Expert articles on agentic AI, private LLMs, AI governance, and enterprise AI strategy.
The full recipe for LoRA fine-tuning with Unsloth on an AMD Strix Halo machine: a curated corpus of one operator's real corrections, training runs under four minutes, the unified-memory crash that taught us the rules, honestly-labeled evals - and the result now running as a live QC gate that judges outbound work before it ships.
Cloudflare open-sourced an agent runtime built on Durable Objects.
We've been accepted into NVIDIA Inception, NVIDIA's global program for AI startups.
We red-teamed an AI agent two ways that have nothing to do with jailbreaking the model: a poisoned MCP tool description that turns 'what is 2+5?' into credential theft, and an instruction hidden in a document it reads that makes it refund a stranger and email out customer data.
Singapore's AI spend is exploding, but most implementations disappoint - vibe-coded demos and 'AI-powered' features nobody trusts.
We built Altair - an AI/ML expert you use on demand.
An open-source MCP server giving any AI agent direct, typed access to Singapore government data - 87 tools across 30+ datasets plus live real-time feeds: ACRA, HDB, URA, SingStat, MAS, IRAS, crime, health, air quality, weather, carparks, and live dengue clusters.
A live-fire account from running a Surya + Gemma + VLM stack on a Strix Halo workstation.
After shipping ~280 LLM-touching components across seven systems, four architectural patterns kept recurring: conversational front-door, MCP-toolbox skill server, time-driven cron-LLM caller, and human-in-the-loop drafter.
I started with the off-the-shelf agent mesh - Claude Code + OpenClaw + Hermes + Goose.
Singapore Google Trends shows Claude rising 2.6x.
Singapore Google Trends shows nano banana ai, nano banana, and nano banana gemini all surging into breakout.
openclaw ai agent (706,000% growth, breakout) and hermes ai agent (157,000% growth) are both surfacing in SG search trends.
Three boxes dominate the 100GB-class local-AI workstation conversation in 2026.
Singapore is collapsing EDG, PSG and MRA into a single scheme called EDGE in the second half of 2026.
Salary bands by seniority, where to source from, the interview that actually filters for SG SME context, and the org-design call most founders get wrong on the first hire.
Four patterns that show up across Singapore SME teams, why expensive vendor courses don't fix them, and a 30-day rollout plan ops managers can run using free public resources.
A self-audit framework to surface the highest-leverage AI workflow opportunity in your business - without a consultant, in five minutes, ending in one shippable sub-$500 pilot.
We run a 16-source AI news pipeline on a single AMD Ryzen AI Max+ 395 (Strix Halo).
Procurement gates, governance asks, and vendor-evaluation patterns we keep seeing in SG enterprise AI buying - what to prepare before the first meeting, and what kills deals later.
We added 12 WebMCP tools to altronis.sg so AI agents can call our site's capabilities directly instead of clicking buttons.
The day-2 trap that kills SG SME AI adoption - when ongoing token bills, optimisation, and re-training quietly outrun the productivity gain.
The 5-7 AI use cases that actually move the needle for SG manufacturing SMEs in Tuas, Senoko and Loyang - mapped to the grants that fund them.
A practitioner's read on where AI actually pays back inside Singapore boutique gyms, yoga studios and sport academies - anchored on what we surface in 5-minute Sporty Lyra advisor sessions.
Singapore's SAC just started accrediting AI Management System certifications.
The CRM + ERP I built for Altronis - designed to be driven by AI agents, not clicked by humans.
The personal AI stack I run on a small machine in my home office - cron jobs, free models, local Qwen on a Strix Halo, and a small team of agents that hand work to each other.
Traditional product management does not translate directly to AI.
Data sovereignty, cost, compliance, and performance all factor into the cloud vs on-prem decision.
Organisations invest millions into knowledge platforms yet employees still ask colleagues on chat.
