Published · By Missan Technology Solutions, part of the Missan group
Why readiness matters more than the tool
The tools are rarely the problem. A capable AI agent or automation still needs a clear process to follow, data it can read and a person who will own the result. When one of those is missing, the project drifts, and the business concludes that AI does not work.
The checklist below has six areas. For each, answer honestly, then score yourself using the traffic-light method at the end. It takes about half an hour with the right people in the room.
1. Processes: is the work suitable?
Good first candidates are repetitive, rules-based and high-volume. Ask of each candidate process:
- Does it happen often enough to matter, such as daily or weekly?
- Could a new colleague follow written steps and get it right?
- Is the input digital (email, forms, PDFs, a system record) rather than paper or phone calls?
- Are the exceptions known, so you can say when a human must step in?
If you cannot write the steps down, the process is not ready. Documenting it is the first piece of work, and it is worth doing whether or not you automate.
2. Data: quality and access
Check the data the chosen process depends on, not your whole estate.
- Where does it live, and who can access it today?
- Is it consistent enough to rely on, for example one customer record rather than three versions?
- Is sensitive information (personal, financial, contractual) identifiable and separated?
- Can a system read it through an export or an integration, or is it locked in a spreadsheet on one laptop?
Data permissions deserve special care. Tools such as Copilot work within the access your people already have, so loose sharing becomes a visible problem quickly. We explain why governance comes first on our Microsoft 365 and Copilot page.
3. Systems and integrations
- List the systems the process touches: email, CRM, ticketing, accounting, line-of-business software.
- Note which have an API or a supported connector, and which do not.
- Identify who administers each system and can approve access for an automation.
- Decide whether a service account, rather than a person’s login, will be used.
Automation that works around your existing tools is usually safer than replacing them. If a key system has no integration route, that is a risk to price in early.
4. People and ownership
Every automated process needs a named owner on the business side. That person decides what good looks like, reviews exceptions and signs off changes. Without one, nobody notices when the output drifts.
- Is there a sponsor with authority to change the process?
- Do the people doing the work today understand what is changing and why?
- Is there time in their week to review early output?
5. Risk and review
Decide in advance where a human checks the work. Drafts that a person approves carry far less risk than actions taken automatically.
- What is the worst outcome if the system gets one case wrong?
- Which decisions must always have human sign-off?
- Is there a log of what the system did and why?
- Have you agreed in writing how your data is handled, including that it is not used to train public models?
6. Success measures
Agree the measures before you build. Choose two or three you already have a baseline for: hours spent on the task, time to first response, error or rework rate. If you do not know today’s figure, measure it for a few weeks first. Otherwise a pilot can only produce an opinion, not evidence.
What usually goes wrong
- Starting too big. A company-wide programme collapses under its own scope. One process, one owner and one measure is easier to judge.
- Automating a broken process. If the manual version produces rework, the automated version produces it faster.
- No review step. Output goes unchecked until a customer notices a mistake.
- Buying licences first. Tools bought before a use case is chosen tend to sit unused.
Each of these is cheap to avoid if you spot it during the checklist rather than after the build.
A simple traffic-light self-score
Score each of the six areas for your best candidate process:
- Green: you can answer the questions with specifics, and the answers are good.
- Amber: you can answer, but there are gaps you know how to close.
- Red: you do not know, or the honest answer is no.
Mostly green: you are ready for a small, fixed-scope pilot. Mostly amber: fix the amber items first, and run a pilot alongside if the stakes are low. Any red in processes, data or ownership: pause and resolve it before spending on tooling. One red in the other areas is often manageable with a clear plan.
Use the free audit to check your thinking
Our free AI Opportunity Audit asks eight questions about how your business runs and gives you an instant readiness score with a prioritised shortlist of where AI fits first. It takes about three minutes, and you do not need to give an email address to see the result. If you want to go further, an optional call can turn the top opportunity into a fixed-fee pilot.
For what we build once you are ready, see AI automation and AI agents. Missan is the India company of the Missan group, which has been engineering since 2004, and is a Microsoft Solutions Partner for AI Business Solutions.
Questions people ask
How do I know if my business is ready for AI?
Look at the work, not the technology. If you have repetitive, rules-based tasks that are already digital, reasonably documented and owned by a named person, you are ready for a small first project. If the work lives in people's heads or on paper, document it first.
Do we need clean data before we start?
You need data that is good enough for the one process you are automating, not a perfect company-wide dataset. Pick a narrow use case, check the data it depends on, and fix only that. Waiting for all your data to be tidy usually means never starting.
Should we start with Copilot or a custom automation?
It depends on where the repetitive work sits. If it is inside Microsoft 365 (email, documents, meetings), Copilot may be the quicker route, provided permissions are in order. If it spans several systems, such as a CRM, ticketing and accounting, a custom workflow or agent is usually a better fit.
What does the free AI audit give me?
Eight questions about how your business runs, an instant readiness score and a prioritised shortlist of where AI fits first. You do not need an email address to see your result, and an optional call can turn the top opportunity into a fixed-fee pilot.
What if our score is low?
A low score is useful information, not a verdict. It usually points to a specific gap, such as undocumented processes or scattered data, that you can close before spending money on tools. Many businesses do the groundwork first and pilot later.
Related services
Keep reading
Cost & planning
Offshore vs in-house developer cost: how to compare properly
Every cost line to include on both sides, a simple spreadsheet model, and the pitfalls that distort the comparison.
Security & compliance
Microsoft 365 security checklist for MSPs
A repeatable baseline for client tenants: MFA, admin roles, email, backup, logging, offboarding and review.
White-label MSP
How to choose a white-label MSP partner
A checklist for buyers covering scope, tooling access, security, response targets, pilots and exit terms.