Skip to content
Createfirm

AI readiness: 10 questions to ask before you invest

Ten AI readiness questions for business owners covering goals, process, data, people, risk and budget, with what a good answer looks like for each one.

By the Createfirm teamPublished 7 min read

Illustration of an AI readiness checklist: a clipboard with ten check boxes, several ticked in gold, beside a compass

Before you invest in AI, ask ten questions across six areas: goals, process, data, people, risk and budget. If you can answer each one in a sentence or two, you are ready for a first project. If you cannot, the unanswered question tells you what to fix first.

What does AI readiness actually mean?

Readiness is the set of conditions that let an AI project deliver value instead of stalling. Published frameworks agree it goes well beyond technology.

Cisco’s AI Readiness Index scores organizations on six pillars, listed in its methodology as strategy, infrastructure, data, governance, talent and culture. The NIST AI Risk Management Framework, a voluntary framework first released on January 26, 2023, organizes the risk side into four functions: govern, map, measure and manage.

Those frameworks are detailed. The ten questions below translate the same ideas for an established small or mid-sized business. NIST’s own Playbook says it is neither a checklist nor a set of steps to be followed in its entirety, so take what fits.

Goals

1. What business problem are we solving?

Microsoft’s planning guidance for AI adoption says to start with business problems before you consider AI at all. It suggests looking at where results miss expectations and where people spend time on repetitive tasks.

A good answer names an activity and an expected result in one sentence, such as “answer after-hours service inquiries so fewer leads go unanswered.” It also confirms the activity happens often enough to justify the effort.

2. How will we know it worked?

Microsoft’s executive guidance says leaders should define the business outcomes AI is expected to support and how the organization will measure the value of AI investments.

A good answer names one or two numbers you already track (response time, hours spent, missed calls, cost per lead) and their current values. Without a baseline, you cannot show improvement. For more on this, see how to measure the ROI of AI projects.

Process

3. Can we describe the current process step by step?

AI works on a process, so you need to know what the process is. Microsoft’s guidance distinguishes between work where the same input should always lead to the same result and work where inputs are unstructured and some variation in output is acceptable. That distinction shapes which kind of tool fits.

A good answer is a written list of steps, who does each one, which systems are touched, and where the exceptions are. If two employees describe the process differently, settle that first.

4. Where does a person stay involved?

The NIST framework asks organizations to define and document processes for human oversight, and to assign responsibility for disengaging a system that behaves in ways inconsistent with its intended use.

A good answer lists what the AI may do on its own, what needs human approval, and when it hands off to a person. It also names who can switch it off.

Data

5. Do we have the data this use case needs, and is it accurate?

Microsoft’s planning guidance recommends documenting data sources, formats, quality and accessibility, and notes that data quality directly affects AI performance. Data is also where the gap between leaders and everyone else shows up in Cisco’s research: the 2025 index reports that 76% of its top-scoring group, which it calls Pacesetters, have fully centralized data, compared with 19% overall.

A good answer points to the specific documents or records, confirms there is one current version of each, and names the owner. How to prepare your business data for AI walks through the steps.

6. Who is allowed to see that data?

The NIST framework maps legal and other risks from third-party software and data. For a business, that means knowing what you are handing to a vendor and what an AI tool can reach internally.

A good answer states which data the system can read, which it cannot, whether any of it is personal information, and what the vendor does with it.

People

7. Who owns this, and do they have the time and skills?

Under its Govern function, the NIST framework calls for clear, documented roles and responsibilities and for personnel to receive AI risk training. Microsoft’s guidance lists three ways to close skill gaps: train your people, hire, or work with a partner.

A good answer names one accountable person, the hours they can give each week, and how the people who will use the tool will be trained. Microsoft also recommends pairing access to approved AI tools with training.

Risk

8. What could go wrong, and how much risk will we accept?

The NIST framework asks organizations to determine and document their risk tolerance, and to weigh expected benefits against potential costs, including non-monetary ones. It also includes feedback processes so users can report problems.

A good answer lists the realistic failure cases (a wrong answer to a customer, a missed handoff, exposure of private data), what each would cost, and what you will do to limit it. It includes a way for staff and customers to flag problems. For a fuller treatment, see AI risk management for small business.

9. Have we considered a simpler option?

The NIST framework’s Manage function includes considering viable non-AI alternatives. Microsoft makes a related point: choose the solution that meets your requirements with the least operational responsibility, because greater customization generally requires more specialized skills and more ongoing ownership.

A good answer explains why AI fits this problem better than a form, a template, a rule-based automation or a process change, and why the option you chose is no more complex than it needs to be.

Budget

10. What is the full cost, and where do we start small?

Budget covers more than the license. Microsoft’s guidance notes that different AI solutions have distinct access patterns that affect cost, and recommends a proof of concept to validate feasibility and value before full-scale development. It suggests starting with internal, non-customer-facing projects to limit risk, and adding 20 to 30 percent contingency time to initial estimates.

A good answer lists setup, usage or licensing, staff time, content upkeep and review time. It defines a small first project with a fixed scope and a date to decide whether to expand.

How to use your answers

Score each question honestly.

Your answer What it means Next step
Clear and written down Ready Move on
Known but not documented Close Write it down and get agreement
Unknown or disputed Gap Make closing it your first task

Cisco’s research points the same way on strategy: its 2025 index reports that 99% of Pacesetters have a well-defined AI strategy, compared with 58% overall. Clear answers come before strong results.

You do not need ten perfect answers to begin. You need clear answers to questions 1, 2 and 5, and a plan for the rest. Note that the NIST framework is being revised, so check its page for the current version before you rely on specific wording.

The short version

AI readiness comes down to ten questions about goals, process, data, people, risk and budget. Write your answers down. Where an answer is missing, that gap is your first project, and closing it will make every later AI investment easier to judge.

Frequently asked questions

What is an AI readiness assessment?
It is a structured review of whether your business has what an AI project needs to succeed: a clear goal, a defined process, accessible and accurate data, people with time and skills, an approach to risk, and a budget that covers ongoing costs. Published examples include Cisco's AI Readiness Index, which scores six pillars.
How do I know if my business is ready for AI?
You are ready for a first project when you can name one frequent, measurable business problem, describe the current process step by step, point to the data it needs, name an owner, state what the AI may and may not do without human review, and say how you will measure the result against cost.
Is the NIST AI Risk Management Framework mandatory?
No. NIST states that the AI Risk Management Framework is intended for voluntary use. Version 1.0 was released on January 26, 2023. Its companion Playbook says it is neither a checklist nor a set of steps to be followed in its entirety, so a small business can borrow the parts that fit.
What should a small business do first before investing in AI?
Pick one business problem and write it as a short use case that names the activity and the expected result. Microsoft's planning guidance suggests starting with where results miss expectations or where people spend time on repetitive tasks, then proving the idea with a small internal project before any customer-facing rollout.

Sources

Every fact in this article was checked against these pages on November 14, 2026.

  1. AI Risk Management Framework | NIST, National Institute of Standards and Technology
  2. AI RMF Core - AIRC, National Institute of Standards and Technology
  3. Playbook - AIRC, National Institute of Standards and Technology
  4. Plan for AI adoption - Cloud Adoption Framework | Microsoft Learn, Microsoft
  5. AI adoption - Cloud Adoption Framework | Microsoft Learn, Microsoft
  6. Cisco AI Readiness Index, Cisco
  7. Cisco AI Readiness Index Methodology - Cisco, Cisco

Keep reading