Jeffrey Tan is the Founder & Architect of RECATOOLS and a Singapore-based technologist behind RECASYS. With a career rooted in digital infrastructure, cybersecurity, and enterprise technology, he built RECATOOLS to make powerful utilities free, fast, private by design, and accessible to everyday users, developers, students, and businesses across Southeast Asia.

His editorial focus covers ASEAN technology trends, cybersecurity readiness, AI enablement, digital infrastructure, and practical tools that bridge the gap between enterprise-grade capabilities and daily digital tasks. RECATOOLS reflects his belief that useful technology should be simple to access, quick to use, and built with regional context in mind.

21
Articles
AI & ML
Primary beat
May 2026
Writing since
~179 min
Total reading
All 21 AI & ML 11 ASEAN Tech 6 Cybersecurity 4

Articles · AI & ML Showing 1–10 of 11

A green banker's lamp lighting an open ledger of figures on a boardroom table at night, empty chairs around it
AI & ML

If You Put Engineers Inside Your Customers, Do You Still Have a Software Company?

The question every product leader has about forward deployment is financial, and almost nobody selling the model will answer it. Embedding humans looks like converting software gross margin into services gross margin, with a software company's valuation attached. One company has run this model for fifteen years and files quarterly accounts, so I opened the filing. The answer is more encouraging than the fear — and the mechanism behind it is the part worth copying.

31 Jul 2026 · 11 min read
A magnifying glass resting on a ledger page of handwritten figures
AI & ML

What a Forward Deployed Engineer Is, and Which Numbers to Believe

Every claim about this job travels faster than its source. The famous "729% growth" is arithmetically correct and almost universally misread — those are index points against a January 2025 baseline, not job postings, and the outlet that broke it published a correction saying so. The "95% of AI pilots fail" figure that justifies the whole hiring boom is a preliminary working paper whose own data shows off-the-shelf tools succeeding four times out of five. Here is what the role actually is, what genuinely happened this year, and which numbers survive being checked.

31 Jul 2026 · 12 min read
A thick bound report with a cracked spine lying open on a boardroom table at night, lit by a single desk lamp, empty chairs around it
AI & ML

Before You Hire Anyone to Deploy AI, Read the Report They Are Quoting

Every pitch for embedded AI engineering rests on one statistic: 95% of enterprise AI pilots fail. I downloaded the paper and read all 26 pages. It contains a complete buyer's playbook that nobody quotes, and several findings that argue against hiring anyone at all yet — that the 95% describes only custom and vendor-sold systems, that being smaller is an advantage rather than a handicap, and that the cheapest demand research available to you is already happening inside your own company for free.

31 Jul 2026 · 13 min read
An empty tiered lecture theatre at night, rows of vacant seats, a blank blackboard and a lit lectern nobody is standing at
AI & ML

Nobody Is Training the People This Job Actually Needs

I read the entry requirements of every Forward Deployed Engineer programme I could find, then read what Singapore's publicly funded system offers instead. Six of seven private programmes require that you already write software; the one university screens on degree discipline, the one thing a twenty-year industry veteran cannot go back and acquire. Meanwhile the region subsidises two-day courses that end, in their own words, at a prototype for engineering teams to expand. Nobody sells the stretch in between, which is the whole of the job.

31 Jul 2026 · 12 min read
A laptop on a steel workbench on a factory mezzanine at night, work gloves and hand tools beside it, machine shop equipment out of focus below
AI & ML

The Best Forward Deployed Engineers May Not Be Engineers

An AI practitioner in Beijing argues that the best Forward Deployed Engineers he has worked with were not computer science graduates but 行业大佬 — industry veterans who spent fifteen years inside the business before they came near a model. I tried to falsify that by reading the two job postings that matter most: Palantir's, who invented the role, and Anthropic's, hiring for it now. He is right that the credential is not the gate. He is wrong that you can skip the code. And the hiring bar moved in his direction while nobody was watching.

31 Jul 2026 · 12 min read
A darkened server hall at night, rows of cabinets receding down a concrete aisle with status lights glowing faintly.
AI & ML

Where Does My Data Actually Go?

"Do you use my data?" is two questions wearing one coat. Training asks whether your text enters a future model and is usually a setting. Retention asks how long a copy sits on the provider's systems, is usually fixed, and is usually not yours to change — so a provider can truthfully say it does not train on your data while holding a copy for months. Underneath both sits the question geography decides: whose courts can reach that copy. Checked against the vendors' own pages rather than a comparison table, which is where this got interesting.

31 Jul 2026 · 9 min read
Woodblock-style line illustration on aged ivory paper: a bundle of blank bamboo slips bound with vermilion cord, laid open on a low table beside a writing brush.
AI & ML

The Chapter Everyone Skips: 孫子兵法 and the AI Capital War

Business writing quotes 知彼知己 and misquotes it — the received text reads 百戰不殆, not the 百戰百勝 that Sun Tzu explicitly calls 非善之善者也 two lines earlier. Chapter 2, 作戰篇, is the one nobody quotes, because it is about logistics and money: 夫兵久而國利者，未之有也. Against US$660–690bn of 2026 hyperscaler capex and a BIS annual report naming an AI capex bust among the top three threats to financial stability, it is also the only chapter that describes what is actually happening. Traditional and simplified side by side, two textual witnesses, our own English, and four places where the frame breaks.

26 Jul 2026 · 19 min read
A digital network graphic illustrating data protection and privacy legislation.
AI & ML

Singapore's Regulator Just Opened the GenAI Data Question — Founders Should Read That as a Signal, Not Wait for the Fine Print

On 2 June 2026 the PDPC opened a public consultation — closing 1 July 2026 — on proposed guidelines for using personal data in generative AI. Most of the detail is still in draft, but for any SME building on AI, the direction of travel is the real signal. A founder's strategic read.

7 Jun 2026 · 6 min read
A humanoid robot rendered against a digital background.
AI & ML

Meta Business Agent Shows AI Moving Into the SME Operating Layer

Meta Business Agent is not just another chatbot launch. It signals a shift where WhatsApp, Messenger and Instagram become AI-enabled operating layers for SMEs — with new questions around trust, controls and human escalation.

7 Jun 2026 · 6 min read
Dramatic night view of Marina Bay Sands, Singapore, with glowing cityscape and reflection.
AI & ML

Agentic AI Governance Must Move From Policy Slides Into Runtime Controls

As AI agents move from experiments into business workflows, governance can no longer stay at the policy level. The next maturity step is operational control: identity, boundaries, monitoring and auditability.

6 Jun 2026 · 6 min read
Editorial Policy →