★ Whitepaper · v1.1 · Methodology Update

What the in-depth interviews changed about how we build this thing.

Whitepaper v1.0 (Interim) proposed the cassette as a working theory. v1.1 is the update after six weeks of in-depth interviews with TKC's directors, project managers, finance, procurement, and the M365 stack operators. The theory survived. One thing was missing: a named, repeatable ordering for how the work actually has to proceed. The interviews named it. This document is that ordering.

Whitepaper v1.0 (Interim) วางสมมุติฐานคาสเซตต์ไว้เป็นทฤษฎีใช้งานได้ v1.1 คืออัปเดตหลังการสัมภาษณ์เชิงลึกตลอด 6 สัปดาห์ — สัมภาษณ์ผู้บริหาร · PMO · ฝ่ายขาย · จัดซื้อ · การเงิน · ไอที · ทีมที่ดูแลระบบ M365 ในปัจจุบัน ทฤษฎีหลักไม่เปลี่ยน แต่สิ่งที่ขาดคือ ลำดับการทำงานที่ตั้งชื่อแล้ว ทำซ้ำได้ การสัมภาษณ์ตั้งชื่อให้แล้ว · เอกสารฉบับนี้คือ ลำดับนั้น

v1.1 · 16 May 2026 v1.0 Interim PDF → Installment 02 report → devlog → live cassette →

§ 01 · The diff

What changed from v1.0 to v1.1เปลี่ยนอะไรจาก v1.0 มา v1.1

Three things. Nothing else.

Area	v1.0 (Interim · 6 May 2026)	v1.1 (Methodology · 16 May 2026)
Methodology	Implicit — the cassette was built but the order we built it in was never spelled out.	Named. Four steps in order: Question · Streamline · Optimize · Automate. The interviews showed that organisations skip step 1 and pay for step 3.
Source basis	The five structural tensions from the original strategic diagnostic + the 4C framework.	Same basis. Plus the in-depth interview corpus — kickoff (17 Apr), management deep-dive (21 Apr), PMO alignment (14 May), platform integration (15–16 May), and the running observation log.
What it asks of the reader	"Read the theory. The cassette is the example."	"Run the four steps on whatever you're about to scope next. Question every requirement. Then build."

v1.0 stays in archive as the foundational document. v1.1 lives next to it. They are not in conflict — v1.1 names what v1.0 was already doing.

v1.0 คงไว้เป็นเอกสารพื้นฐาน v1.1 อยู่ข้างๆ ไม่ขัดกัน — v1.1 แค่ตั้งชื่อสิ่งที่ v1.0 ทำอยู่แล้ว

§ 02 · The methodology

The four steps, in orderสี่ขั้นตอน เรียงตามนี้เท่านั้น

Across every interview the same anti-pattern recurred. Organisation receives a TOR. Organisation skips ahead to Optimize. Organisation optimizes a process that should not exist. The result is an expensive system delivering a feature nobody needs at a quality nobody can verify.

จากการสัมภาษณ์ทุกครั้ง · เจอ anti-pattern ซ้ำๆ — องค์กรรับ TOR · ข้ามไป Optimize ทันที · สุดท้าย optimize สิ่งที่ไม่ควรมีอยู่ก่อน ผลคือระบบราคาแพงที่ส่งฟีเจอร์ที่ไม่มีใครใช้ · ในคุณภาพที่ไม่มีใครตรวจสอบได้

The four steps below are the antidote. They must run in this order, not any other. Skipping or reordering them is the failure mode every interview confirmed.

Step 01

Question Requirement

ตั้งคำถามกับสิ่งที่ขอ — ก่อนทำ

Read the TOR clause by clause. For every requirement ask, "must this exist?" before "how do we deliver this?" Most of the cost on a 200-page TOR sits in clauses written because that's how it was written last time, not because anyone now needs it.

The TV-portal proof point: a national broadcasting TOR scoped at ฿200,000,000 over six months was reduced — by interrogating every clause and removing the ones whose purpose could not survive "why?" — to a working, hostable prototype in 45 minutes at ฿0 cost. The cuts were not magic. They were just questions.

