The AI Readiness Check: Core Work

Why this matters

There is enormous pressure right now to "adopt AI" from vendors, conferences, and LinkedIn posts about how competitors are already using it. Most founders respond in one of two ways: they rush to buy tools they do not need, or they avoid the topic entirely because it feels overwhelming. Both responses waste time and money.

AI is useful when it is applied to a process that already works. If your operations are inconsistent, your data is scattered across WhatsApp groups and spreadsheets, and the team does not follow a documented process, then adding AI will automate the chaos faster and at scale. That is the wrong direction.

This chapter helps you figure out where you actually stand before you spend a single dirham on AI tooling.

A founder you might recognise

A founder runs a 40 person property management company in Dubai. After attending a tech conference, she signed up for three AI tools in the same month: a chatbot for tenant enquiries, an AI writing assistant for the marketing team, and a "smart" scheduling tool. Three months later, the chatbot gives wrong answers because the knowledge base was never properly set up. The writing assistant is used by one person who was already a strong writer. The scheduling tool sits unused because the team prefers their existing WhatsApp group. The founder spent over AED 15,000 (USD 4,085) and has nothing to show for it. The tools worked as advertised. The foundation underneath them was not ready to receive them.

What the founder actually needed first was a documented complaint-handling process and a single source of truth for tenant records. Without those, no chatbot in the world could give the right answer. The failure was the sequence.

The core exercise: The AI readiness scorecard

This is an honest assessment of where the business stands. You are answering one question: is the business ready to benefit from AI, and if so, where?

Step 1. Score your process maturity. For each of the following, rate yourself 1 (not at all) to 5 (fully in place):

• Our key business processes are documented and followed consistently
• Handoffs between team members have clear checklists or triggers
• We can describe our client journey in specific steps, with concrete handoffs at each one
• New hires can learn how the business works from documentation alone, with shadowing as a supplement

Add up your score. If you are below 12, your priority is process mapping and systems work, not AI. Go back to Part 4 and strengthen your foundation first. If you score 12 or above, test your readiness assumptions in the simulator.

Step 2. Score your data readiness. Rate yourself 1 to 5 on each:

• Our client information is stored in a central system, with no copies spread across inboxes and phones
• We can pull a report on key business metrics (revenue per client, project timelines, team billable rate) within 30 minutes
• Our financial records are up to date and categorised consistently
• We have at least 6 months of historical data in a structured format that goes beyond email threads

Add up your score. If you are below 12, your priority is getting your data house in order. AI tools need structured, accessible data to deliver value.

Step 3. Identify your top three repetitive, high-volume tasks. These are tasks that are done the same way every time, happen frequently (daily or weekly), and do not require deep human judgment. Examples: sending appointment reminders, generating standard reports, categorising incoming enquiries, drafting routine client communications from a template.

Step 4. For each task, answer three questions:

• Is this task currently done the same way every time, or does it vary based on who does it?
• Would a mistake in this task cause serious damage, or is it low-risk?
• Is the input to this task already digital and structured, or does it require interpretation?

Tasks that are consistent, low-risk, and have structured inputs are your best candidates for AI. Tasks that vary, carry high risk, or require interpreting unstructured information should stay human until your process and data maturity improve.

Step 5. Make your decision. Based on your scores and task analysis, choose one of three paths:

• Not ready yet. Your scores are below 12 in either area. Focus on the earlier parts of this playbook. Come back to this chapter in 3-6 months.
• Ready for one focused pilot. Your scores are 12 or above in both areas and you have identified at least one strong candidate task. Pick the single best candidate and research tools specifically for that use case. Budget a small amount and set a 60-day trial with clear success criteria.
• Ready for broader adoption. Your scores are 16 or above in both areas, you have documented processes, structured data, and multiple candidate tasks. Consider working with an implementation partner to build an AI roadmap.

Where to focus by team size

• 10 to 19 people: Complete the scorecard. Most businesses at this size are not ready for AI beyond personal use.
• 20 to 34 people: If your process and data scores are above 12, you are ready for a focused pilot.
• 35 to 50 people: You should have at least one active AI pilot with measured results before investing further.

---

Working prompts

Use these to pressure-test the scores you just gave yourself. They are designed to expose the gap between what you wish was true and what is actually true.

Process prompts

• If a key team member left tomorrow, which AI-eligible process would suddenly become non-functional?
• Which of your processes has changed three or more times in the past six months without being rewritten?
• Where do you find the team using a workaround that is not in the documented process?
• Which step of your highest-volume process still requires a judgment call when a rule could carry it?
• If you handed your process documentation to a new hire today, which page would they get confused by first?

Data prompts

• Where is your most important client data stored, and who can access it without asking you?
• If a regulator asked for the last 12 months of transactions sorted by client type, how long would it take to produce that report?
• How many systems hold a copy of the same client record, and which one is the source of truth?
• What percentage of your operational data lives in WhatsApp threads, voice notes, or someone's inbox?
• Could you measure the impact of an AI tool with the data you have today, or would you have to start tracking from zero?

Task prompts

• Which task does the team complain about most because it is repetitive and obvious?
• Where does the team spend more than two hours a week on copy-paste work that could be templated?
• Which task is currently done by the most expensive person on the team because nobody else has the context?
• What is the one document, report, or message you find yourself writing the same way every week?
• Which task fails because of human inconsistency when human judgment is doing fine?

People and risk prompts

• Who on the team is most threatened by the idea of AI, and what would change their mind?
• Who is most excited about AI, and is that excitement grounded in your reality or in LinkedIn posts?
• If an AI tool made a mistake on the task you are considering, what would the consequence be?
• Which client would notice and complain if the response they received was generated by a tool?
• What is your fallback plan if the tool you adopt is shut down, repriced, or absorbed by a competitor in 18 months?

