize First.
Frequently asked questions
Is superintelligence in glass shops real already?
In practice, the term describes using AI to organize requests, measurements and information about glass, not a machine that decides on its own. Current systems collect data, follow scripts and prepare drafts. Technical decisions remain with the team.
Can a quote be made from photos alone?
A photo helps understand the request, but it does not replace a measurement with a reference and, in many cases, a site visit. Without a scale and a sketch, the margin of error grows. The safest use is to prepare the visit with the photo, not to close a price based on it.
Who approves the quote before it goes to the client?
It should be a person with technical responsibility in the company, following a clear checklist and recording the decision. The agent can prepare the draft and flag open points, but it should not send a final price without that approval. This rule needs to be written into the process.
Can AI choose the glass type on its own?
It is not advisable. The choice depends on the purpose, the installation site, the applicable standards and the guidance of the technical lead. AI's role is to ask, record and flag incompatibilities so the team can decide.