Driving AI Transformation at LINE MAN Wongnai

A session at LINE THAILAND DEVELOPER CONFERENCE 2026

Transcript

1. TOWARDS AN AI-DRIVEN COMPANY Driving AI Transformation at LINE MAN Wongnai Virot “Ta” Chiraphadhanakul VP of AI, LINE MAN Wongnai
2. Digitalize Everyday Life Services 10M+ Monthly active users #1 Market share in food delivery and F&B POS 1,200 Employees #1 Revenue / headcount in the industry
4. 80% of our code is now AIassisted.
5. How did we do it?
6. How did we do it? And...did we actually ship 80% faster?
7. Today’s Agenda 1 2 3 Adoption Way of working System How we got everyone on board How the work is changing Build a system that holds it together
8. 1 ADOPTION Getting everyone on board Scaling AI coding agents across engineering
9. From Pilot to Everyday Work AI coding agent adoption journey at LINE MAN Wongnai WHEN WE STARTED AFTER LESS THAN A YEAR 36 users 420 users 78% Weekly Active Users 94% Weekly Active Users
10. Our AI Usage Far Outpaces Global Peers ~30% Higher weekly usage than peers 61% Use MCP (peer median: 27.4%) 80% of our code is now AI-assisted
11. How did we scale? If you wait until people are ready, it won't happen. Start small Go top-down Inspect everything A pilot group showed us what works, and what doesn't, before we scaled. Once we were ready, it was a leadership decision. We wanted everyone on it. We measure every team's usage and provide necessary supports.
12. People don't do what you expect, but what you inspect. Make adoption visible Learn from the signal Measure the outcomes A Slack bot shares every team's usage, so everyone knows where they stand. High-usage teams show what works. Low-usage teams get the help they need. Go beyond active users: Merge Request throughput and the AI share of committed code.
13. Empowering Engineers with Best-in-Class AI Tools + Best tools for everyone Freedom to pick Always re-evaluated Premium seats for every engineer. A tool is cheaper than an engineer’s time. Engineers use what fits their workflow, and share what works. We keep one company default and re-test it. When something better ships, the default moves.
14. The bottleneck has shifted from coding to everything around it. AI is a sports car. Our SDLC was a dirt road built for horses.
15. 2 WAY OF WORKING The Shifting Roles From completing tasks to collaborative problem solving Adapted from “The 4 Patterns of AI-Native Dev” by Patrick Debois
16. From Producer to Manager Less typing, more reviewing and steering. Engineers Manage a team of agents: review AI pull requests, verify correctness, set strict constraints. PMs Let AI analyze customer conversations, cluster feedback and surface trends. Designers Direct agents to polish UX directly in the codebase, blurring the line with front-end.
17. From Implementation to Intent Describe the goal, let AI build. Engineers Write the specs and tests. Agents iterate on the code until the tests pass. PMs Swap static PRDs for raw problem transcripts and clear success criteria. Designers Define constraints, visual intent and success criteria instead of rigid wireframes.
18. From Delivery to Discovery When building is cheap, the risk is building the wrong thing. Engineers Explore several implementations in parallel and keep what works in reality. PMs Bring working prototypes to alignment meetings, not pitch decks. Designers Make rapid AI mockups, and let customers iterate on the interface themselves.
19. From Content to Knowledge Expertise becomes the edge, not just code. Engineers Capture architecture decisions and incident learnings so agents don't repeat past mistakes. PMs Bring working prototypes to alignment meetings, not pitch decks. Designers Codify taste and user review feedback into principles agents can follow.
20. AI isn't replacing us. It's changing where we create value. Producer → Manager Implementation → Intent Less typing, more reviewing and steering. Describe the goal, let AI build. Delivery → Discovery Content → Knowledge Test more ideas, faster. Expertise becomes the edge, not just code. Source: Patrick Debois, “The 4 Patterns of AI-Native Dev”
21. Everyone is now a builder. That’s exciting ... and a little scary.
22. 3 SYSTEM Towards AI-Native SDLC Building a system that enables teams to move fast, safely
23. Humans are now the slowest part of the process. So...?
24. Move Humans Off the Critical Path Keep them where judgment matters. 1 2 3 AI that collaborates AI that knows AI that ships So humans stay in their workflow, with no copy-paste. So humans don't have to provide context every time. So humans step in only where judgment matters. AI bots Queryable company · Closed loops AI Software factories
25. AI THAT COLLABORATES Meet People Where They Already Work Write code and open a merge request while the team iterates in the Slack thread.
27. AI THAT KNOWS Make the Company Queryable The whole organization, legible to AI. 01 Every important action leaves an artifact, so the company’s AI keeps learning and improving. 02 Meetings get AI notes, and communication moves out of DMs and emails into channels where agents can join. 03 Anyone can talk to the data to get metrics and insights, with access control matched to risk.
28. When the company is queryable, managers no longer need to route information.
29. AI THAT KNOWS From Open Loop to Closed Loop Feed the output back, and every cycle gets better. Open Loop Closed Loop Decide, execute, rarely measure. Monitor the output, feed it back, adjust.
30. AI THAT SHIPS Build AI Software Factories Humans define success, agents build. Automate the manual tasks and enable agent-to-agent handoffs, limiting human intervention to specific review gates. HU M A N A I AG E N TS Specs and tests Build and iterate Define what success looks like Generate code until the tests pass SHIPS
31. AI THAT SHIPS Build a System You Can Test and Trust. AI will do things you don't expect. Verification Observability Review gates Specs and tests define success before agents write any code. See what agents did, and why, on every run. Humans step in at the decisions that carry real risk.
32. With great power comes great responsibility. Smaller teams with more leverage means each person owns more.
33. TOWARDS AN AI-DRIVEN COMPANY Driving AI Transformation at LINE MAN Wongnai Virot “Ta” Chiraphadhanakul VP of AI, LINE MAN Wongnai