AI readiness

Get your data and security ready before AI touches them.

AI is only as good as the information you give it, and only as safe as the systems it connects to. Most AI projects that stall do so here, not in the technology. This is the checklist I work through with clients before anything is built.

Step 1 · Your data

Five steps to data AI can actually use.

You don't need perfect data across the whole company. You need clean, organized data for the one workflow you're starting with.

  1. 01

    Find it

    List where the data for that workflow lives: CRM, help desk, accounting, spreadsheets, inboxes, shared drives. Note who owns each one.

  2. 02

    Clean it

    Remove duplicates, fill in missing fields, and make sure the same thing is always written the same way. "CEO" and "Chief Executive Officer" should be one title, not two.

  3. 03

    Label it

    Sort past examples into the categories you want AI to recognize. Tag old support tickets as billing, technical, or product questions, and AI can learn to route new ones.

  4. 04

    Protect it

    Mark what's sensitive, strip out personal details AI doesn't need, and limit which tools and people can reach the rest.

  5. 05

    Connect and maintain it

    Connect the cleaned sources, test the results against real examples, and give someone the job of keeping the data accurate.

Data readiness questions

The questions I ask about your data.

If you can answer most of these for the workflow you want to improve, you're ready to start. If you can't, that's where the first few weeks of work should go.

  • What decisions do we want to make faster or better with this data?
  • Where does it come from, and how often is it updated?
  • Who owns it, and who is allowed to see it?
  • How complete is it? Which fields are usually missing?
  • Do our systems use the same definitions for customers, deals, and revenue?
  • Do our tools share data with each other, or does someone copy it by hand?
  • Which privacy rules apply to us, such as HIPAA or state privacy laws?
  • Would the team trust a number or a decision that AI produced from it?

Step 2 · Sensitive data

Know where your sensitive information lives.

These are the areas I flag in every business before AI is connected to anything. Each needs business-grade tools that don't train on your data, access limited to the people who need it, and personal details removed wherever they aren't required.

Employee records

Personal details, Social Security numbers, payroll, benefits, health information, and performance reviews.

Financial data

P&L, balance sheets, bank and card details, tax filings, forecasts, and unpublished pricing and discounts.

Customer data

Contact details, purchase history, payment information, support conversations, and account logins.

Intellectual property

Your methods, frameworks, product designs, research, trade secrets, and training materials.

Contracts & partners

Client and vendor contracts, NDAs, partnership terms, and data-sharing agreements.

Internal records

Emails, meeting notes and recordings, board minutes, strategy plans, audit results, and incident logs.

Step 3 · Security basics

Twelve security basics to have in place first.

None of these are AI-specific, and that's the point. Connecting AI tools to your systems gives anyone who gets in more reach, so the basics matter more, not less.

  • Multi-factor authentication on email, banking, cloud storage, and every business app, with an authenticator app rather than text codes where possible.
  • A password manager, with a unique, long password for every account and default passwords changed on every device.
  • 3-2-1 backups: three copies of your data, on two kinds of storage, one kept offsite, tested by actually restoring them.
  • Automatic updates for computers, phones, apps, routers, and firewalls.
  • Access by role. People get the data their job needs and nothing more. Admin rights are rare.
  • Offboarding that removes access the same day someone leaves, from every tool.
  • Phishing training so the team checks senders, hovers before clicking, and is wary of urgent requests.
  • A call-back rule for unusual requests to move money or share data, confirmed by phone, not email.
  • Encrypted devices with automatic screen lock and remote wipe if a laptop or phone is lost.
  • No sensitive work on public Wi-Fi without a VPN or a personal hotspot.
  • A data retention policy so old sensitive records are deleted instead of piling up.
  • An incident plan that everyone knows: who to tell, what to shut off, and who to call.

Thinking about an AI chatbot?

Six questions to answer before you buy one.

A chatbot is often the first AI project companies want, and it can cut support load quickly. It can also frustrate customers if it's rushed. These are the questions I work through with you first.

  • What share of your questions are repeats with documented answers, and what does each one cost you to answer today?
  • Which tasks could a customer finish with a bot, step by step, and which are your highest-volume ones?
  • When should the bot hand off to a person, and can you control that?
  • Can it connect to your help desk, CRM, and order systems, and how is customer data stored and used?
  • What does it say when it doesn't know? It should admit it, not make something up.
  • Can you test it on your 20 most common questions, across every channel your customers use, before you commit?

FAQ

Questions about AI readiness.

How do I know if my business is ready for AI?

A business is ready to start with AI when it knows which workflow it wants to improve, the data for that workflow is in one or a few systems and is reasonably clean, sensitive information is identified and protected, and basic security like multi-factor authentication and backups is in place. You do not need perfect data across the whole company to begin.

Does my data need to be perfect before I use AI?

No. It needs to be good enough for the one workflow you are starting with. Clean and label the data that workflow depends on, start with a person reviewing the results, and improve the data as you go.

What business data should not go into AI tools?

Be careful with Social Security numbers, health records, payroll details, passwords, bank and card numbers, unreleased financials, trade secrets, and anything covered by a confidentiality agreement. Only use these with business-grade AI tools that don't train on your data, with access limited to the people and systems that need it.

What security should a small business have in place before using AI?

At minimum: multi-factor authentication on every business account, a password manager with unique passwords, automatic software updates, regular tested backups, access limited by role, and basic phishing training for the team. These protect you whether or not you use AI, and they matter more once AI tools are connected to your systems.

Is an AI chatbot right for my business?

A chatbot is a good fit when a large share of your customer questions are repeat questions with documented answers, your help content is up to date, and you can set clear rules for when the bot hands a conversation to a person. Test it on your most common questions before you buy, and check how it handles questions it can't answer.

This checklist is general guidance, not legal, compliance, or cybersecurity advice. Regulated businesses should confirm requirements with a qualified professional.

Not sure where you stand?

The AI Operations & Profit Audit includes an AI and automation readiness review, so you know what to fix before you spend on tools.