Do not buy an AI system because the technology looks impressive. Buy only after one business problem is measurable, the workflow is documented, source information is controlled, employees know the handoff rules, and a failed action has a safe fallback. Score the 20 requirements below before requesting a build.
AI readiness is an operating question—not a technology question
A business is ready for AI when it can describe the work clearly enough to control the outcome. That means the owner knows what should trigger the workflow, which facts must be collected, what the system may do, what requires approval, what counts as success, and what happens when the system is unsure. Buying software before answering those questions usually creates an expensive demonstration instead of a dependable business system.
This Fayetteville NC AI readiness assessment is for local owners deciding whether to move from experimentation into customer-facing automation. It does not ask whether the business is “innovative.” It asks whether the business has enough operational clarity, data control, staff ownership, and measurement discipline to launch safely.
How to score the assessment
Score each statement from 0 to 2. Use 0 when the requirement is missing, 1 when it exists informally or inconsistently, and 2 when it is documented, owned, and testable. There are 20 statements for a maximum of 40 points.
The 40-point Fayetteville NC AI readiness scorecard
1. Business problem
- We can name one costly or frustrating workflow.
- We have a baseline: volume, delay, errors, missed contacts, or labor time.
- We know who experiences the problem—customer, employee, owner, or all three.
- We can explain why the current process fails without blaming “people” in general.
2. Process clarity
- The workflow has a clear start and finish.
- Required information is known.
- Decision rules and exceptions are written.
- The human escalation path is named.
3. Data and access
- Source information is accurate enough to use.
- The business controls the relevant accounts and logins.
- Sensitive data is identified and limited.
- A correction and update process exists.
4. People and ownership
- One person owns the workflow.
- Employees know what AI will and will not do.
- Approvals and handoffs are assigned.
- Someone will review failures after launch.
5. Measurement and control
- Success is defined with an observable metric.
- A test set represents real customer language.
- There is a rollback or manual fallback.
- The business can export its records and results.
Interpret the score honestly
| Score | Readiness level | What it means | Next move |
|---|---|---|---|
| 0–14 | Not ready to automate | The business problem or process is too unclear. AI would hide confusion rather than remove it. | Document one workflow and establish a baseline before requesting a build. |
| 15–24 | Ready for a controlled prototype | The core need is visible, but ownership, data, or exception rules are incomplete. | Prototype with internal review; do not let the system make irreversible decisions. |
| 25–33 | Ready for a narrow production pilot | The workflow is defined and testable, with some governance gaps. | Launch one use case with limits, monitoring, and a human fallback. |
| 34–40 | Ready for a managed rollout | The business has clear rules, owners, data, and measurement. | Expand only after the first workflow proves reliable under real conditions. |
A high score is not permission to automate everything. It means the business can choose a narrow use case and control it. A low score is not a failure. It is an early warning that process work will create more value than another subscription.
Five red flags that override the total score
Even a business with a strong total should pause when one of these conditions exists:
- No owner for the workflow. When everyone is responsible, nobody updates the knowledge, reviews failures, or answers escalation questions.
- The system must guess to be useful. If prices, service areas, availability, policies, or qualification rules are not settled, automation will create inconsistent answers.
- The business does not control the account. A vendor-controlled phone number, domain, calendar, advertising account, or data store can become an exit trap.
- There is no manual fallback. Customers need a valid next step when an integration, provider, internet connection, or automation fails.
- Success means “it feels good.” A pilot needs a measurable result such as response time, complete lead records, booked requests, reduced duplicate entry, or fewer unresolved calls.
Readiness is not the same as urgency
A painful problem can be urgent and still not be ready for automation. The fastest route may be one week of process cleanup followed by a smaller, safer build.
Choose a first use case with the readiness triangle
The best first project sits where three conditions overlap: the work happens often, the rules are clear, and failure is recoverable. Routine customer questions, structured lead intake, after-hours message capture, appointment requests, review follow-up, and internal document search often fit that triangle when the business information is accurate.
| Candidate workflow | Frequency | Rule clarity | Failure recoverability | Pilot judgment |
|---|---|---|---|---|
| Answer approved service questions | High | High when knowledge is maintained | High—handoff to staff | Strong first candidate |
| Collect estimate-request details | High | High with a required-field checklist | High—owner reviews before quoting | Strong first candidate |
| Automatically issue custom prices | Medium | Often low | Low if customer acts on a wrong quote | Delay unless pricing rules are fixed |
| Diagnose a mechanical, medical, legal, or safety problem | Variable | Low and high-risk | Low | Do not automate as advice |
| Confirm appointments through a connected calendar | High | High when integration and rules are tested | Medium—must expose failed writes | Good after technical validation |
The existing guide on choosing what a Fayetteville business should automate first goes deeper on prioritization after this readiness score is complete.
A 10-business-day readiness repair plan
Days 1–2: Record the current process
Follow five to ten real transactions from start to finish. Capture delays, missing information, duplicate entry, exceptions, and employee workarounds.
