By KYN AI Advisory Team — AI implementation specialists, Singapore
The first 30 days of a KYN engagement split in two. Days 1 to 14 are the review: an inventory of your current stack and interviews with the people who use it. Days 15 to 30 turn the review into a written plan and the start of the first build. A full build ships in under 90 days. The 30-day mark is not a delivery date. It is the point where you have a plan you can read and a slice you can click through.
What happens in the first 14 days?
The review has two parts, and neither involves building.
The stack inventory. We list every tool the team uses, who owns it, what data it holds, what it costs and what connects to what. For a typical SME this includes a CRM or a spreadsheet standing in for one, an accounting package, a shared inbox, WhatsApp, a document store, and a handful of exports that someone runs by hand each week.
The interviews. We speak to the people who do the work, not only the person who signs. A finance lead explains the month-end. A salesperson shows how a lead moves from first message to quotation. An operations manager shows the weekly report and where its numbers come from. We ask to see the actual file, not a description of it.
The output is a short list of workflows, ranked by three tests:
- How often does it run?
- How many minutes does it take each time, and who does it?
- What breaks when it goes wrong: a late invoice, a missed follow-up, a wrong figure in a board pack?
A workflow that runs daily, takes an hour and can send a wrong number to a client goes first.
What do you need to prepare?
Less than most founders expect.
- Access. A named person who can show us each tool. We do not ask for shared passwords. Where we need access to a system, it is granted to a named account that you can remove later.
- Two or three real examples per workflow. The last quotation, the last reconciliation, the last report. Real examples expose the exceptions that a description hides.
- A decision-maker for the plan. One person who can say yes, no or later to each proposal.
- A list of what you will not change. Some tools stay because the team likes them or a contract binds you. Say so early.
Budget about two hours per interviewee in the first fortnight, split into short sessions.
What do you get at day 14?
A written review with four sections.
| Section | Contents | |---|---| | Inventory | Every tool, owner, data held, monthly cost, integrations | | Workflows | Each workflow as steps, with minutes, owner and failure mode | | Proposal | What to build first, what to leave, what to retire | | Constraints | Data location, approvals required, regulatory points |
The proposal names the first system, the agents inside it and the actions that will wait for a named person. It is a plan you could give to another developer. If you decide not to proceed, you keep it.
What happens in days 15 to 30?
We set up the foundations and build one working slice.
- The cloud account. A cloud account in your name, with named users through the provider's identity service. No shared logins. We hold a role you can revoke.
- The data model. The tables for the first workflow, with row-level security so each user sees only what their role allows.
- A slice, not a screen mock-up. A real record page or a real list, reading real data, that the team can open. For a sales system this might be the lead list with a reply queue. For finance it might be the bills inbox. For operations it might be one department's morning report.
- The approval rules. Which actions run unattended, which wait for a person, and who that person is.
- The audit log. Every action recorded with the actor, from the first day, so the log is never a retrofit.
At day 30 we sit with the team and walk through the slice. Their corrections at this point are cheap. The same corrections at day 80 are not.
What is not done by day 30?
Be clear about this, because it sets the right expectation.
- Not the full build. The full build ships in under 90 days from the start. Days 31 to 90 add the remaining workflows, the agents, the integrations and the reports.
- Not full data migration. We migrate a sample first and check it against the source. The full load comes once the model is right.
- Not every agent switched on. Agents begin in draft mode, writing into a review queue. They move to unattended only for actions you allow, after you have checked the output.
- Not a handover of the retainer. The monthly retainer, with a minimum of twelve months, covers changes, monitoring and fixes after launch.
How is it priced?
A one-off build fee, then a monthly retainer of at least twelve months for changes, monitoring and fixes. The quote comes after the review, because the review is what tells us the size of the job. There is no price list. See the guide to what an AI sales agent costs per month for how running costs of the agents differ from the build and retainer.
How do you judge the day-30 slice?
Ask the team four questions while they click through it.
- Does the screen show what you look at first each morning?
- Is any field missing that you currently keep in a side sheet?
- Which action here would you not trust the system to do alone?
- What would you do differently in the next step: reorder, merge, remove?
Write the answers on the plan. They set the order of the next 60 days.
What can go wrong in the first 30 days?
Four patterns recur.
- The wrong person is interviewed. The owner describes the process as designed. The clerk describes it as run. Interview the clerk.
- Scope grows in the review. Every conversation surfaces a new wish. The ranking in the review exists to hold the line: one first system, then the next.
- The data is worse than expected. Duplicates, free-text fields, missing dates. We find this in the sample and plan the cleaning as its own task.
- Nobody owns the decision. Reviews stall when three people share the yes. Name one.
Where KYN fits
KYN runs this sequence for SMEs in Singapore, Hong Kong, Indonesia and Malaysia. It is the same sequence we used for a manufacturer that moved eleven systems onto three on one AWS account, and for KYN's own books. Everything runs in a cloud account in your name. Rights in the custom code and configuration pass to you on full payment, and you can revoke our role while everything keeps running. See infrastructure, pricing and case studies.
Frequently asked questions
Do I pay for the review?
The price is quoted after the review, as part of the build fee and retainer. We do not publish a price list. Ask on WhatsApp for how the review is handled for your case.
Will my team stop working during the review?
No. Interviews are short and use real work. Most people spend a couple of hours across two weeks, and nobody stops using their current tools.
Can I stop after the review?
Yes. The written review is yours. You can give it to another builder or act on it yourself.
Checked 30 October 2026.
Curious whether Marketing agents fits your business? Talk to KYN on WhatsApp — no forms, just a conversation.
