AI projects in progress

Can AI combined with long-term, real records help me understand myself more clearly and make steadier trade-offs across work, learning, creation, and life?

Its purpose is not to make life management more complex, but to use real actions, records, reading, AI, creation, and projects to understand myself over time and build an autonomous, rhythmic life that can sustain exploration. The system should serve real life and real projects rather than system-building itself.

• Material is scattered across Notion, Evernote, public posts, Xiaohongshu, reading notes, journals, and projects, without meaningful connections between them.
• Saved material rarely becomes personal judgement, and the reasons behind past decisions are easily forgotten.
• When work, reading, exercise, English, content, and AI exploration coexist, it is difficult to decide what matters most now.
• Each AI collaboration requires re-explaining personal context, goals, and long-term projects instead of building continuous context.
• The system must not become another large construction project; it should move real life and projects without adding excessive maintenance.
• Long-term records may also allow ideas such as a one-person sample and game-like life visualisation to develop naturally.

Once AI continuously understands the clothes I own, my body data, try-on feedback, and real wearing habits, can it help me form a stable personal style more effectively than continuing to buy new clothes?

This is not about building a more complete digital wardrobe or recommending more things to buy each day. I want to test whether AI that truly knows what I own, what I wear, what I only like in theory, and my body and daily contexts can activate my existing wardrobe, reduce duplicate purchases, and gradually clarify my style judgement.

• I already own many clothes, but new items still trigger purchases based on whether one piece looks good rather than duplication or wardrobe fit.
• Generic styling advice does not know what I own, recommends unavailable items, and cannot activate the existing wardrobe.
• Colour alone is insufficient; cut, material, weight, season, body proportions, weather, and real contexts determine whether an outfit works.
• Product and reference images are not real wear; try-ons, returns, comfort, frequency, and reasons for avoiding an item must be remembered.
• I want personal style to mean understanding myself and building repeatable outfits, not continuing to buy more clothes.

Can design knowledge, components, page patterns, usage rules, and project reviews be organised into a UI/UX capability that AI can reliably understand, retrieve, and apply?

I want AI to help organise requirements, research material, and generate drafts while the designer retains direction and final interface refinement. After each project, reusable components, page patterns, decision rules, and lessons should remain so the next project does not start from zero.

• Moving toward freelance work, remote collaboration, and personal products requires rebuilding and demonstrating combined UI and AI capability while improving solo efficiency.
• AI can read PRDs, organise sources, research competitors, structure pages, and generate drafts, but it cannot replace business understanding, direction, and professional judgement.
• What deserves preservation is not only the final interface but also core problems, open questions, information structure, accepted and rejected options, feedback, and revision reasons.
• Components, page patterns, design rules, and project reflection should become personal UI assets reusable in future work.
• Structured Figma should connect with MCP and code generation to reduce implementation loss, but only after validation through real and repeated workflows.

Can a designer's long-developed tacit judgement be learned by AI and gradually become reusable, executable design capability that can eventually participate in real project delivery?

I am not testing whether AI can quickly generate a good-looking interface, but whether a designer's accumulated tacit experience—understanding requirements, finding problems, organising information, selecting options, and making trade-offs—can be made explicit and learned and applied by AI. I also want to start by training a small model, first becoming familiar with the basic steps of model training, then gradually testing whether personal design data genuinely affects the model's judgement. The long-term goal is to explore whether an independent designer can work with an AI design twin that increasingly understands them, taking on more analysis, options, and execution while the designer focuses on direction, critical decisions, and final quality.

• General-purpose AI can already generate interfaces and analyse products, but I still need to explain my background, experience, workflow, and preferences each time.
• Design quality depends on more than drawing interfaces; it also depends on understanding requirements, unpacking the business, organising information, and making many trade-offs.
• More than a decade of design work has produced many tacit judgements, but most remain in experience rather than data a machine can learn from.
• If those judgements can be recorded and structured, AI may move from applying general design knowledge toward approximating my way of deciding.
• UI Skill, past projects, design discussions, component standards, and project reviews can already provide knowledge and data for Design Twin.
• I also want first-hand experience of training and fine-tuning language and multimodal models to understand how data and parameters shape behaviour.
• The long-term question is whether a designer working with a personal AI twin can raise the ceiling of independent client work and end-to-end delivery.