Free AI Toolkit graphic featuring an eCreek IT team member and friendly robot against a blue technology background.

We learned a lot about AI for Colorado businesses at our first AI Simplified webinar on October 29, 2026. Colorado business owners are not asking whether to use AI anymore. They are asking how to use it safely, which tool to pick, and how to get past rewriting emails.

eCreek CTO Colter Strahan and Director of Business Development Elan Jones spent an hour with local business leaders covering what goes wrong with AI, how to do it right, and what is now possible for a 10 to 20 person shop. Here is what we learned about putting AI to work in Colorado businesses, and what we are doing about it.

Key takeaways:

  • Most employees already use AI at work, often on personal accounts the business cannot see.
  • The 4 D’s of AI fluency (Delegation, Description, Discernment, Diligence) work in any AI tool.
  • A simple Stoplight Rule tells your team what data is safe to put into AI.
  • The best way to stop personal AI use is to give people a better approved tool.
  • The biggest wins come from rebuilding a workflow, not polishing a paragraph.

In this article:

Lesson 1: Your team is already using AI. Do you know where?

Shadow AI is employees using AI tools the business has not approved, often on personal accounts, with no record of what company data went in. Microsoft and LinkedIn found that 78% of AI users bring their own AI tools to work, and that was in 2024.

Elan’s view in the session: the real number today is close to everyone. AI now shows up at the top of every Google search, so people use it without thinking of it as “using AI.” As Colter put it, shadow AI is the new shadow IT.

The mistakes are not just big-company problems. A few real examples came up:

Lesson 2: The 4 D’s of AI fluency make any AI tool work better

The 4 D’s of AI fluency are Delegation, Description, Discernment, and Diligence, a framework from Anthropic’s AI Fluency: Framework & Foundations course, developed with professors Rick Dakan and Joseph Feller. They work the same in ChatGPT, Microsoft Copilot, Claude, or any other tool.

  1. Delegation: Should AI do this, and which part?
  2. Description: Did I give it what it needs?
  3. Discernment: Is this right, and how do I know?
  4. Diligence: Would I put my name on this?

Colter walked through a quoting example. A customer asks for a quote on a job you have done a dozen times. Delegate the legwork: AI digs through your last 10 similar jobs, while you decide pricing, margin, and whether you want the job. Describe it like you would brief a new estimator: give it the customer, your three closest past quotes, your margin rules, and your terms. Discern by spot-checking one number; if it is off, do not trust the rest. Then apply diligence: your customer may not care that AI drafted it, but they will care if it is wrong.

Lesson 3: Use the Stoplight Rule to decide what data goes into AI

The Stoplight Rule sorts information into three colors so staff know what is safe to share with an AI tool. We borrowed it from one of our clients because it is easy to remember.

  • Green: General, public, or non-sensitive content. Use freely.
  • Yellow: Names, client or business names, and other sensitive details. Anonymize first.
  • Red: Passwords, personal information, HIPAA data, financial data. Never put it in.

Alongside the Stoplight Rule, we recommended four habits for any business:

  1. Adopt one company-approved AI tool instead of a mix of personal accounts. Our ChatGPT vs. Claude vs. Copilot comparison can help you choose.
  2. Write an AI acceptable use policy and have staff sign it.
  3. Close AI access when employees leave, just like email.
  4. Know your AI scope: who is using AI, for what, and where.

Want a deeper framework for managing AI risk? The NIST AI Risk Management Framework is a free, voluntary guide, and our AI governance team can help you put it into practice.

Lesson 4: AI for Colorado businesses is more than rewriting emails

The biggest opportunity in AI for Colorado businesses is rebuilding a workflow, not polishing a paragraph. Most businesses use AI for drafting emails, summarizing meetings, and writing social posts. The bigger wins are replacing the Monday spreadsheet with a live dashboard, getting two systems that do not talk to finally connect, or replacing rented software that never quite fit.

In the session we walked through sample AI Opportunity Blueprints. For a manufacturer, that meant an order-to-ship dashboard pulling from its ERP and SharePoint every hour, with late jobs rising to the top. For a nonprofit, it meant a grant outcomes dashboard that drafts reports from case management data.

We also shared what eCreek has built for itself, because we use what we sell:

  • A custom CRM that replaced a paid subscription, including our own email deliverability system that replaced a service costing about $800 a month.
  • A service desk dashboard pulling about 10 systems together. A vendor quoted $500 per connector, roughly $5,000 a month.
  • TARS, an automated ticket triage system rebuilt in a couple of weeks, replacing a tool that cost about $1,500 a month.
  • An EOS tracker, a learning management system, a knowledge base manager, an employee pulse survey tool, and Anton, our private AI.
  • One-click root cause analysis after an outage, pulling from tickets, chat channels, and monitoring.

That is the kind of work our AI development and AI automation services do for clients.

