Stop Answering the Same Questions: The AI Policy Bot for Architecture Firms
“How much PTO do I have?”
“Can I expense parking at the client site?”
“Can I work remotely next Friday?”
“What’s our policy on using ChatGPT?”
None of these are bad questions. Employees need answers.
The problem is who has to answer them.
In a 10-, 25-, or 50-person architecture firm, questions like these frequently end up with the same handful of people: a principal, office manager, operations leader, or project manager. And when that person is working on a proposal, reviewing a project, solving a client problem, or doing deep design work, even a simple question becomes an interruption.
Multiply that across dozens of employees and hundreds of questions a year, and you start to see the opportunity.
The information probably already exists
Most firms already have an employee handbook. They may also have expense policies, remote-work guidelines, AI policies, standard operating procedures, design standards, and other internal documentation.
The problem is that employees don’t necessarily know where those documents live.
And even when they do, finding the answer might mean opening a 40-page PDF, searching for the right term, reading several sections, and trying to interpret what the policy actually means.
The easier alternative?
Ask the person who already knows.
That is exactly the behavior we wanted to change at ArchIT.
What if employees could just ask the firm?
We built an internal Policy Bot that our team can access directly through Microsoft Teams.
Instead of messaging Boris with a question about PTO, an employee can ask the Policy Bot:
“How much PTO do I get?”
The bot searches the firm’s approved documentation, returns the relevant answer, provides a confidence level, and points the employee back to the source document.
That last part matters.
For policies, procedures, AI governance, expenses, or other areas where accuracy is important, we don’t want an AI system creatively inventing an answer.
We want it retrieving the answer the company has actually approved.
That is why the system is deliberately grounded in the firm’s own documents rather than treating the open internet or the AI model itself as the source of truth.
Sometimes “I don’t know” is the most valuable answer
There is another benefit we didn’t fully appreciate until we started using the system ourselves.
The Policy Bot exposes gaps.
During our demonstration, we asked:
“Are we allowed to use ChatGPT for work?”
The bot couldn’t answer correctly.
That’s useful.
ArchIT actually has an AI policy, but the bot didn’t have access to it in the right place. Instead of making something up, the failure told us we had a documentation or configuration problem to fix.
We’re now working toward a system where unanswered employee questions can be surfaced automatically to the appropriate manager.
Think about what that creates over time.
Instead of a principal answering the same question 20 times, leadership can answer it once, improve the policy or SOP, and make that knowledge available to everyone going forward.
Answering the question repeatedly is an interruption. Building the system that answers it permanently is management.
The bigger opportunity: institutional knowledge
Policy Bot is only the beginning.
Architecture firms accumulate enormous amounts of knowledge over decades:
How does our firm approach this type of project?
What have we learned from previous school projects?
What standards do we use?
How do we handle this particular CA process?
Where is our digital design manual?
What went wrong on a previous project, and what did we change because of it?
Much of that knowledge is buried in documents, folders, previous projects, or—most importantly—in the heads of experienced people.
Imagine if an employee could ask a question and tap into more of what the firm already knows.
That doesn’t eliminate mentorship. Far from it.
But it can make mentorship and institutional knowledge scalable.
Senior people can spend less time pointing employees toward documents and answering routine questions, and more time providing judgment, coaching, design input, and experience—the things that are much harder to automate.
That’s where this becomes much more interesting than another AI chatbot.
You’re starting to build a usable knowledge layer for the firm.
And for architecture firms trying to get practical value from AI, that may be one of the most valuable places to start.
If you have questions or need help please reach out to us.
ArchIT specializes in providing IT services for architecture, design, and engineering firms.