---

Founder exercise

Set aside 60 minutes. Bring the team member who would be most affected by an AI rollout, plus one team member who is sceptical of the idea. Their disagreement is useful data.

Part A: Score the business honestly (15 minutes)

Walk through Steps 1 and 2 of the scorecard above. Score each statement out of 5. Do not inflate. A process documented but followed by only half the team scores 2, even if the documentation alone tempts you to write 4. The score is only useful if it is accurate. Add up the totals for process and data. Write both numbers on the wall.

Part B: List the candidate tasks (15 minutes)

Brainstorm every repetitive, high-volume task in the business. Aim for at least 10. Then narrow to the three that are done most often, take the most time, and require the least judgment. Write each one on a sticky note with three columns: how often, how long, who does it.

Part C: Apply the three filters (15 minutes)

For each of the three candidate tasks, answer the three questions from Step 4. Mark each task green (consistent, low-risk, structured input), yellow (mixed), or red (variable, high-risk, or unstructured). Only green tasks are real AI candidates. Yellow tasks need process work first. Red tasks should stay with people.

Part D: Pick the path and write the next move (15 minutes)

Based on your scores and your green tasks, choose one of the three paths in Step 5. Write down what you will do in the next 30 days. If you are not ready, name the one weakness you will fix first. If you are ready for a pilot, name the task, the team member who owns it, the budget cap (a specific number such as AED 2,500 (USD 680), with no "small" or "modest" placeholders), and the success criteria you will judge it against in 60 days.

---

What success looks like

You know exactly where your business stands on the readiness spectrum. You have stopped feeling pressured to "do something with AI" and started making grounded decisions based on your actual operational maturity. If you are ready, you have one specific, well-chosen pilot in mind. If you are not ready, you know precisely what to work on first, and that clarity is worth more than any tool subscription.

The signal that you got this right: in 12 months, you can point at a specific number that improved (hours saved, response time, error rate) because of the AI choice you made. A measurable number, on a dashboard the team trusts.

The current stack we recommend (as of April 2026)

The playbook is principles-first because tools shift fast. But founders ask what to actually use, so here is the short answer for April 2026. Full detail lives in Appendix E.

Claude is the model. Anthropic ships three sizes. Opus 4.7 for hard reasoning and anything that touches money or strategy. Sonnet 4.6 for the everyday work: drafting, summarising, client emails, the bulk of team usage. Haiku 4.5 for high-volume simple work like classification or quick lookups. You access Claude through the chat at claude.ai, the desktop apps, or the Claude API when you want it embedded inside another tool.

Claude Code is the personal engineer. It runs in a terminal on your laptop. For a non-developer founder, it is the closest thing to having a junior CTO. The team uses it to build the small internal tools that do not need a SaaS subscription. A quote generator. A hiring scorecard. A client-onboarding checklist that fills itself in.

n8n is the workflow layer. Open source, self-hostable on a VPS, with 400-plus integrations and a built-in Anthropic node. Open source matters because you own your data, you can run it on your own infrastructure, and no vendor can switch off your operations. Pair n8n with the Claude API and you have a quiet automation layer that the team does not need to learn. It just runs in the background.

MCP is the protocol that lets Claude talk to your tools. Notion, Gmail, Calendar, Drive, Supabase, n8n. Mention it because it will be everywhere within a year, and the platforms you already use are adopting it now.

The decision is not which tool to buy. The decision is which layer of your business is ready for which tool. Run the readiness scorecard above first. Then pick the layer.

Common mistakes

1. Letting a vendor define your AI strategy. Tool vendors will always tell you that you need their product. The readiness assessment happens before any sales conversation, with the founder driving the questions.
2. Choosing an AI tool by capability instead of by task. Start with the task you want solved. "Reduce appointment confirmations from 2 hours per day to 15 minutes" is the kind of starting point a tool can be matched to. "We need an AI chatbot" is too abstract for any tool to land against.
3. Skipping the people side. Even when a task is a strong AI candidate, the team member who currently does it needs to understand what is changing and why. Introducing tools without bringing people along creates resistance and quiet workarounds that undermine the investment.
4. Buying multiple tools at once. Three pilots running in parallel is three failures waiting to happen. Pick one, run it for 60 days, learn what works, then move to the next one.

When to move on

This is the final foundational chapter. If your scores indicate readiness, move forward with a single focused pilot and revisit your scorecard after 60 days. A score that says you are not ready is also useful: it points at the part of the playbook that addresses your weakest area. Return there and work through it with the team.

---

ARCAS lens

This is the chapter where the People then Systems then AI sequence is named directly. People build the business. Systems make it consistent. AI accelerates what is already working. When founders skip the first two layers and go straight to the third, the result is the AED 15,000 (USD 4,085) of nothing-to-show-for that opens this chapter.

There is a story from IKEA Festival City that captures the sequence. The store ran on a system designed so that even if the power went out tomorrow, the team could still check out customers. The technology made the experience faster, and the people-first design meant the foundation could survive without it. Technology multiplied a working operation. AI works the same way. Apply it to a process that works and it makes the team faster. Apply it to a process that does not work and it makes the chaos faster.

Run the layers in the wrong order and the money spent on the third one moves the business backwards.

---

Start now: Quick self-assessment

Rate each statement from 1 (never true) to 5 (always true):

Score 24 or above: You are ready for a focused pilot. Pick one task and run it for 60 days. Score 15 to 23: There are gaps to close before AI is a sensible spend. Work through the founder exercise. Score below 15: Your biggest wins are in the earlier parts of the playbook. Come back to this chapter in 3 to 6 months.