Step 02

Streamline (Simplify)

ทำให้เบาที่สุด — เริ่มจากศูนย์

Once Step 1 has named what must exist, start from zero. Not from the existing system. From zero. What is the minimum set of components that solves the named problem? If a layer was not justified by the answer to Step 1, it does not come in.

The cassette itself is the demonstration. From zero we kept Postgres (source of truth), Sheets (human-readable mirror), and Next.js (rendering). Nothing else. No auth provider, no analytics SaaS, no tracking pixels. The reason every other "platform" feels heavy is that Streamline was skipped — every component lives there because it was already there.

Step 03

Optimize

เพิ่มประสิทธิภาพ — เฉพาะคอขวด

Now measure. Only after Step 1 has named the right thing and Step 2 has built it at minimum size, look at performance. Find the bottleneck. Optimize that single gap. Not the whole thing.

The repeat failure across every interview: organisations start at Step 3. They optimize first, before they have questioned. They speed up the wrong process. They buy a faster version of the bottleneck without asking whether the work upstream of the bottleneck should be happening at all.

Step 04

Automate

ออโตเมต — เฉพาะที่รันสะอาด 3 ครั้งติด

The last step, never the first. Automate only when the manual version has run cleanly three times in a row. Automating a process that has not stabilised bakes the instability into code that is harder to change than the process.

The cassette enforces this rule structurally: every state-change writes to Postgres synchronously and mirrors to Google Sheets fire-and-forget. The mirror runs because the underlying flow has been verified by hand first. If the manual path is broken, the automation amplifies the breakage. If the manual path is clean, automation just removes the typing.

The four steps are not a checklist. They are a sequence. Out of sequence they are worse than doing nothing.

§ 03 · From interview to step

What each interview added to the methodologyการสัมภาษณ์แต่ละครั้งเพิ่มอะไรเข้ามา

Methodology does not write itself in an office. It comes out of the room where the actual work is being done badly. Below: the interview → the finding → the step it shaped.

17 April 2026 · Kickoff

DEPA × TKC alignment · contract orientation

Finding: The contract objectives were strategic (ecosystem · capability · delivery efficiency · digital-infrastructure positioning) but no one in the room could name the operational ordering that would deliver them. This is where the absence of Step 1 first became visible — the org had requirements but had not questioned them.

21 April 2026 · Management deep-dive

TKC senior management · whole-org structured walk-through

Finding: Project work crosses departments routinely; resource visibility does not. Sales, PMO, procurement, finance, technicians each see their own column. No one sees the row. This shaped Step 2 — the cassette must be one URL, one DB, one mirror, one game world. Not a federation of tools each owned by a function.

14 May 2026 · PMO alignment

PMO director · Khun Nuntawan Phoonkerd

Finding: The PMO ships an executive dashboard PDF every quarter. The PDF is the contract surface. This shaped Step 3 — the cassette's Project Health page is now page-5 parity with that PDF, polling a single endpoint (/api/db/project-health) that feeds both the executive tiles and the per-project cards. One optimization point. Not the whole system.

15–16 May 2026 · Platform integration

Khun June (PMO) · Khun Tom (advisor) · Khun Earth (CEO) · May (M365 stack lead)

Finding 1: The org runs five parallel platforms (Microsoft Planner, SharePoint, Power Automate, Power BI, ad-hoc Excel/Slack/Jira). Each was added because the previous one was not enough. None were removed.

Finding 2: A ฿4,000 single-day trip from headquarters to a provincial governor runs through ~12 steps including ERP request, two-line approval, AirAsia booking, paper receipt return, finance reconciliation. The process is correct. The process is also why the company cannot move at competitive speed.

This shaped Step 1 + Step 4 — interrogating the ฿4,000 trip clause by clause names which steps could be removed (Step 1); automating what survives is then the last move (Step 4), not the first. The director workshop scheduled for end-May / early-June will run this exact exercise on the company's real customer journey.

