A stage for systems in motion
AI Engineer Singapore brought the conference’s first Asian edition to Singapore for three days of talks, workshops and demonstrations. Kairos and 65 Labs organised the demo stage, creating a dedicated place for builders to put working systems in front of the room.
A live product makes the workflow, latency, failure modes and human hand-offs visible.
The invisible work behind a live demo
A useful demo stage gives each project enough context to be understood quickly, keeps transitions tight and leaves room for the moments when real software behaves like real software.
The format let builders show what they had made, explain the decisions underneath it and give the audience something concrete to question.
| Demo | What it showed | Focus |
|---|---|---|
| Cyber Sierra | AI-assisted workflows for cyber risk, compliance and audit operations. | Enterprise cybersecurity |
| Nebius × Solo Tech — AGIBOT X2 | Walking, dexterity and reasoning workloads using NVIDIA Sonic, GR00T and Nemotron on Nebius. | Physical AI infrastructure |
| Cliya AI — Agentic frontends | Frontend patterns that let AI agents understand and act through an application. | Agentic interfaces |
| Sapin-sapin — AI that speaks Filipino | A decade of datasets and model work for Philippine languages. | Local-language AI |
| Thesys — Generative UI | Live-generated interfaces and a frontend framework for AI-native products. | Generative UI |
| Steel — Web agents at scale | Managed browsers and production loops for agents completing complex web tasks. | Browser infrastructure |
| Voltade — AI video tools | Interfaces and MCP tools that preserve creative context across AI filmmaking workflows. | AI media |
| Byterover — Memory layer for agents | Persistent, structured memory shared across tools, models and teams. | Agent memory |
| BukuWarung — Merchant experience | Agents reducing manual work across support, logistics, CRM and engineering. | Operations automation |
| Ottodot — Company OS | A real-time operating system unifying business data, analytics and AI-assisted decisions. | Internal operations |
| Sonzai — Designing for persistence | Memory pruning and consolidation for agents that improve across long-running work. | Long-running agents |
| Clustly — Agentic micropayments | Agents using pay.sh and x402 on Solana to send instant tips on X. | Agent payments |
| Level3AI — The Intent Trap | Production lessons for designing intent hierarchies that agents classify reliably. | Agent reliability |
Proof changes the conversation
When the system is visible, conversation moves from speculation to judgment: where does it create value, what evidence supports the output, and what would need to be true before someone could trust it in production?
That is why real work and working systems remain Kairos’ preferred form of proof.