Plain chat, fresh each time. This is where you first feel that the thing is genuinely intelligent - and first sense that something is missing. Both feelings are correct. The missing part has a name, and the levels above are about it.
You open a chat, ask something, get an answer, close the tab. Every conversation starts from nothing, and every answer is built to satisfy anyone who might have asked that exact question. That’s not a flaw to fix later. It’s just what Level 1 is, and most people live here for years without naming it.
Hold the shift clearly, because it’s bigger than a new website: a search engine looks things up; this thinks alongside you. It drafts, weighs, explains, argues back. The old advice was to fact-check its every claim - today’s tools search the web themselves and mostly have the facts. What they cannot have is your situation. The gap you’re managing isn’t accuracy anymore; it’s aim.
Give it ten honest hours, on work that actually matters to you. Somewhere in those hours comes a late night where you suddenly see what this thing is and can’t quite sleep on it. That’s not a problem - it’s the toll booth. Everyone who gets genuinely good passes through it.
And the biggest skill at this level was never how to use it. It’s noticing when - the moments it belongs in your work, and the moments it doesn’t.
ANY TOOL
Paste the anti-sycophancy instruction as your first message in a new chat, before you ask anything else. No settings, no menus, nothing to find. It’s in the Library, and typing or pasting it in as plain text is all Level 1 requires. One honest caveat: pasted this way it holds for that conversation - the tools’ automatic memories may pick it up, may not. Making it stick every time is exactly what Level 2 is for.
ANY TOOL
Ask it something you genuinely don’t know. Quiz it like a party trick and you’ll get the party-trick version of what it can do. Bring it a real problem, with the messy details, and you’ll meet the actual tool.
ANY TOOL
Flip the interview: “Before you answer, ask me whatever you need to do this well.” One sentence, and it changes everything - the machine starts pulling your context out of you instead of guessing at it. This is the single best move at Level 1.
ANY TOOL
Take the sixty-second tour of the window. There’s a model picker (the brains, in sizes), a microphone (talking works, and often works better), a paperclip or plus for files and photos, and a sidebar holding every past chat. You don’t need any of it yet - you just need to know it’s there.
ANY TOOL
Try saying it instead of typing it. Use the voice mode built into your tool, and talk through the problem the way you’d tell a colleague. More of the real context comes out of you speaking than typing - rambling is a feature here. (When you get serious, a dictation tool like Wispr Flow makes talking your default everywhere, in every app.)
ANY TOOL
Ask it for help with itself. “What can you actually do? What are you bad at? How should I ask you things?” It answers honestly, and it’s the fastest orientation you’ll get anywhere - including here.
ANY TOOL
Notice the next time it agrees with you fast. Don’t fix it yet. Just notice it. That noticing is most of what Level 1 is training.
The same four pieces of work recur at every level - and they are your WILD Intelligence, working. The name on each card is the job; the capacity behind it is what the reps build. The framework itself ↗
The skill that outlasts every prompt trick is discernment: which moments belong to the machine, and which are yours to sit in. Wisdom here is telling leverage from erosion before you type - the email it should draft, versus the hard conversation you need to have thought your own way through. Bring it something true. Also learn what not to bring it.
Even in plain chat you can decide how you’ll be worked with: challenge me, ask before assuming. That’s Intention in miniature - your first boundary, and the first moment the tool is yours instead of everyone’s.
Anyone can try a tool once. Changing how you actually work is the Leadership rep, and it’s emotional before it’s technical. A week of honest use beats a month of reading about it.
Little to maintain yet - the Discovery rep is watching where it stuns you, where it stumbles, and the moment it agrees a little too fast. That fast agreement is a mirror with no one behind it, handing your own assumptions back as a face. Catching it is the rep that pays off at every level above.
You’re ready for Level 2 when you’re tired of explaining the same context every single conversation, when you’ve caught it agreeing with you more often than it should - and when you’ve put real hours in. The ladder pays by the rep, not by the read.