Lesson 5: Start with one AI win you can measure

The fastest path to AI value is picking the one task you would hire someone to do and automating that first. Elan’s was a weekly sales and marketing report he disliked building, now compiled by an agent on demand.

Colter shared a story about an accounts payable manager who could not picture what AI could do for her. Her daily routine: check a shared inbox, save each transaction to a shared drive, stamp the PDF, and update the accounting software. Once that workflow was connected, she immediately saw three more tasks it could take on. One connected workflow opens the door to the next.

Easy first wins we suggested:

  • A daily brief that reads your inbox and calendar.
  • A weekly roll-up of sales and marketing numbers.
  • An order-to-ship dashboard that pulls from your ERP and file shares.

Lesson 6: Private AI keeps company data on your side of the fence

Private AI is a hosted AI environment that keeps your data inside your own secure network, with role-based access and a full audit trail. It can route to multiple models, including GPT, Claude, Gemini, and open-source models.

Colter showed how eCreek builds a private AI stack: users log into one chat portal, a model proxy routes requests, and each team gets its own sandbox. Accounting agents only see accounting data. Sales agents only see sales data. Processing can run on an in-house server or through hosted APIs, and the data stays inside the private, encrypted network either way.

He also flagged a pattern he sees in client offices: production tools running as chat artifacts inside ChatGPT or Claude. Artifacts are great for prototypes and personal productivity. Tools your business depends on need proper AI hosting and security.

What Colorado business owners asked us about AI

The live Q&A showed where business leaders are stuck. These are the questions, with the short answers.

How do I stop employees from using personal AI accounts?

You cannot stop it 100%, because everyone has AI on their phone. The best approach is giving people an approved AI tool that is more useful than their personal one, connected to the company data they need, backed by a signed AI acceptable use policy. As Colter said, if you did not give staff company email, they would use Gmail. AI works the same way.

How should a small team pick one AI tool to standardize on?

Get the team together and decide now, instead of defaulting to whoever adopted first. Choose a tool where the business controls user accounts: an organizational ChatGPT or Claude plan, or a private AI environment. If you are starting fresh, start with private AI.

Is a business account on ChatGPT or Claude safe enough?

It is safer than free accounts, which may use your data to improve models. Business plans say they do not train on your data, but that data is still stored outside your environment. Our rule of thumb: if you are subject to a compliance framework or legal requirement, keep regulated data out of public AI providers and use a privately hosted environment. See our guides for healthcare teams and law firms.

How do I connect AI to systems with Social Security numbers or financial data?

Use a private AI stack with separate, permissioned sandboxes per team, so agents only reach the data that team is allowed to see. The data stays inside your private, encrypted network.

When should I not use AI?

Be careful using AI for work you do not understand yourself. You need enough expertise to apply discernment and diligence to the output. Not sure about a specific task? Our free Bot or Not? game helps you decide whether AI, a person, or both should handle it.

Why does AI make things up, and will that ever be fixed?

AI predicts the most likely answer, so it can be confidently wrong. Larger models tend to be more accurate but slower and more expensive. Hallucinations will probably never disappear completely, which is why discernment and diligence matter.

Do I need to be an eCreek client to work with eCreek on AI?

No. Any Colorado business can use the free eCreek AI Toolkit or request a free AI Opportunity Blueprint.

What we’re doing with what we learned

Every question in that Q&A pointed to the same need: simple, practical tools business owners can use today, without a consultant in the room. So we turned the webinar into a free resource on AI for Colorado businesses that anyone can use.

The eCreek AI Toolkit for Colorado Businesses puts four tools in one place:

  • The AI Maturity Model: a two-minute quiz that places your business on a five-level scale of AI use and suggests your next step.
  • Bot or Not? The AI Decider Game: a five-minute game, solo or with your team, that helps you decide whether AI, a person, or both should handle real workplace tasks. It is built on the 4 D’s.
  • AI Policy Toolkit: an editable, plain-language acceptable use policy covering approved tools, sensitive data, review steps, and staff training.
  • AI Prompt Guide for Small Businesses: a reusable prompt framework with worked examples and a printable cheat sheet.

Stay tuned for more on the toolkit this week.

Your next step

  • Get the toolkit: ecreekit.com/ai-toolkit
  • Watch the replay: AI Simplified, October 2026
  • Join the next session: Where Your Business Stands: The AI Maturity Model, Thursday, January 21, 2027. Register here.
  • Get a free AI Opportunity Blueprint: tell us what slows your team down, and we will come back with ideas built on the systems you already use, from QuickBooks and Microsoft 365 to Procore, Sage, and your ERP or EMR. You do not need to be an eCreek client. Book a free consult or explore our AI services.

eCreek IT Solutions is a Denver-based managed IT, cybersecurity, compliance, and AI services provider for Colorado businesses, and was named to CRN’s inaugural 2026 AI Innovators list.