Archive
Invited speaker and panelist at finance, technology and AI events across Singapore. Topics include finance automation, AI adoption, expense management, agile finance tools and practical AI agent workflows.
Presented a practical finance workflow built around Hermes Agent and the moomoo API, turning account data into a clear view of cash and the actions behind it.
Focus: Cash visibility, finance automation, agent workflows and working with brokerage data through an API.
Hosted a live, demo-led update on the Herald and Pantheon releases and what changed for attendees already running Hermes Agent.
Focus: Bot Mode, teams of named agents, memory, scheduling, steering and a safe update path to Hermes Agent 2.0.
Spoke to a workshop with 241 attendees registered and set the prepared deck aside for a live Pi demo.
Focus: Community extensions, permission gates, a six-year notes archive moved into CouchDB and read through self-hosted Obsidian, and low-cost coding on Bitdeer AI models.
Showed how a local Qwen model and Hermes Agent can work with an Obsidian-based second brain while keeping the notes and model close to home.
Focus: Local Qwen, Obsidian, Ollama, private notes and hands-free agent workflows.
Shared an accountant's journey into deploying OpenClaw and Hermes AI agents, then moving deeper into local AI inference, with three real business workflows.
Focus: Third-party risk assessment, local document redaction, public-registry debtor alerts and choosing cloud, open-weight or local models for each workflow.
Ran our first Intermediate Hermes Agent workshop with The AI Capitol and the community, with a few participants joining from overseas. Five hours and we covered more than just installing an agent.
Focus: Hermes Agent in a Docker sandbox connected to Telegram, a private VPN so the dashboard is reachable anywhere while your data and keys stay on your own box, SOUL.md as the guardrail boundary, local AI models, workflows, multi-agent orchestration and reusable skills and tools.
Spoke at Hermes AI Agent Night: The AI Operating System at SGInnovate tonight, with thanks to Lionel from The AI Capitol and SGInnovate for hosting.
I shared how I use Obsidian as a company wiki and demoed it for sales and GTM professionals: context, meeting notes and follow-ups all in one local database. I also walked through Karpathy's LLM Wiki, modified with Dr Vivian Balakrishnan's OKF idea to save tokens as a rulebook for my agents' second brain.
Over the last few weeks the same questions kept coming up: what do I use as my second brain, and how do I handle data privacy and security. My Obsidian wiki is only reachable by devices I let onto my secure Tailscale network. For confidential data, I run local models to turn messy inputs (emails, voice notes, and so on) into a structured Karpathy-style wiki, building it from scratch on the machine with no third-party LLM or service looking at the data.
We keep seeing models go off the rails in the news, acting without a human in the loop when instructions are not bounded by clear safeguards. Full control over how data is used is not a nice-to-have, it is a necessity, and it matters most for enterprises in regulated industries.
Good to see attendees in the community already running Obsidian fully local and starting to sync devices with the free, community-built LiveSync plugin. My slide deck is here: obsidian.procss.online.
Most conversations about AI agents stop at the prompt. This session started where the real work begins, at the infrastructure that has to carry those agents once they leave the demo.
Focus: Local versus cloud inference, what actually runs on the box, GPU and memory limits, redaction before any model reads your data, and the finance workflows that prove the setup.
Presented a hands-on session on using Hermes Agent to keep finance work private. Thank you Lionel Sim (The AI Capitol) and Jessica D. for hosting the House of AI event at Mandala Club, and to the founders, builders and AI experts who shared the stage.
Focus: Redact before any AI reads it — Presidio runs locally on the machine, no private data leaves for inference, a human reviews the redacted output and every step is logged. Hermes Agent as orchestration, with the takeaway that automation only earns trust when privacy controls and human review are built in from the start.
Delivered face to face, the second edition of this five-hour practical, hands-on workshop focused on moving beyond AI theory and building a working Hermes Agent with the framework from Nous Research.
Focus: Safe local agent management, Hermes setup, LLMs and context windows, prompts, workflows, persistent memory, tools, skills and automated reporting.
Demonstrated practical AI agent workflows for finance and operations at an evening exploring how agents move from prompts to production.
Focus: Multi-currency balances, spending alerts and daily summary.
Introduced Hermes Agent to Singapore builders and operators through practical workflows for durable, model-agnostic AI agents on AWS, with real examples from finance operations and client response work.
Focus: Trust-and-verify agent operations, prompt-to-execution workflows, debtor tracking, First Gazette alerts, client questionnaire automation, Hermes vs OpenClaw and governance for production AI agents.
Shared building a repeatable cybersecurity questionnaire response tool that can draft responses factually with evidence from IT controls configured within Microsoft 365 — a practical example of how AI can help with third-party risk assessment responses.
Focus: Cybersecurity questionnaire automation, Microsoft 365 IT controls, third-party risk assessment, evidence extraction and bridging AI capabilities with custom workflows.
The first and largest community-driven Hermes Agent workshop in Singapore, delivered face to face. This five-hour practical session covered building, deploying and managing AI agents using the Hermes framework from Nous Research.
Focus: Agent installation, context management, tool integration, workflow orchestration and automated reporting.
The AI Capitol event for finance professionals building their AI tech stack. Shared practical perspectives on where Claude fits in finance workflows, from research and analysis to memos, models, reporting and document review.
Focus: Practical Claude use cases for finance teams, with human oversight and workflow discipline.
Singapore's first Hermes Agent meetup, with approximately 300 sign-ups. A practical talk covering local-first AI, agent onboarding, operating discipline, hardware, OpenClaw, Hermes and real business workflows.
Examples included clinic discovery, PII removal and local meeting transcription.
Panel discussion on why more finance teams are layering agile tools and solutions over their ERP systems.
Topic: Why more finance teams are layering agile tools & solutions over their ERP.
Panelist at the 2024 edition, covering how finance teams should harness AI for spend and cash management.
Topic: How should finance teams harness AI for spend & cash management?
A 15-slide interactive presentation covering the six core reasons to automate expense management processes.
Topic: 6 Reasons to Automate Your Expense Management Processes.