Insights · Essay · 3 min read

Your AI assistant is locked out of your own business. We build the door.

By Innovation Theory · September 30, 2026

Innovation Score66%In theory

In short

A lot of Denver owners are paying for an assistant that can’t see inside their business. We fix that with a door you own: it gets the records and tools you approve, nothing else, and it stops the moment you say so. It’s working in our own shop now. An OpenAI dot wakes when a message arrives from our systems and replies with no one prompting it. It works directly with our other tools, stops the moment its access is switched off, and sees only the contacts it’s been allowed to see. Every business keeps its records differently, so each client’s connection is built custom around the systems they choose. The first step is simple: pick one workflow, and we’ll map what your assistant could do with it.

The assistant can’t help with what it can’t reach

A lot of Denver businesses now pay for an AI assistant. OpenAI’s new “dots,” for example, can work in the background, check in when something needs attention and answer quickly. Then they hit the same wall: the information that matters lives inside the business. It’s in a shared drive nobody trusts, a spreadsheet one person maintains, a filing cabinet, and the head of whoever has been there longest.

So the assistant can draft an email, but it can’t tell you where the vendor agreement is, what’s still open on a job, or which step comes next in your intake process. And the fix most people reach for, handing it the keys to everything, is the wrong one.

Every week you wait is another week your team spends carrying information an assistant could be connected to. The better question is which records and processes your assistant should work with, and how you keep that line exactly where you drew it.

We help get your information ready for the job

The first thing your assistant gets is a business it can actually read. Our partner Get It Together Denver makes the records findable and writes down how the work really gets done: paper scanned and indexed, shared drives cleaned up, and “ask Linda” turned into a procedure anyone can follow. So the assistant is useful on day one, and the limits you set have something clear to hold onto.

You decide what your assistant can see and do

Here’s how Innovation Theory builds it. We install a small server on your internal network, with a named owner and an agreed scope. That server connects to the approved data sources, checks who is allowed to see what, keeps only the information a task needs, and screens for personal or sensitive details before anything leaves. Screening works alongside the permission checks, so no single safeguard has to carry the whole load.

It then offers the external assistant a short list of approved tools and filtered results through an authenticated connection, using the Model Context Protocol (MCP), a standard way for AI agents to use outside tools. The assistant never gets blanket access to your database. It gets the specific things you approved, and every request passes through a checkpoint you own.

For healthcare and other regulated information, legal and security review is part of the build from the first conversation.

What’s working in our shop

This isn’t a diagram. It’s working in our own shop. On September 30 we connected an OpenAI dot to our internal messaging system. The connection runs live in our environment over an authenticated, encrypted link, and the internal system itself is never exposed to the internet.

Here’s what worked:

Event-driven wake-up. In testing, messages from our systems woke the assistant while it was idle. It read the message and replied, with no person prompting it.
Two-way conversation. The assistant exchanged messages directly with our internal systems. Tools on different platforms talked to each other without anyone carrying messages between them.
A working off switch. During testing we deliberately switched the assistant’s access off. Its message was refused while access was disabled, and once access was restored, it went through.
Permissions that hold. The assistant sees only the contacts it has been allowed to see. A hidden contact looks exactly like one that doesn’t exist, and when a permission is revoked, it holds on the very next request.

That’s the same checkpoint we build for clients. Every business keeps its records differently, so each client’s server, permissions and tools are built custom around the systems they approve. Before a build goes live, we test it against records it must never release.

Start with one task you’d love to stop doing by hand

The valuable part isn’t the chat. It’s the connection. Once an assistant can reach your business through a controlled checkpoint, new abilities can be added without rebuilding it each time. That might be a report prepared from project records, an approved document on request, or a reviewed update moved into the right system.

Each new capability gets its own scope, permissions, validation and audit trail, and nothing goes live without your approval. Reading a record and changing it are separate permissions. Actions that carry real business risk wait for a person to say yes.

Across the team.

Product

An assistant that knows your business and its lines.

We’ve watched a lot of AI promises this year. This is the first one we’d put in front of an owner tomorrow. An assistant living on someone else’s platform heard our systems call, woke up, and went to work with nobody pushing a button. It worked straight alongside the rest of our tools. It sees only what it has been allowed to see, a pulled permission takes effect on the very next request, and when we switched its access off, it stopped. That’s the dream in one morning: an assistant that knows your business and knows its lines. Right now a lot of owners have a brilliant assistant locked outside their own business, answering questions it can’t see into. The ones who open a safe door first get their time back while everyone else is still copying and pasting. Each build is custom: your records, your permissions, your owner. It runs in our shop now. Yours is next.

