Tubegraph

Comparison of Poly and VZ

4 min read

From: gregisenberg

This article compares Poly and VZ, two AI product design tools, based on a first-time user experience with Poly and existing familiarity with VZ [00:00:00]. The comparison aims to assess their effectiveness in generating production-ready designs and front-end code from natural language prompts [00:01:10].

Poly: AI Product Designer

Poly is an AI product designer from a Y Combinator (YC) startup, touted for its ability to produce beautiful designs and production-ready code from simple English descriptions [00:00:02].

Key Features and User Experience

  • Prompt-Based Design: Functions much like ChatGPT but for design, allowing users to write what they want to build [02:45].
  • Input Methods: Supports microphone input and image uploads, enabling users to upload hand-drawn wireframes or inspiration images [03:04].
  • User Interface: Autocreates a first project upon signup [02:26]. Offers light and dark mode [08:00]. Provides options to edit elements like font, radius, and colors after generation [08:11].
  • Credit System: Free users receive 250 credits, with pages costing 50 credits and components 25 credits each [08:26].
  • Design Persistence: Keeps a history of designs on the right-hand side, and includes a rollback button [16:46].

Evaluation of Poly’s Performance

The user tested Poly by attempting to design a YouTube analytics and prediction SaaS product [04:15].

  • Initial Output: The first attempt generated only a landing page for the product, not the actual functional interface [11:08]. Buttons like “Get Started” and “See Demo” were non-functional, leading to a black screen [10:40]. The design did not achieve the requested “glassy” aesthetic [10:26].
  • Generation Time: Pages can take up to a minute or longer to generate, and the tool lacks a progress bar, making the waiting process less transparent [09:01].
  • Prompt Specificity: The tool requires precise design terms (e.g., “glass morphism”) to yield desired aesthetic results [13:33]. Without explicit design terms, the output may not align with the user’s aesthetic taste [14:17].
  • Iterative Design: It required multiple prompts to refine the design [15:52]. The second attempt yielded a more detailed product interface with relevant elements like analytics charts, although some details (like “Image” format for YouTube) were still inaccurate [23:45].
  • Strengths: The generated product interface for the analytics tool was considered “pretty cool” and “helpful,” and it included features like a cumulative engagement chart and statistical confidence metrics [23:54]. It also built out the code, which is seen as a significant advantage [27:05].

VZ: AI Product Designer

VZ is presented as a well-known product in the AI designer space [00:36]. The user has prior experience with VZ [01:18].

Key Features and User Experience

  • Transparency in Generation: VZ provides reasoning and a “thinking” process during generation, which makes the user feel more at ease compared to Poly’s “black box” approach [02:05].
  • Interactive Feedback: Allows users to ask questions and receive feedback on design elements (e.g., whether a design feels “glass-morphic”). The tool will respond and offer to enhance the design [24:42].
  • Output Quality: In a comparative test, VZ also generated a design that was very similar to the provided inspiration image [22:23]. It built out the product interface, including elements for AB testing YouTube content versions [22:37].
  • Design Nuances: It lacked the ability to hover over charts to see specific data points, a feature present in Poly’s output [26:50].
  • Overall Impression: Visually, VZ was perceived as “a little bit nicer” than Poly [27:17].

Conclusion and Recommendation

Both Poly and VZ are capable AI product designers [28:16].

  • Poly Strengths: Generates production-ready code, and its final design iteration provided a detailed and helpful product interface with unique hover functionalities on charts [23:54].
  • VZ Strengths: Offers more transparency in its generation process, provides interactive feedback, and generally feels more “at ease” to use [21:05].

The speaker’s recommendation leans slightly towards VZ due to its feedback mechanism and slightly nicer visual design [28:03]. However, the rapidly evolving nature of AI tools means that the better product today might be surpassed tomorrow [27:52].

Ultimately, to get the most out of these AI product designers, it’s recommended to use a combination of tools [28:30]. Similar to how users might leverage multiple AI tools like ChatGPT or Perplexity AI for different purposes, combining Poly and VZ could yield the best results [28:39]. This multi-tool approach allows users to capitalize on the specific strengths and nuances of each platform [29:04].

Graph View