Start with a better question.

Artificial intelligence

Explore local and cloud inference, multimodal interfaces, developer tooling, and practical AI integrations. Useful consumer tasks, clear limits, and control over personal data guide the questions.

A question to investigate

When does running a model on a personal device offer a better experience than sending a request to the cloud?

What an evaluation should capture

Task completion, response time, memory use, uncertainty, and data leaving the device.

Connected technology

Investigate sensors, IoT, smart-home platforms, device communication, and connected ecosystems. Interoperability matters most when the network is unreliable or an account is unavailable.

A question to investigate

Can a useful smart-home routine remain understandable and dependable across different device ecosystems?

What an evaluation should capture

Discovery, response time, recovery, permissions, and behavior without internet access.

Consumer computing

Evaluate computing platforms, operating systems, hardware capabilities, peripherals, and device ecosystems through the tasks people actually perform.

A question to investigate

How do on-device capabilities change the tradeoffs between responsiveness, battery life, and useful functionality?

What an evaluation should capture

Resource use, interaction latency, compatibility, and setup complexity.

Cloud & distributed systems

Explore APIs, event-driven systems, networking, cloud infrastructure, and distributed applications. Focus on how backend decisions affect the reliability and privacy of consumer experiences.

A question to investigate

What should happen when a device reconnects after a long period offline?

What an evaluation should capture

Retry behavior, consistency, duplicate events, recovery time, and operational cost.

Human-computer interaction

Explore how people interact with software, voice interfaces, AI systems, devices, and automation. Accessible defaults and understandable system behavior are part of the design question.

A question to investigate

Can context make an interface more useful without making its actions unpredictable?

What an evaluation should capture

Keyboard access, task clarity, discoverability, reversibility, and cognitive load.