Short lessons on getting the most out of Beanly Studio — and on the thinking behind good learning design.
What Beanly is, and how a project moves from blank page to AI-ready brief.
A free Claude Artifact you run in your own Claude account — a workspace for the thinking that comes before you build a training: the analysis and the design. Here’s the problem it solves: AI can now generate genuinely slick courses, whether that’s Claude’s design tools, iSpring, or Articulate AI. But “slick” isn’t the same as “solves the problem” — feed those tools a vague brief and you get a polished course that looks great and changes nobody’s behaviour. Beanly fixes the input first: you work through the analysis and design with Beanly, a built-in AI coach, and come out with a clean, structured brief you can hand to any AI tool — or a person — to actually produce the course. Good analysis in, genuinely good learning out.
Almost the opposite. Most AI learning tools start at content generation and assume the thinking is already done. Beanly sits one step earlier: it doesn’t generate your course, it gets your analysis and design right so whatever you build next is worth building.
Instructional designers and L&D folks of every stripe — freelancers scoping client work, in-house teams, trainers, consultants, and anyone using AI to build learning who wants the foundation to be sound. New to the field? Beanly keeps you from skipping the parts that matter most. Experienced? It’s a fast, structured way to do the rigorous stuff without rebuilding your templates every time.
Jack and the Beanstalk: plant one small bean — a solid analysis — and something grows far beyond its size. Growth, a little magic, good foundations. The mascot is a magic bean with round glasses and a sprout. Yes, it blinks. Yes, on this page its eyes follow your cursor. And no, it won’t just nod along to a weak learning objective to make you happy.
Four moves, one good brief: (1) Start a project — name the challenge and who it’s for. (2) Work the analysis — answer Beanly’s questions on the gap, the audience, the behaviour. (3) Shape the design — turn that into clear decisions. (4) Hand it to AI — export a brief full of real context and build with the tool of your choice.
Beanly covers the whole project, start to finish: working out the real need, checking honestly whether training is even the right fix, shaping the design, drafting the build materials, planning the rollout, and planning how you’ll measure impact — plus a final step that packages everything into one AI-ready brief. It goes deepest at the front end (analysis and design). And no, you don’t have to use all of it — use what your project needs. Many people start with just the analysis and the brief.
Beanly is the guide inside the tool. It asks the questions a good colleague would, and pushes back when something doesn’t add up — a sparring partner, not a yes-machine. What it won’t do is write your content for you: it returns feedback, questions and suggestions, not finished work. The craft stays yours (and the output stays honest).
No. It structures your thinking and challenges weak inputs, but the decisions and the craft are yours. If anything, it frees you up for the parts only a person can do.
Yes, free — nothing to buy, nothing to install, no API keys, no account with me. It runs inside your own Claude account, so the only thing in play is your Claude plan.
You can open Beanly and use the coach on free Claude. For real work, Pro is the honest recommendation: saving your projects across sessions needs storage that Claude gives you on Pro and up, and the free tier’s tighter limits can cut a longer session short. On free you can still work and export your project as a backup; on Pro it just saves as you go.
The version you already have can’t become anything but free — here’s why. I build Beanly alongside a full-time job, and a Claude Artifact has real limits, so I won’t over-promise that it’ll keep evolving as a free Artifact forever; I’d rather keep my options open for where it goes long-term than make a promise I can’t keep. But the important part I can promise: it’s a free Claude Artifact, and once it’s open in your own Claude account you can edit it and save your own copy — so the version you have stays yours, for as long as Claude Artifacts exist in this form. I literally can’t put a paywall on the copy you already have. One honest caveat that’s out of my hands: Artifacts are Anthropic’s technology, not mine — they could change how Artifacts work, or retire them, at any time. So “forever” is only as durable as the format itself. That’s the deal, stated straight.
Everything you enter stays in your own Claude storage, private to you — nothing is sent to me or any third party beyond Claude itself. For anything beyond that, your data is governed by Anthropic’s terms for your Claude plan, not by me.
Yes — please make it a habit. Your work lives in your own Claude storage, and a reset of the artifact can wipe it. Exporting a backup is one click and a file on your machine (it’s also how free-tier users carry work between sessions). Future-you will be grateful.
Each document exports as Word, Markdown, or JSON — and to a formatted Google Doc if you’ve connected Google Drive in Claude. You can also brand exports with your own colour and logo for client-facing docs. The main thing you get is the brief: one document, ready to paste into any AI tool, that gives it the full context it needs to produce something usable instead of generic. Paste it into Claude, ChatGPT, Articulate, Rise, iSpring — whatever you build in. (Direct PDF/PowerPoint aren’t supported inside an Artifact yet; for PDF, export to Word and “Save as PDF.”)
The backbone is ADDIE, with a SAM view for those who work iteratively, plus Cathy Moore’s Action Mapping for design and Kirkpatrick’s four levels for evaluation. You don’t need to know any of them up front — Beanly explains each step as you go, with tips and a glossary. It’s opinionated in a good way: it nudges you toward observable behaviours over vague “understanding,” practice over information dumps, and honesty about what training can and can’t fix.
It’s an early public beta. English-only and light-themed for now; no direct PDF/PowerPoint/Excel export yet; storage is finite and a reset can wipe it (back up!). And the coach, like any AI, can be wrong — treat its checks as a smart second opinion, not gospel. You stay the expert.
If a project disappears, restore it from your exported backup (this is why the backups matter). For bugs and ideas, use the “Give feedback” link inside the tool — honest, rough-edge feedback is genuinely wanted during the beta.
It’s in early access — request a spot below, and this page is where the launch link and walkthroughs appear first.
Julian Brunke — a training & learning-design practitioner in Berlin (iSpring Course Creation Contest winner, webinar speaker, featured case study, mentor). Built by someone who actually does this work, not a vendor selling a platform.