Airbnb, a leading online marketplace for hospitality services that allow travelers to look for suitable accommodation during short and long vacations, has waded into the fast-growing arena of artificial intelligence. The company is working on one of the latest AI systems that will allow its product engineers to get ideas directly from the drawing board and converting them into actual products almost immediately.
The Airbnb AI, which is still under development, targets simplifying the product design routine by narrowing down to testing the principle functional prototypes as opposed to simply relying on half-assessed modules from different development teams.
Airbnb AI design, a solution to reduce lags in product development
Benjamin Wilkins, the Airbnb design technology lead, argues that the time required to test any idea should be zero. He further posited that at the moment, each step in the design process and the generated outcome is a dead end. The task ends when a discipline/ one level completes a section of a project and hands it to another team.
The conventional flow in any product development and design begins from stakeholders’ meetings through stages such as design engineering, explorations, mockups, prototypes, and final products.
At the heart of this lengthy process are complexities associated with translating the shared meaning to make the original goal achievable. This causes a lot of friction and drags that can quickly fade interest or lose the original meaning.
By focusing to clear the friction on the production workflow, Airbnb tech team concluded that the most productive method of streamlining product development is automation to allow machines to learn through artificial intelligence.
The company commenced by building the initial prototype with hand-drawn components as data and open source codes learning algorithms for training their Airbnb AI. The process was not very difficult because Airbnb already had a carefully documented design model with multiple components and guidelines.
The subsequent step was integrating the design and allowing Airbnb artificial intelligence to classify different components.
Airbnb AI tests give satisfactory results by recognizing Airbnb standards
The main assumption by the lead design teams is that after thorough training and fine-tuning, the artificial intelligence would easily understand the preset Airbnb standards sketched design components and render them into real source codes. The team was surprised to find the results highly satisfying.
The Airbnb AI has demonstrated immense potential. Wilkins explains that they have also tried the technology at different live code-prototypes levels in the Airbnb ai lab to do the following:
- To live-code product prototypes from multiple whiteboard drawings.
- To convert high-fidelity mocks targeted component specifications for the Airbnb engineers.
- To translate production codes into specific design file replication by designers.
If everything goes as Airbnb plans, the AI-assisted prototyping will be fully integrated and replicated in the company’s in-house product development packages.
Note that Airbnb AI sketch technology is not the first in the fast unfolding field of artificial intelligence. UIzard Technologies, a Copenhagen based startup, demonstrated its software that can interpret raw product-services mockups and apply latest AI to translate them to source code.