REFERENCE 13 / 14 · GLOSSARY · 4 MIN
Two ecosystems
Same ideas, different names. The ChatGPT↔Claude decoder — what each word means and what it's called across the aisle.
Two companies raced to ship the same ideas, and each one names its own. ChatGPT and Claude aren't different tools doing different jobs — they're two vocabularies for one underlying practice: a place to edit the AI's output beside the chat, a scoped workspace that remembers files and instructions, an agent that writes code in your repo. Learn the concept once and you understand both products. What's left per tool is just a name.
The decoder above is that dictionary, side by side. Read the concept first — the plain, vendor-free sentence beneath the names, written to outlive a rename. Then read across for what each company currently calls it. Set your platform above and your own column gets marked; leave it on both and you see the full pairing, which is itself worth sitting with once.
The rule of thumb: learn the concept once, learn the word per tool. When ChatGPT and Claude both ship "Projects" under the same name, that's confirmation you already understood the concept before either company named it. When the names diverge — as of August 2026: Artifacts against writing & code blocks, Claude Code against Codex — the divergence is mostly the label. The underlying job hasn't moved — though the trims around it (what each product retrieves, exposes, and limits) do differ and do move. Learn the concept once; verify the trims per product.
One honesty note, and it matters here more than almost anywhere else on this site: every name above is dated, not evergreen. Both companies rename and repackage constantly, so a pairing that's accurate today can be wrong by next month. That's what the date in the header is for — check it before you repeat a name out loud, and verify against the product itself if it's been a while.
Try it live: Meet both workspaces sets up ChatGPT and Claude side by side and sends you off with your own translation card.
Say it back in one line: what's worth learning once, and what do you look up per tool, every time?