Ape Tutoring will launch Motiff, a design tool for the AI era, for UI designers

Text | Zhang Ziyi

Editor|Peng Xiaoqiu

36Kr exclusively learned that Ape Tutoring will launch a professional tool for UI designers - Motiff in the second half of this year, which is developed by Kangyun Holding Group. This new AI design tool is led by Yang Yuanzu, vice president of Kanyun Holding Group and president of Motiff.

Ape Tutoring's internal ambition for this product is to become "Figma's next generation product", after all, Ape Tutoring has used AI to create three popular products in the education industry: Little Ape Search Questions, Little Ape Oral Arithmetic, and Zebra.

Previously, the outside world was more familiar with Kanyun Holdings for its consumption layout, including coffee, down jackets and confinement centers. But Motiff became the first product for Ape Tutoring to involve the enterprise service track.

According to 36Kr, Motiff's technology originated from the AI lab established by Ape Tutoring in 2014, which was the industry's first artificial intelligence research institute.

Since the double reduction, according to the policy requirements, Ape Tutoring has split its business segment, and its largest product line, online live class classes, has been transformed into a non-profit organization, which is no longer included in the business map of Ape Power.

As early as July 2022, Ape Tutoring announced an internal organizational structure adjustment email and established Kanyun Holding Group Company, and the group structure also surfaced: including four major sectors: education technology, digital publishing, corporate services, and new consumer goods.

In this regard, 36Kr interviewed Yang Yuanzu, vice president of Kangyun Holding Group and president of Motiff, as the former head of production and research of Ape Counseling, he led the establishment of Ape Tutoring AI Lab, responsible for product research and development and AI technology research and development of Kangyun Holdings for a long time, and led the production and research team to continuously create products such as ape counseling and zebra.

In this interview, he answered several core questions: Why is Ape Tutoring involved in the SaaS space, and what can Motiff do? In the face of the aggressive Chat-GPT, and the revolutions of the century it could bring, how will AI change the field of design?

Motiff wants to build "AI that understands design systems"

36Kr: What was the motivation for making this product at that time?

Yang Yuanzu: In mid-2021, Microsoft released a product called GitHub Copilot, which is a product that assists in writing code. I was amazed that this product actually proved that AI is very effective in a professional production tool.

When we look at a new opportunity, we try to look at the industries that we can understand and that match our core capabilities.

For Ape Coaching, the entire development process of software and products is what we understand, not just how to develop products. Over the past decade, we have served hundreds of millions of users with products that are well aware of the pain points, needs, and improvements in the product design process. We have also always believed in the value of AI to improve products and services, especially in the past few years, we have successfully applied AI to popular products such as Little Ape Search Question, Little Ape Oral Arithmetic, and Zebra.

We realized that production tools could very well usher in a new era—the era of AI. So in the end, I chose to make a product like Motiff.

36Kr: How does AI play a role in Motiff?

Yang Yuanzu: The first simple idea was to explore the designer's personal work scene and use AI to improve efficiency. For example, the designer needs to draw a page, in order to align, even if only one pixel is added, 30 elements must be adjusted, and each element must be adjusted by one pixel, which is extremely cumbersome.

What AI can do: make these tedious processes more structured and adjusted. For example, when we want to adjust all elements by 1-2 pixels, the AI can quickly and automatically identify the elements that need to be adjusted and make structural adjustments. This is the first step of debugging.

Furthermore, we recognize that AI can exert more power in improving the overall efficiency of a design team than improving the designer's individual efficiency. For example, AI can assist teams in creating design systems; The ability to automatically extract common, common styles and components, which used to take weeks to sort out, can now be done quickly with machine learning, greatly reducing the threshold for building and using a design system.

Another potential scenario is that AI can learn the design system of the product, and after learning it, it can do mechanical design execution. For example, when AI learns the team's design system, we only need to give a relatively rough interface draft in the tool, and AI may improve this interface into an interface that meets the design system and meets production needs.

36Kr: Assuming that AI can design a design system through algorithms by itself, will it lower the career threshold for designers?

Yang Yuanzu: No. First of all, AI is not to create a design system, it is actually to use and learn the original design system, and do relatively mechanical and relatively certain work.

What AI can't do? For example, it is difficult to understand your business needs - why do I need to emphasize this element on this page at this time? AI can't do real aesthetics, nor can it accurately judge what kind of design is easier to use for users, when AI is gradually able to handle mechanical and tedious work, the remaining creativity and user-friendly work still need to be thought and completed by people. From this level, it does not lower the threshold of designers, but has higher requirements for the ability of designers.

ChatGPT cannot replace existing workflow patterns in a short period of time

36Kr: When you talk about this, I think of ChatGPT, you said that AI may do some basic functions, and now people think that ChatGPT can also replace doing some creative work.

Yang Yuanzu: After the release of ChatGPT, we tried to see if we could make ChatGPT directly do UI design. The current conclusion is that the latest version of GPT already has the ability to enter a paragraph and let it directly create a page, which can indeed output a page directly, not a UI design draft, but directly output an HTML code, or even run directly.

Going back to the essence of this matter, the problem we want to solve is to make a screen-based, interactive product from a larger workflow, this product should convey some business requirements, it should be good-looking, easy to use such core values. I think there are some things that machines have not been able to do yet, such as aesthetics, how to have a deeper understanding of this business? These are things that current AI is difficult to handle.

The current large model technology has two problems: one is that the adjustment of very detailed details is not easy to achieve, and the overall amazing performance ability that everyone sees now is still the appearance. The more detailed control behind it may be more difficult to go up.

The second is to assume that a large model, or an interactive form such as ChatGPT, generates a framework, and people make some adjustments and improvements on it, which is equivalent to the combination of the output of the large model and human creative labor. The difficulty with this model now is that it doesn't improve efficiency in the process. It is possible that a designer building an idea from scratch and then changing it with a large model may be more efficient.

But I think it (the big model) may open up a completely different new workflow, and the scenario that the original low-code platform is trying to solve may be well implemented by the big model.

The design tools market is "winner takes all"

36Kr: What is the market size for design tools?

Yang Yuanzu: Actually, there is no particularly official or authoritative data, I think it will be millions of users. The use of our tool, plus downstream, plus product managers and the corresponding front-end engineers for subsequent delivery, may be at the level of several million.

36Kr: This market doesn't seem to be particularly large. If we decide to enter this market, how much market share do you expect to eat?

Yang Yuanzu: It may be a little early to say this, this production tool actually has a certain winner-take-all attribute, or has the Winner takes all attribute.

The migration of the entire industry is now in its third generation. The first generation was PS more than a decade ago, when PS was used to design the website interface (Web) and process bitmaps.

The second generation is Sketch, which focuses on user interface design. The PS large-scale shift to Sketch was around 2015, and it also marked the entry of the entire design tool into the vector era.

The third generation is Figma, which ushered in an era of collaboration. Figma was first released in 2015 and was only recognized in 2017 and 2018, but basically by 2020 and 2021, 80% of the world's users switched from Sketch to Figma.

According to our judgment, the next visible era of the industry will be AI. Therefore, many daily business production research and UI design teams within Cloud Holdings are already using Motiff, such as zebra, ape programming, flying elephant planet, etc., and ape coaching related teams are also using.

We firmly believe that in the AI era, Motiff can exert great AI value, and it will also have an essential improvement in productivity. So we're ready for long-term investment in terms of technology, team, and funding.