Days 3–4: Define the decision rules
Write the approved answers, required fields, routing rules, prohibited claims, approval points, and conditions that force a human handoff.
Days 5–6: Fix source data
Correct service lists, hours, pricing boundaries, location details, calendars, contact routes, staff assignments, and account access.
Days 7–8: Build the test set
Create normal, vague, combined, urgent, hostile, misspelled, after-hours, and unsupported requests using real customer language.
Days 9–10: Set measurement and fallback
Record the baseline, choose two or three pilot metrics, name the reviewer, define the fallback, and decide what would stop the pilot.
What local readiness looks like in Fayetteville
Fayetteville businesses often serve customers who are moving, deploying, commuting, working irregular hours, managing rental property, repairing homes, maintaining vehicles, or trying to book service quickly. That makes speed valuable, but speed without local operating rules is dangerous. Service areas, gate access, travel fees, emergency boundaries, appointment windows, seasonal capacity, and who is actually on call must be defined before a system speaks for the business.
The City of Fayetteville describes current economic-development programs aimed at supporting small businesses, private investment, job creation, and corridor revitalization. That local growth focus makes operational discipline more important, not less: a business should know whether an AI project increases capacity, improves service, or simply adds another tool. See the City’s official small-business resource page.
One-page readiness worksheet
Complete this before requesting a proposal
- Workflow:
- Current monthly volume:
- Current failure or delay:
- Required information:
- Approved actions:
- Actions requiring approval:
- Human escalation owner:
- Success metric:
- Manual fallback:
- Stop condition:
The owner interview that exposes false readiness
Before approving a pilot, ask the owner or process leader to answer these questions without a vendor in the room. The answers should be specific enough that two employees would make the same decision. If the response depends on “common sense,” “it varies,” or “we will know when we see it,” the rule is not ready yet.
| Question | Evidence of readiness | Evidence of a gap |
|---|---|---|
| What exact event starts the workflow? | A specific channel, request, status, date, or system event. | A broad goal such as “when a customer needs help.” |
| Which facts are required before action? | A short required-field list with validation and source. | Employees collect different information depending on who answers. |
| What may the system say with certainty? | Approved knowledge tied to an owner and update schedule. | Old web copy, scattered documents, or employee memory. |
| What may the system change? | Named records and actions with success evidence. | General permission to “handle it” or “take care of the customer.” |
| What must reach a person? | Defined exception, risk, complaint, uncertainty, or customer request. | Escalation only after the customer becomes frustrated. |
| What happens during failure? | Manual route, honest action state, employee alert, and recovery owner. | The customer receives a success message or dead end. |
| What result justifies the cost? | Baseline, target, review date, and decision threshold. | The team plans to judge adoption by impressions or usage alone. |
Keep the completed interview with the project scope. It becomes the first version of the operating rules, test plan, employee training material, and post-launch review. When an answer changes, update the rule and rerun the affected tests instead of quietly changing the system in production.
Build the minimum evidence package
A readiness score is more credible when each point is supported by evidence. Assemble one folder containing the current service and policy sources, five to ten representative customer interactions, the intake form, workflow map, role assignments, account inventory, baseline metrics, test cases, fallback instructions, and the name of the person authorized to approve launch. The package does not need to be polished; it needs to be current and usable.
This evidence prevents the same questions from being answered differently during sales, design, testing, training, and support. It also gives the business leverage: another qualified provider can understand what was intended, employees can verify the rules, and the owner can distinguish a system defect from an unresolved policy decision.
Sources and standards used
This framework uses the practical risk pattern in the NIST AI Risk Management Framework: govern the use, map the context, measure performance and risk, and manage what happens next. The scorecard is an original small-business implementation tool, not a NIST certification.
For a broader local implementation path, review Fayetteville Artificial Intelligence and the 90-day Fayetteville AI adoption roadmap.
Frequently asked questions
What score means my business is ready for AI?
A score of 25 or higher can support a narrow production pilot when no red-flag condition is present. The workflow still needs limits, testing, ownership, and a human fallback.
Can a one-person business be AI-ready?
Yes. The owner can hold several roles, but the rules, source information, approval points, measurement, and fallback still need to be explicit.
Should I clean up my process before hiring an AI company?
Yes. A vendor can help map the process, but unresolved pricing, service, routing, and ownership decisions should not be delegated to software.
Does a high score mean I should automate customer decisions?
No. High readiness means the business can control a pilot. High-risk, regulated, safety-sensitive, or irreversible decisions may still require a qualified person.
How often should the scorecard be repeated?
Repeat it before each new workflow, after major staffing or policy changes, and when a pilot expands into a new channel, location, or action.
Get the process clear before the software gets expensive.
Fayetteville Artificial Intelligence can map one real workflow, expose the missing rules, and build a controlled AI system around the business you actually run.
Editorial standard: practical, business-specific, customer-facing, and honest about limitations. Examples are illustrative unless explicitly identified as measured business data. Updated when technology, local operating conditions, or implementation standards materially change.