reply to self: I iterated on the interactive 2's complement teaching tool on a bit and integrated "Claude-in-claude" aka "Claudeception" Here's the result: ht

reply to self: I iterated on the interactive 2's complement teaching tool on a bit and integrated "Claude-in-claude" aka "Claudeception" Here's the result: https://claude.ai/public/artifacts/782698c8-654d-4c37-96eb-647b7b206d9e https://claude.ai/public/artifacts/782698c8-654d-4c37-96eb-6... It even gives dynamic question suggestions based on your current conversation! I even try to optimize by getting the suggestions alongside the current response. But if this fails, we use two calls to the LLM: the conversation completion happens first, then the suggestions happen in a 2nd call.