← All services
ServiceAsia Pacific

AI Agent Design & Implementation

What it is

Building practical AI agents that automate repetitive workflows, reduce manual effort, and free your team to focus on what matters. From first proof-of-concept to full deployment, Francis helps operations-heavy organisations get AI working for them without the disruption.

Who it is for

Operations-heavy organisations, particularly small and mid-sized companies, that want to remove manual work with practical AI agents.

Proof from the field

  • A live presentation, Building an AI Agent from Scratch, walks through an AI agent workflow from problem framing to live deployment (June 2026). See it in the Talks section.
  • Certified Prompt Engineer (GSDC, 2025), AI Agent Certification (2025) and Business Workflow Automation with AI (NTUC LearningHub, 2025).
  • Example engagement: an AI agent for a business owner who buys items himself. He records each purchase and the agent sends it straight to the finance team, so claims worth $1,000 or more a month are no longer missed.
  • Recognition: 2nd runner-up out of 11 teams at the Singapore-Sarawak Friendship AI Business Design Hackathon 2026. Read the story.

Who you work with

Francis Lee spent 30 years at Sealed Air Corporation, rising from Engineering Assistant to Director, and is PMP-certified. APACEdge is his independent consulting practice, based in Singapore and serving clients across Asia Pacific.

In practice

The first AI agent Francis built for a client was for a business owner who was losing money because his purchases were never submitted for reimbursement. The agent records what he buys and sends it directly to the finance team. He gets the money back, and his business spending is properly tracked.

  • Job interview agent: built with Kimi and Claude and shared with job seekers, at interview.apacedge.asia. It has two levels: a standard set of questions, and a version tied to the Kimi API so the questions vary and become more challenging. Job seekers practise a real interview instead of rehearsing in their heads.
  • Food stall complaint agent: built for food stalls that are penalised by delivery platforms when customers complain about spillage or a missing chopstick or spoon. The agent scans the issue and drafts an email to the delivery platform asking for the penalty to be reversed. Stalls report being penalised on about 20 to 40% of complaints.
  • Courage AI Playground: co-facilitated with three other facilitators for a group of 20+ people, co-hosted by e2i and The Courage Chapter at the e2i hub in Ang Mo Kio, building an AI agent for job search. Participants arrived unsure what AI could do for a job search and left asking to do more with AI agents.

Lesson learned: a simple agent can win back money that would otherwise be lost, and help people prepare for an interview properly instead of improvising.

How an engagement works

  • First contact: Francis speaks with you, learns what you need and what is not working today, and suggests several options for you to review.
  • Week 1: identifying the pain points. A common issue clients arrive with is data readiness, so this is checked early.
  • What you get: it depends on the need, from straightforward verbal advice to a simplified workflow supported by an AI agent built on Claude.
  • Timeline and price: typically within a month, depending on complexity. Pricing scales with the complexity of the project, from a few hundred to several thousand Singapore dollars (SGD). Complex projects use a retainer so the solution sticks.
  • Who it is usually for: business owners, who decide whether to invest in AI.
  • Delivery: remote or on-site, in English or Chinese, and Francis will travel to any country if needed. Based in Singapore, taking projects globally.

How to engage

Engagement models: AI agent projects, advisory retainers, project-based consulting, workshop facilitation and a peer advisory network. Enquiries are answered within 48 hours.

Related services: AI Workflow & Process Simplification → Workshop Facilitation & AI Training →

Common Questions

What does AI agent design and implementation involve?

It means building practical AI agents that automate repetitive workflows and reduce manual effort, from first proof-of-concept to full deployment.

Who is AI agent design for?

Operations-heavy organisations, especially small and mid-sized companies that want to remove manual work without disrupting how they run.

How do I get started?

Use the contact form or WhatsApp button on the APACEdge homepage and describe the workflow you want to simplify. Enquiries are answered within 48 hours.

How long does a typical engagement take?

Typically within a month, depending on the complexity of the project and the needs of the client.

How much does it cost?

Pricing depends on the complexity of the project, from a few hundred to several thousand Singapore dollars (SGD). Complex projects use a retainer so the solution sticks.

What was the first AI agent built for a client?

It was an agent for a business owner who bought items himself and often missed claiming them. The agent records each purchase and sends it directly to the finance team, so he gets his money back and his spending is tracked properly.

What other AI agents has Francis built?

A job interview agent built with Kimi and Claude and shared with job seekers, with a standard level and a harder level that varies the questions, and an agent that helps food stalls scan a customer complaint and draft an email asking the delivery platform to reverse a penalty.

Has Francis taught people to build AI agents?

Yes. He co-facilitated the Courage AI Playground with three other facilitators for 20+ people, co-hosted by e2i and The Courage Chapter in Ang Mo Kio, where participants built an AI agent for job search.

Start a conversation