Creative Design with AI: 8 Techniques to Escape Predictable Outputs

X postLenny RachitskyOpinion

Lenny Rachitsky highlights insights from former Apple design and engineering lead Anshu Chhabra (@anshuc) on unlocking AI's creative potential in design. Because LLMs are next-token predictors, they naturally default to the average; eight practical techniques aim to overcome this limitation.

Key points

  1. The core conflict: LLMs predict what typically follows, whereas exceptional design relies on unexpected choices and rule-bending.
  2. According to Chhabra, most users only tap into 1% of AI's creative capacity.
  3. Iterative prompting: Use seed strings for variety, craft more ambitious prompts, and establish positive feedback loops using subagents.
  4. Visual richness: Incorporate image and video generation into design workflows.
  5. Curation: Strip out redundant elements, eliminate identifiable AI tells, and rewrite copy by hand.

Why it matters

Designers and developers often dismiss AI tools as producing generic 'AI slop'. Chhabra's framework demonstrates that the limitation often stems from default prompting and that deliberate constraints, multi-agent workflows, and human editing are needed to generate distinctive UI/UX work.

Analysis

The central observation explains why standard generative workflows fail at high-level design: next-token prediction inherently aims for the statistical mean. Chhabra's method addresses this by forcing variance through seeds, expanding the ambition of prompts, using subagents for iterative critiques, and maintaining human oversight over text and clutter.

Original post

Lenny Rachitsky

@lennysan · 1 September 2026

I'd always thought AI was terrible at design, but after reading today's 🤯 post by @anshuc, I realized I was just doing it wrong. "AI models are capable of amazing creativity, but that creativity gets stifled. LLMs are trained to be next-token predictors: they look at a sequence t.co/MCy1d9XYiJ

3.2k likes110 replies8.6k bookmarks904.1k views

View on X
More posts in the thread (2)
  1. @HighwayBila @anshuc @clairevo is on the case
  2. @BDoma @anshuc whoa! It's learning!
Selected replies (5)
  • @BDoma @lennysan @anshuc I used the prompt, before I ever got to the part about generating a random seed, and guess what? AI went and generated a random seed! Wild. t.co/Nt8cC6tLsQ
  • @HighwayBila @lennysan @anshuc This needs to be a podcast episode.
  • @ezramechaber @lennysan @anshuc Have you seen @pwnies’s work with Diffui yet? Really something else.
  • @rcanepag @chadboyda @lennysan @anshuc I bought it. Did a quick test with GPT-5.6 Sol and asked to review the copy of a marketing page with the skill. The results were very surprising!
  • @Betraydan @SebTurn @lennysan @anshuc @jameygannon @dcbruck Do you have a link to this?

Mentioned links

Links and tools from this entry

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:Lenny Rachitsky

    Stripe Head of Design Katie Dill on Preventing Zombie UIs with AI

    Speaking at Lenny's Summit, Katie Dill, Head of Design at Stripe, addressed the risk of creating generic 'zombie UIs' with AI. She argued against merely using AI to build existing interfaces faster and presented four principles to elevate design standards and creativity.

    6.2kbookmarks301.8kviews

    AI & LLMs· Talk

  • X post:Katie Dill

    AI in Product Design: Beyond Doing the Same Things Faster

    In a talk at Lenny's Summit, Katie Dill argues that AI should be used to raise the ceiling of design quality rather than just speeding up the production of existing work.

    683bookmarks200.8kviews

    AI & LLMs· Talk

  • X post:Adham Dannaway

    Agent Design Skills: UI Polish and Accessibility for AI Agents

    Designer Adham Dannaway highlights a curated set of agent design skills created by Jakub Krehel. The collection equips AI agents with specific domain knowledge covering UI polish, typography, OKLCH colours, accessibility, layout, and UX copy.

    492bookmarks28.6kviews

    AI & LLMs· Resource

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.