Working claims, evidence, uncertainty, and what would change the view.

Evidence

• People already use AI as a thinking partner before choosing a tool.
• Knowledge work has too many handoffs between search, writing, scheduling, analysis, and execution.
• The bottleneck is increasingly context design, trust, and workflow fit rather than model access alone.

Uncertain about

How much autonomy people will actually trust in high-stakes work without clear controls and audit trails.

Building consequence

Products should help people move from intention to useful output without losing context in the middle.

Evidence

• AI search behavior is teaching users to expect synthesis, not only blue links.
• Credibility depends on source visibility, citations, and user-controlled verification.
• A good result increasingly includes what to do next, not only what to read.

Uncertain about

Which interfaces will earn enough trust for complex decisions: search boxes, browsers, agents, or embedded workflows.

Building consequence

Search tools should make source-backed decisions easier, faster, and more inspectable.

Evidence

• Social profiles fragment identity and push visitors into platform-specific context.
• Crawlable pages, feeds, structured data, and machine-readable summaries give work durable surfaces.
• Future agents need canonical contact, citation, and routing information from owned sources.

Uncertain about

Which agent-facing conventions will become durable standards versus useful-but-optional affordances.

Building consequence

This website should be the canonical identity graph for Jack Lau's work, writing, tools, and public context.