Innovation Laboratory
Three streams reshaping human-AI collaboration: Voice interfaces that speak naturally, Agents that work autonomously, and Context that shapes intelligence.
Three streams are reshaping how I work and what I'm building: Voice interfaces that remove friction, AI agents that work alongside me, and context engineering that makes it all effective.
This laboratory tracks my exploration and experiments - building up expertise in public.
This work is grounded in the Fitzgerald Principle—holding opposing ideas (automation + humanity, voice + text, agents + control) in productive tension.
Voice UI
Voice is the next interface - real-time, accessible, human. No typing, no screens, just conversation.
AI Agents
From prompts to orchestrated workflows - systems that work alongside you, not just tools you use.
Context Engineering
Architecting what AI knows before it reasons - the meta-skill that makes Voice + Agents effective.
The Thesis
Voice + Agents + Context = Exponential Leverage
These three streams amplify each other. Each layer makes the others more powerful:
- •Voice makes agents accessible - no coding, just talking
- •Agents multiply what voice can accomplish
- •Context engineering makes both reliable and effective
Laboratory Focus
Building expertise through hands-on exploration and shipping experiments.
Cross-Pollination Studio
Exploring where Voice, Agents, and Context intersect with business domains and creative applications.
Cross-Pollination Studio
Where traditional business, web3 innovation, and strategic thinking intersect to create breakthrough solutions
Michelin Restaurant Recognition
Applied systems thinking to elevate culinary experience and operational excellence
35% Growth in Web3 Platform
Leveraged traditional business metrics to optimize decentralized analytics
S$1M+ Revenue Growth (2x)
Cross-domain insights from hospitality to tech venture scaling
4 Startups Co-founded
Pattern recognition across industries driving systematic venture creation
Voice AI Laboratory
Voice UI
Building up - Full ElevenLabs platform exploration
"Voice UI democratizes AI access - no typing, no screens, just conversation"
ElevenLabs Platform Mastery
Full platform exploration: voice synthesis, cloning, conversational AI agents, sound effects, and music generation
Voice-First Prototypes
Building conversational interfaces through vibecoding - Vercel deployments coming
Platform Capabilities Exploring:
Key Learnings:
- • Voice removes friction from human-AI interaction
- • Real-time synthesis enables conversational AI at human pace
- • Full audio stack (voice + effects + music) opens creative possibilities
AI Agents Laboratory
AI Agents
Claude Code mastery + learning multi-agent patterns
"AI Agents transform knowledge work - from tools you use to systems that work alongside you"
Claude Code Workflow Mastery
Active - Daily UseDeep expertise in agentic coding - context engineering, task decomposition, tool orchestration
Multi-Agent Orchestration
LearningLearning to coordinate specialized agents via ElevenLabs conversational AI and Claude Code subagents
Agent Capabilities Building:
Key Learnings:
- • Agents need context, not just prompts
- • Specialization beats general-purpose for complex tasks
- • Orchestration is the new programming paradigm
Context Engineering
The Meta-Skill
Architecting what AI knows before it reasons
Context Engineering is not "prompt engineering" - it's the broader discipline of architecting the informational environment where AI understanding emerges. This makes both Voice UI and Agent workflows effective.
For Voice
Context shapes how voice agents understand intent and maintain conversation flow
For Agents
Context determines what agents know, how they reason, and when they hand off
For Systems
Context engineering creates reliable, predictable AI behavior at scale
My Approach: Three-Hat Framework (Systems Architect / Product Manager / Implementation) for rapid context switching - validated across 3 client projects.Learn more →
Sustainability (Keeping Warm)
Sustainability
AI + Climate convergence - watching the space
Previously explored how AI scaling constraints (energy, chips, data) intersect with sustainability imperatives. Key insights now inform broader AI literacy work.
Watching:
- 👀 AI infrastructure energy efficiency
- 👀 Model selection trade-offs (capability vs carbon)
- 👀 Nuclear renaissance for AI datacenters
Key Insights (Still Valid):
Smaller Models, Smarter Choices
Specialized models often outperform larger ones for specific tasks while using 90% less energy.
Constraints Drive Innovation
Energy and chip constraints are forcing efficiency innovations that benefit everyone.
May revisit when Voice UI and AI Agents foundations are established. The convergence thesis remains valid - just deprioritized for now.
Three Focus Areas
Building expertise where the future is heading: Voice interfaces, autonomous agents, and the context that makes them work.
1. Voice UI Mastery
- • ElevenLabs full platform exploration
- • Voice synthesis and cloning
- • Conversational AI agents
Building up: Vibecoding voice-first prototypes
2. AI Agent Workflows
- • Claude Code daily practice
- • Multi-agent orchestration patterns
- • MCP tool integration
Validated: Three-Hat Framework across 3 projects
3. Context Engineering
- • Information architecture for AI
- • CLAUDE.md patterns
- • Knowledge systems design
Meta-skill: Makes Voice + Agents effective
My Competitive Advantage
Voice-First AI
Real-time synthesis with ElevenLabs full stack
Agent Orchestration
Claude Code mastery with multi-agent patterns
Context Engineering
Architecting what AI knows before it reasons
What I'm Exploring
Current Questions
- • How do voice + agents combine for real workflows?
- • What context patterns make agents reliable?
- • Where does vibecoding create the most leverage?
- • How do specialized agents hand off effectively?
Join the Laboratory
Interested in Voice AI experiments? Building with agents? Exploring context engineering patterns? I'm learning in public and happy to share what I'm discovering.