Dev reviews 15 AI predictions for 2026: 6 right, 2 wrong, security and memory moats confirmed
tibo_maker · x · 2026-10-25
Developer Thibo (215k followers) reviewed the 15 predictions he made 10 months ago for 2026: 6 right, 2 wrong, 7 still open.
Correct:
- AI security became a mess — a scan found 5,000 vibe-coded apps leaking medical records and company docs
- AI content labeling and provenance standards never stuck
- AI moved inside the tools (code, browser, inbox, CRM); the chat box is just one entry point among many
- Memory is the moat — the AI tools he still pays for are those that remember his context
- Agent marketplaces never took off
Wrong:
- Agents still can't shop like humans — nearly every AI-assisted purchase ends with a human clicking buy
- He said gen video peaked in 2025; Revid then had its best year ever
Too early to call: whether Google overtakes everyone (half right, it's close), humanoids (great demos, nothing buyable yet), and 5 others.
More from AGI Musings
- Lex Sokolin podcast: AI, Crypto and the rise of on-chain AI agents — LexSokolin · 2026-10-25
- Shopper notes assistants now pause 5 seconds like LLMs before answering — dioscuri · 2026-10-25
- Juniors' new path to expertise: doing the verification work managers no longer have time for — danielrock · 2026-10-25
- Mathematician pushes back on doom narratives: AI isn't bulldozing pure math — littmath · 2026-10-25
- AI workshop for 100+ lawyers: watching an associate acquire a 'founder's edge' — jkubicki · 2026-10-25
- AWS CEO Andy Jassy says AI agents will force companies to rewrite almost all their code — VraserX · 2026-10-25