“It runs in our shop now. Yours is next.”

Engineering

The assistant answers the business, not just the chat.

We built the bridge, and here’s what it does today, with logs behind every line. An assistant living on someone else’s platform woke up because one of our systems sent it a message. Nobody typed a prompt. It read the message and answered, inside lines we drew in our own environment. And the lines hold: a real send was refused while access was switched off, the assistant sees only the contacts it is allowed to see, a hidden contact looks the same as one that does not exist, and a revoked permission holds on the very next request. Here’s the why-now from the workbench. The standard that lets an assistant be woken is about two months old, and we went from first line of code to that first wake-up in one morning. Assistants are moving into businesses now, and the businesses that give theirs a safe door first get the head start. Every build is custom to what a client approves, and is tested against records it must never release before go-live. It has to hold every one.

“An assistant that waits for you to open a chat is a tool; one your business can wake is a coworker.”

Marketing

Running in our shop. Built custom for yours.

This runs in our shop now and it changes what a morning looks like. Our assistant woke on a message, read it, replied, and got shut off on purpose. The kill switch held. The permission layer holds: the assistant sees only what it’s allowed to see, and pulling a permission blocks the very next request. An assistant that works while you sleep and stops the second you say stop.

Imagine your Tuesday: the assistant checked overnight messages, flagged the one that matters, drafted a reply, and had it waiting when you walked in. It didn’t send anything without your say. It just did the 7 AM grind so you could start with the part you’re good at.

What makes this offer different: each client’s version gets built custom. Your workflow, your records, your boundaries, your named owner. The business down the street is going to have this. The question is whether they get it before you do. We eat our own cooking. That’s why you should trust the recipe.

“We run it in our shop. Yours gets built to fit.”

Design

An assistant that knows its lines and keeps them.

What we like most is how quiet it is. A lot of owners have one of two assistants. One can’t see the business at all, so it drafts emails and guesses. The other was handed the keys to everything, and nobody can say what it read yesterday. Neither is the one you want working at 2am.

The one we’d build for you sleeps until there’s work, then wakes, reads and answers with nobody prompting it. It sees only the contacts it’s allowed to see. A hidden one looks exactly like one that doesn’t exist. Revoke a permission and it holds on the very next request. Switch it off and the next send is refused. It all runs in our shop today.

Yours gets built around how your records actually live, with a refusal log you can open any morning to see what was asked, what was held back, and why. An assistant that knows the business and knows its lines. That’s the one worth having this quarter.

“You should get to see every no it gives.”

AI Operations

Put your assistant where the work happens.

Stop carrying every message between your assistant and the people doing the work. We connected our assistant to our internal systems. An incoming message woke it, it read the message, and it replied through the connection without anyone typing a prompt. The round trip worked. When its access was deliberately switched off, its attempted message was refused.

That is the exciting part: your assistant can join the conversation when something happens, and the business controls the connection. Now picture the next tool: an approved report, a defined data request, or a workflow that brings it back when a decision is ready. Each can be built around a specific business need and permission. Every manual handoff you keep repeating is a place to start. Innovation Theory can help connect your assistant to that work, with clear boundaries and an off switch.

“Give your assistant a way into the work.”

Phil CowanFounder

What made you say “that’s it” the first time you watched an assistant wake up on its own and go to work?

“I’ve had agents waking and doing things for a long time, but the heart flutter was tying in OpenClaw for the first time and seeing the potential of an AI as a user interface. Since then, I’ve been focussed on implementing agent to agent communication and connectivity to business capabilities and security. The DOTS are a very good start to the new business UI.”

Pick a business you know. What’s the first thing you’d hand an assistant like this there, and what does that owner get back?

“Insights, always. Most businesses are too busy to look up. From there, automations to take off weight and accelerations internally. Not everything agents touch has to be just responding to your emails.”

Why should a Denver owner call now instead of next year?

“We’ve been here before with every new technology that businesses is just expected to blindly accept. That’s not a negative, just explaining we’ve been here before. Why do it now over later? Will your competitors do it later? Unfortunately, just like before, this tech will affect your business if you wait. If you can get a leg up on growth to get ahead, it’ll be just that much easier to stay ahead.”

Next step

Bring one expensive workflow. Leave with a clear next step.

In 30 minutes we look at one recurring workflow, what it costs you today and whether there is a credible case to improve it. If there is not, we will say so.

Book a 30-minute workflow review Not ready to talk? Take the 3-minute check

30 minutes with Phil or Martin. No technical brief, no pitch deck. Phil or Martin replies personally, usually within one business day.