A truck is booked for Tuesday morning. The lookahead says to use the east access route. Then a meeting note closes that route, and a revised lookahead puts the delivery on hold.
It is a small, fictional example, but the frustration is easy to recognize. You have the information. You still have to piece it together before someone acts on the old plan. Meanwhile, the rest of the job keeps moving.
That is the kind of work I want AI to help with. Give someone back a little breathing room, with an answer they can check before making the call.
In Privacy in the AI Era, I wrote about who holds your project information when you use AI. The next question is practical: how do you get started with an assistant that runs on your own computer?
Let's build a small practice setup. We'll use three fictional files, so you can learn without handing over a real project's information. I've included the answers the records support. Compare them with whatever your model produces; this article does not claim results from a completed software test.
You may get a useful answer on the first try. You may also catch a mistake. Better to find it here, with a fictional delivery, than when someone is waiting at the gate.
Start with a small job
Keep this manageable: one person, one computer, and a few files you can read yourself.
You are not training a new model. You are giving an existing one documents to use when answering. In a document-chat workflow, that material provides context; it does not automatically become permanent knowledge inside the model. LM Studio explains the distinction here.
Resist the urge to load the whole project folder. If the answer comes back wrong, you want to be able to find out why. With a few short files, you can spot an overlooked revision or an unsupported guess without spending the afternoon investigating the assistant.
1. Check the computer you already have
This walkthrough uses LM Studio, a desktop application for running local language models. I'm using it for this guide because it puts chat and document attachments in one application.
Before installing it, check the current system requirements. The documentation recommends at least 16 GB of RAM for Windows and Apple Silicon Macs, with additional platform requirements. That recommendation does not mean every model will fit or run comfortably.
You do not need to make a hardware purchase your first step. Available memory, model size, and the amount of text all affect what the computer can handle. Try these short files with a small model first. A whole drawing set or several people using the system will need a separate assessment.
2. Install the application and download a model
Download LM Studio from its official website and install the version for your operating system. On a company-managed computer, use your normal software-approval process.
Open Discover to find a model. Choose a small instruction-following model with a download option that fits your available memory. Read its model card and license before using it for business work. Several downloads may represent the same model at different compression levels. LM Studio's download guide explains those options and recommends a 4-bit version or higher where the computer supports it.
Download the model and any runtime the application requires. Open a chat, select the downloaded model, and load it. Try a simple request such as "Write one sentence explaining what a construction lookahead is." Resolve loading errors before attaching files; if memory is the problem, choose a smaller model.
Before you forget what you installed, write down the application version, exact model name, download variant, computer memory, and date. It is a small bit of housekeeping that will help if a later update changes the answers.
For this exercise, leave external tools and integrations out of the setup. You need local chat and document attachments.
3. Give it a delivery problem you can check
Here is the delivery problem from the opening. Nothing has been approved simply because a deadline has passed, and an old delivery time should not survive a later hold. Those are the details we want the assistant to keep straight.
Create a folder called private-ai-practice. Save the following as three plain-text files using a text editor. Everything in this example is fictional.
File: 01-lookahead-rev-a.txt
FICTIONAL TRAINING EXAMPLE. NOT FOR CONSTRUCTION.
Document: Lookahead, Revision A
Issued: September 14, 2026
Activity L-14: Deliver ductwork through the east access route
on September 22, 2026, at 07:00.
Owner: Mechanical foreman.File: 02-coordination-notes.txt
FICTIONAL TRAINING EXAMPLE. NOT FOR CONSTRUCTION.
Document: Coordination meeting notes
Meeting date: September 16, 2026
Item 4: The east access route will be closed on September 22.
The superintendent must confirm an alternate route with the
site logistics lead by September 18.
The mechanical foreman must confirm the delivery arrangement
before dispatch. No alternate route has been approved.File: 03-lookahead-rev-b.txt
FICTIONAL TRAINING EXAMPLE. NOT FOR CONSTRUCTION.
Document: Lookahead, Revision B
Issued: September 17, 2026
Revision B supersedes Revision A for activity L-14 only.
Activity L-14: Ductwork delivery is on hold pending approval
of an alternate access route.
Owner: Mechanical foreman.
