The best Perplexity alternatives in 2026
Perplexity is good at what it does. People leave for four reasons: privacy defaults, hidden model choice, search-box-not-workbench, cost. Best answer to each.
oran.chat blog
Notes from the workbench
Comparisons
Head-to-head looks at the AI tools and models you're choosing between.
Which AI chatbots train on your conversations?
ChatGPT, Claude, Gemini, and Perplexity all train on consumer chats by default. None train on API traffic. Where each setting lives, and how to turn it off.
Codex Sites vs v0, Bolt & Lovable: prompt-to-app
An honest four-way among prompt-to-app builders: Codex Sites hosts inside ChatGPT, while v0, Bolt, and Lovable each deploy and own code differently.
Codex Sites vs Vercel & Netlify: when to use which
An honest head-to-head: Codex Sites is prompt-to-host inside ChatGPT, while Vercel and Netlify are git-to-host production platforms.
The best AI voice assistants in 2026
Voice finally works: sub-100ms latency and native audio reasoning. A field guide to the AI voice assistants worth talking to in 2026 — and when typing wins.
AI browsers tested: Atlas vs Comet vs Dia
ChatGPT Atlas, Perplexity Comet, and Dia turn the browser into an agent that clicks and fills forms for you. Which one earns a permanent tab in 2026?
Playbooks
Workflows, prompt patterns, and ways to get more out of every chat.
Build an internal tool with Codex Sites (a dashboard)
A start-to-finish tutorial that builds an ops request dashboard in Codex Sites from one prompt, then walks save, review, deploy, and workspace access.
Controlling who can see your Codex Sites site
A security-minded guide to Codex Sites access control: every deploy is production, so set the audience before sharing, then widen from owner and admins.
Add a database and file uploads to a Codex Sites app
How to give a Codex Sites app durable data and uploads — when to reach for D1, R2, or both, how to ask for it, and where the binding names live.
Managing secrets and environment variables in Codex Sites
Where each value belongs in Codex Sites — runtime secrets in the panel, binding names in hosting.json, keys in .env — and why you redeploy after a change.
Save vs deploy in Codex Sites: review before you go live
The difference between save and deploy in Codex Sites, why every deploy is production, and how to use the save step as your review gate.
How to deploy an existing project with Codex Sites
A practical guide to hosting a project you already have with Codex Sites: the compatibility check, the deploy prompt, hosting.json, and the save-deploy loop.
Essays
Where we stand on agency, model choice, and thinking with AI.
Confidence is not correctness
The dangerous thing about a language model isn't that it's wrong — it's that it's wrong with total fluency. A short case for keeping your judgment in the loop.
Is Google search dying? What the 2026 numbers say
ChatGPT hit 900M weekly users and ~17% of global queries. Google still sends 190× more traffic. A clear-eyed read of whether search is actually dying.
Keep a model you don't agree with
GPT leads with the consensus; Claude more often surfaces the counterargument. Why the point of using more than one model is disagreement, not redundancy.
Vibe coding and the limits of describing what you want
Vibe coding became a $4.7B habit in 2026 — yet developer trust in AI code fell to 29%. What the gap between describing and knowing really costs.
The case against the all-in-one AI assistant
Why bundling chat, voice, image-gen, browsing, and agents into one product makes each one worse — and what to use instead.
The forgotten skill: knowing when not to ask AI
Some questions are worse for having been asked of an AI at all. A short list of categories where the human answer beats the model answer reliably.