AI for construction
AI won't fix a broken process.
It scales a good one.
You already own the software. AI is the next layer on top of it — applied to the handful of workflows where it actually pays off, with controls your controller and your IT provider can both live with.
Use cases
Where AI actually helps in construction
Dashboards & reporting
Bring information from multiple systems together into useful dashboards and recurring reports — without manually rebuilding them in Excel every month.
Custom operational apps
Build simple, purpose-built tools for the gaps your existing software doesn't cover — work reporting, inventory, tool tracking, timekeeping, safety, ticketing and more.
AP & invoice processing
Read invoices, match information to jobs, POs and cost codes, and flag exceptions for review — reducing repetitive data entry.
Document processing
Read, classify and extract information from contracts, invoices, forms, reports and other documents — then send that information where it needs to go.
Workflow automation
Use AI within repetitive processes to read information, make routine decisions, route requests and complete steps that are currently being done manually.
Internal AI assistants
Give your team a way to ask questions across company procedures, documentation and internal information instead of searching through folders or asking the same person.
Scope of work
The two kinds of AI work I do
Custom tools & AI coding
Small internal apps, scripts and integrations built quickly to close a gap your ERP won't — without a six-figure development project.
- Internal dashboards and trackers
- Integrations between systems that don't talk
- Built fast, documented, handed over
Examples
- Inventory & tool tracking
- Timekeeping & crew entry
- Work reporting & daily reports
- Safety forms & inspections
- Ticketing & field billing — hired trucks, aggregate, T&M, unit rate
- Custom dashboards & reporting
Agents
Automations that run one repeatable step end to end — read, classify, enter, notify — with a human approval gate before anything commits.
- One workflow at a time, scoped tightly
- Human sign-off on every commit
- Audit trail of what it touched
Examples
- AP invoice processing — read invoices, extract information, suggest coding and route for approval
- Document processing — read, classify, rename and file incoming documents
- Report summaries — turn project or financial data into concise management summaries
- Internal knowledge assistant — ask questions across company procedures, documentation and other internal information
Method
How I implement it
- 01
Find the bottleneck
We start with where time actually disappears — not with a tool looking for a job.
- 02
Confirm it's worth automating
If the process is broken or the data is dirty, we fix that first. Automating a mess just makes a faster mess.
- 03
Pilot one workflow
One narrow use case, real data, measurable before-and-after. No company-wide rollout on a hunch.
- 04
Set the controls with IT
Review steps, approval gates, data rules and an audit trail — agreed with your IT group so finance and IT can both sign off.
- 05
Hand over to your team and IT
SOP videos recorded in your system so your team runs it and your IT group can maintain it without me.
Starting point
Most engagements start with one workflow and a short pilot — small enough to prove out, big enough to feel.
What AI is not
- A shortcut around bad data or a broken process
- A reason to replace an ERP that can be fixed
- Unsupervised approvals on payroll, AP or banking
- A black box nobody in the office can explain
Data safety & governance
What leaves your systems
We decide explicitly which data a tool may see, and which stays inside your network. That list gets written down, not assumed.
Who can approve what
Every automated step ends in a named human approval for anything financial. AI drafts; your team commits.
Retention & audit
Vendor retention settings configured deliberately, plus a log of what each automation read and wrote.
Working with your IT
I don't replace your IT team — I work with them.
Your IT group owns your environment, your data and your security posture. My job is to build inside those lines and hand the work back in a state they can support.
01
Built to your guidelines
Any AI tool or agent is reviewed with your IT so it follows internal policies for data, access and approved platforms — no shadow tools, no surprises.
02
IT can take it over
Everything I touch comes with SOP videos and documentation, so your IT team can maintain, change or shut it down without needing me on the phone.
03
Standards for the work I build
If you don't already have AI guidelines, I set the standards for the projects I build — what data the tool can use, where it runs, who approves it and how it gets reviewed — so IT has something concrete to sign off on.
Fees & next step
Billed in 15-minute increments, no retainer and no commitment. Bring me in to look at one workflow and tell you honestly whether AI is the answer.