01 / AGENTSUNVERIFIED EDITORIAL WATCHLIST
AI agents will replace traditional apps
Agents are increasingly able to complete multi-step tasks across services, but reliability, permissions and user trust still limit full app replacement.
EXAMINE TESTS & COUNTERARGUMENTS
What supports the hypothesis
- More software is exposing actions through APIs and agent tools.
- AI assistants can already orchestrate tasks across multiple services.
- Natural-language interfaces reduce the need to learn individual app workflows.
What challenges it
- Many tasks still require deterministic interfaces and explicit user control.
- Authentication, payments and regulated workflows remain difficult to delegate safely.
- Users often prefer visual tools for complex editing and comparison.
Evidence to watch for
- Major operating systems exposing deeper agent permissions.
- A mainstream agent completing purchases and bookings end-to-end reliably.
- App stores shifting toward agent-callable services rather than standalone interfaces.
02 / DEVICESUNVERIFIED EDITORIAL WATCHLIST
The next personal computer will be an AI wearable
Wearables may absorb some phone tasks, but displays, batteries and input constraints still make a complete replacement uncertain.
EXAMINE TESTS & COUNTERARGUMENTS
What supports the hypothesis
- Always-on voice and vision models are a natural fit for wearable hardware.
- Several companies are experimenting with glasses, pins and ear-worn AI devices.
What challenges it
- Early AI wearables have struggled with battery life, latency and usefulness.
- Phones remain difficult to replace for rich visual interaction.
Evidence to watch for
- All-day battery life with useful multimodal AI.
- A major platform vendor shipping AI-first glasses at consumer scale.
03 / BUSINESSUNVERIFIED EDITORIAL WATCHLIST
AI can already run a small company with minimal human help
AI can handle substantial portions of research, marketing, support and operations, but humans still carry legal responsibility and handle exceptions.
EXAMINE TESTS & COUNTERARGUMENTS
What supports the hypothesis
- AI tools can generate sites, campaigns, support replies and operational documents.
- Agent workflows can connect multiple business systems and automate routine work.
What challenges it
- Banking, contracts, identity verification and liability still require accountable humans.
- Long-running autonomous workflows remain error-prone.
Evidence to watch for
- Businesses reporting sustained revenue with very low human intervention.
- Financial and legal services enabling tightly scoped machine agents.
04 / AGIUNVERIFIED EDITORIAL WATCHLIST
AGI is secretly operating inside major AI labs
There is no strong public evidence that a generally capable autonomous system meeting a rigorous AGI definition is secretly operating inside a lab.
EXAMINE TESTS & COUNTERARGUMENTS
What supports the hypothesis
- Frontier labs routinely test systems more capable than public releases.
What challenges it
- Extraordinary claims lack independently verifiable evidence.
- Current systems still show major reliability and autonomy limitations.
Evidence to watch for
- Credible independent documentation.
- Reproducible demonstrations of broad autonomous competence.
05 / ROBOTICSUNVERIFIED EDITORIAL WATCHLIST
Humanoid robots will become common household products this decade
Rapid progress in manipulation and embodied AI makes household robots more plausible, but cost, safety and reliability remain significant barriers.
EXAMINE TESTS & COUNTERARGUMENTS
What supports the hypothesis
- Robotics companies are improving dexterity and learned manipulation quickly.
- Foundation models are transferring language and vision capabilities into physical tasks.
What challenges it
- Homes are highly variable and unforgiving environments.
- Hardware maintenance and safety certification can slow adoption.
Evidence to watch for
- Sub-$20,000 general-purpose robots with dependable home-task performance.
- Large-scale consumer pilots from major manufacturers.