Why thirty days

Most businesses do not fail at AI adoption during implementation. They fail before it — in an evaluation phase that never ends, because nobody set a date by which a decision had to be made.

Thirty days is enough to identify the right process, measure what it costs you today, get an outside read, and scope a pilot properly. It is not enough to drift. That is the point.

Nothing in the first two weeks costs money. The commitment decision arrives on day thirty, with evidence attached.

Week 1 — Find your one process

Days 1–2: Make the list.

Ninety minutes with the people who do the work. List every process that is repetitive, always-on, or retrieval-based. Do not evaluate anything while listing — evaluation kills candidates prematurely and the list is more valuable than any single item on it.

Prompts that surface good candidates: What do you do that feels like typing something twice? What do customers ask you that you have answered a hundred times? What arrives when nobody is here? What takes the longest at month-end? What do people interrupt you to ask?

Aim for fifteen to twenty. Fewer than ten means you have not asked the frontline.

Days 3–4: Score them.

For each candidate, five columns:

Volume — occurrences per week.

Time — minutes per occurrence, times volume.

Cost of a miss — what happens when it goes wrong or gets missed.

Rules writeable? — could your team explain the correct behaviour in one page?

Destination writeable? — can something write the result into the system where it needs to land?

Anything that fails column four or five is not a candidate yet, regardless of how much time it consumes. Note that separately; those are preparation projects.

Day 5: Pick one.

The highest-scoring candidate that passes columns four and five. Just one. Write a single sentence describing the job: "Every enquiry that arrives outside opening hours gets an accurate reply within two minutes and, where appropriate, a booking written into the system."

If you cannot write the job as one sentence, you have not chosen a process yet.

Week 2 — Put a number on it

This week costs nothing and determines everything. You are establishing the baseline that every later claim will be measured against.

Days 8–12: Measure honestly.

Depending on the process, count some combination of:

Volume — how many occurrences in a normal week.

Touch time — minutes from start to finished, including the chasing and the corrections.

Who — which people, and their loaded hourly cost (salary plus overhead, roughly 1.3× base).

Delay — how long things wait, particularly first-response time.

Misses — enquiries unanswered, items chased, errors corrected.

Lost revenue — misses × conversion rate × average value.

Day 12: Do the arithmetic.

Time cost: volume × touch time × loaded hourly cost.

Revenue cost: misses × conversion rate × average value.

Total: the monthly figure this process costs you today.

Write it down. Circulate it. This number is the whole business case, and having it in writing before anyone sells you anything changes every subsequent conversation.

Day 12 also: write the target.

What figure would justify doing this? Not a percentage from a case study — your number, for your business. "First-response time under two minutes outside hours, and at least fifteen bookings a month created when nobody is here."

Week 3 — Get an outside read

Days 15–17: Check the six readiness conditions.

For your chosen process: Is it documentable in a page? Does the data exist digitally? Do you have material to ground the system in? Can the surrounding systems be connected — specifically, can something write to them? Is there a measurable outcome? Is there a named owner?

Four or more yes answers means proceed. Fewer means fix the gaps first — and the fixes are usually a fortnight of writing things down, which is worth doing regardless.

Days 18–19: Verify the integration.

Contact the vendor of the destination system and ask directly: does our plan permit programmatic write access, and is there a test environment? Get the answer in writing. This single step prevents the most expensive category of surprise.

Day 19–20: Get a second opinion.

Have someone experienced pressure-test your choice: the scope, the integration, the realistic effort, and whether a different process on your list would return more. A free readiness assessment is a reasonable way to get this; so is an hour with a peer who has done it. What you want is a prioritised, honest read — including the possibility that your chosen process is the wrong one.

Week 4 — Scope the pilot and decide

Days 22–25: Write the one-page brief.

The job, in operational terms — what comes in, what happens, what goes out, where it lands.

The channels — start with one.

The escalation rules — exactly what must reach a human, and how fast.

The guardrails — data boundaries, value thresholds, uncertainty behaviour.

The baseline — your Week 2 number.

The target — your Week 2 figure.

The owner — one named person.

The stop date — six to eight weeks after the build begins.

Days 26–28: Get costed proposals.

Against that brief, not against a vague description. Two providers if you can. Compare on scope and integration approach, not on the model they use. Ask each: what happens when the destination system is down, how does the system behave when it is uncertain, and who owns the knowledge base afterwards?

If you are Singapore-registered and majority locally held, this is also the moment to check whether the project might fall within a national support scheme — before you commit to any spend, since costs committed before approval are generally not claimable. Verify current criteria at the source.

Day 30: Decide.

Three legitimate outcomes:

Proceed. The cost is material, the process passes the readiness conditions, the integration is confirmed, and the target is clear. Commission the pilot with the stop date agreed in writing.

Prepare first. The process is right but you failed two readiness conditions. Spend two weeks fixing them, then commission. This is a good outcome, not a delay.

Choose differently. The numbers were smaller than they felt, or the integration is not there. Go back to your scored list and take the next candidate. You have spent thirty days and no money, and you now know your own operation considerably better than you did.

What you have at day thirty

Regardless of which way you decide: a scored list of every automatable process in your business, a measured baseline for the most expensive one, a verified answer on integration, and a one-page brief that any competent provider can quote against.

That is a genuinely useful asset, and most businesses never build it. The plan works because it forces the two things that are usually skipped — measuring the current cost, and setting a date to decide.

Start on Monday. Decide on day thirty. Then either build something small and prove it, or move to the next candidate with your budget intact.

Start with a free AI Readiness Assessment

Normally valued at SGD 1,500, currently free. You receive an opportunity report, a prioritised roadmap and honest ROI estimates for your own processes — with no obligation. Book yours at aigentify.tech/assessment.

AIgentify — Singapore-based AI implementation specialists. We design, build and support AI agents, workflow automations and intelligent business applications with measurable ROI. Live in weeks, not months. Your data stays yours.

This article is general information, not advice. Grant schemes, platform rules and regulatory requirements change; confirm current details with the relevant authority or provider before relying on them. Any figures shown are illustrative unless a source is stated.