§ 04 · So what

What this means for TKC, specificallyมันหมายความว่าอะไรกับ TKC โดยเฉพาะ

The director workshop is the next instance of Step 1

End-May / first-week-June, every director-level operator (BD · Sales · PMO · Procurement · Finance · IT) will be in one room for a half-day session. The first hour is the four-step framework. The remaining hours are the directors questioning the company's own customer journey from project initiation to closeout. Every step gets the question: must this exist? The Post-Its that fail the question come off the wall.

ปลายเดือนพฤษภา / สัปดาห์แรกของมิถุนา · director ทุกฝ่ายเข้าห้องเดียว · ครึ่งวัน ชั่วโมงแรกสอน 4 ขั้น · ที่เหลือ director ตั้งคำถามกับ customer journey ของบริษัทเอง — ทุกขั้นโดนคำถามเดียวกัน: ต้องมีไหม? · Post-It ที่ไม่ผ่านคำถามถูกถอดออก

The cassette is the operating example, not the contract

The cassette already runs. It is the example of what happens when the four steps are applied to a problem TKC has been working with three different platforms to solve. The contract names five installments, but the working artifact is one URL, one database, one mirror. That is what Step 2 looks like done correctly. The other surfaces (PMO Control Tower, Project Health, Formation, Match Engine, Tome) are what Step 1 names as necessary after the questioning.

The AI Unit advisory framing

The 15–16 May meeting raised the possibility of forming a TKC AI Unit with Dr Non as founding advisor. The four-step methodology answers the structural risk that unit would otherwise face: built as a function, the unit closes. Built as an engine, it compounds. The methodology is the engine framing. Whatever the unit chooses to work on, the order is Question → Streamline → Optimize → Automate. Without that order, the unit becomes the next platform sitting alongside the five already there.

ประชุม 15–16 พ.ค. เสนอตั้ง AI Unit ของ TKC · ผมเป็น founding advisor 4 ขั้นนี้คือคำตอบของความเสี่ยงเชิงโครงสร้าง — ถ้าตั้งเป็น function หน่วยงานก็จบที่ตัวเอง · ถ้าตั้งเป็น engine มันจะทบเพิ่ม 4 ขั้นคือกรอบของ engine · ไม่ว่ายูนิตจะทำอะไร ลำดับยังคงเดิม

Compounding effect of running the methodology

Per Installment 02's Pain Points report, the cassette already exposes the gap on 11 named structural pains. The methodology is the procedure for closing them individually. Each pain begins with Step 1 (must it exist?), proceeds through the rest, and ends with a Step 4 automation only after the manual answer is clean. The compounding is the point. Each closed pain reduces the surface for the next one.

§ 05 · What's next

What v1.2 will probably addv1.2 น่าจะเพิ่มอะไร

v1.2 should land after the director workshop, where the methodology hits the company's actual customer journey and either survives or gets revised. Likely additions:

Customer Journey overlay. The As-Is map from the workshop, with the questions that should be asked at each handoff. This makes the methodology operational at every gate, not just at TOR-time.
Six-axis team-efficiency signal. The provisional patent draft describes the reconciliation method that turns the four-step methodology into a measurable signal. v1.2 should explain how to read it.
The "Don't optimize Step 1 work" rule. Step 3 has a corollary that emerged in the 15–16 May discussion but is not yet documented: even within Step 3, the gain is largest when you optimize the bottleneck that is closest to a Step 1 questioned-and-kept clause, not the bottleneck that is biggest in absolute terms.
The Step 4 ratchet. Once automation lands, it should not be unwound. v1.2 should describe how to prevent regression — the cassette's Postgres-is-truth rule is one example; the methodology should generalise it.

v1.2 น่าจะออกหลัง director workshop · เมื่อ methodology ถูกทดสอบกับ customer journey จริง สิ่งที่น่าจะเพิ่ม: Customer Journey overlay · Six-axis efficiency signal · กฎ "ห้าม optimize งานของ Step 1" · Step 4 ratchet (ห้ามถอย)