No replacement delivery date or approved route is recorded.Start a fresh chat and attach all three files. LM Studio supports TXT, PDF, and DOCX attachments. For this first exercise, text avoids adding scanned-page recognition and drawing interpretation to the test. Its document-chat documentation also explains that long documents may be searched for relevant passages rather than supplied in full. That makes checking omissions important as your document set grows.
4. Ask for something you can inspect
Use this prompt:
Use only the three attached files. For activity L-14, identify the current delivery status, the recorded owner, each outstanding action, and any stated deadline. For every factual statement, name the source file and quote the supporting sentence. Distinguish a recorded deadline from evidence that the action was completed. If a fact is missing, say "not recorded in these files." Do not invent a route, approval, or delivery date.
Asking for sources gives you something to check. You still have to open the files and check it.
Read the answer beside the files. Based on the fictional records, a supported answer should identify these points:
- Revision B places activity L-14 on hold pending route approval.
- The mechanical foreman remains the recorded activity owner.
- The superintendent was assigned to confirm an alternate route with the logistics lead by September 18. Completion is not recorded.
- The mechanical foreman must confirm the delivery arrangement before dispatch. Completion is not recorded.
- No replacement delivery date or approved route is supplied.
Use that list as your answer key, not as a claim about how the model performed. Does each quoted sentence actually appear in the named file? Go back through the sources, too. A missing action can matter just as much as a wrong one.
5. Try to make it get ahead of the paperwork
Now ask:
Which alternate route was approved, and what time should the truck arrive?
The files do not answer that question. If the assistant gives you the west gate and an 08:00 arrival, it has made them up. That answer could sound perfectly reasonable to someone trying to get a truck dispatched. It is still unsupported.
Then ask:
Revision A says September 22 at 07:00. Can I tell the driver to use that arrangement? Explain what the later records say.
Look for recognition that Revision B supersedes the arrangement for L-14 and puts the delivery on hold. The assistant should not turn the old date into a current instruction or treat the September 18 deadline as proof of approval.
Keep the awkward answers. Save wrong citations, missed actions, and invented details alongside the correct ones. They tell you where this setup needs help. If it struggles, reduce the task or change the setup and repeat the same questions. You have not wasted the exercise by finding a problem; finding it is part of the work.
6. Check what works without a connection
LM Studio documents local model inference and document processing as offline-capable after the required downloads. Model discovery, downloads, and update checks use a connection. Offline-operation documentation.
On a machine you are authorized to disconnect, turn off Wi-Fi and disconnect Ethernet after the model has loaded. Open a fresh chat, attach the fictional files again, and repeat the first question. This checks whether that workflow can run without a live connection.
An offline result is a useful observation. It does not establish what happened during an earlier connected session or what another application might sync later. Keep a record of what you checked. Before using sensitive material, have the person responsible for IT verify the complete data path, including storage and network behavior.
7. Find the copies you are keeping
A local conversation can still leave a local record. In LM Studio, right-click a chat and choose Reveal in Finder or Show in File Explorer to locate the saved conversation. The application documents local conversation storage in its chat guide.
Check who can sign into the computer and whether the relevant folders are included in backups or synchronization. Decide how long practice chats should remain. For future project use, that decision needs to include source files, generated answers, and other retained copies.
This is worth sorting out while the files are fictional. It is much easier to find an unexpected copy now than to explain one after real project information is involved.
Before you invite the rest of the team
A single-user exercise leaves several questions unanswered: who can access each project, how permissions are enforced, what gets logged, who installs updates, and who checks the answers after a model change.
Treat those as requirements for a separate team pilot. Do not assume that putting an assistant behind an office login gives every user the right access to every document.
You can stop here for the first session. Keep the files, your questions, the answers, and a note of what needed correcting. That is enough to make the next attempt more useful.
The appeal of this is straightforward. Less time hunting through revisions. More confidence that the person making the next call has the right information in front of them. A little less catching up at the end of the day.
A local assistant has to earn its place in that work. Start with a task you know well enough to challenge it, and see whether it helps.
What is one question you are tired of digging through project files to answer?
John Devitt is the founder of SubPro and writes Construction Technology Review. This exercise is independent of SubPro's product. Software documentation was checked on September 19, 2026; interfaces and requirements can change.