Why machine learning models which take actions on computers (such as Adept AI) will succeed in the near-term
Earlier this week, we had two fantastic representatives from Adept AI present in our technology strategy class (with Professor Dan Wang) and educate us about the use cases of their AI that takes actions on computer software. Although Professor Wang informed us that most AI VC funding is currently going into either generative AI or specifically into companies that are rapidly improving AI technology, I believe that in the near-term companies like Adept AI that are translating tasks into software commands will have even more success.
I believe that the “sell” of companies like Adept AI is very strong for specific use-cases. In other words, Adept was able to streamline their product-market fit to create a product that when applied is specific, timesaving, and would be easy to market to a variety of industries.
Adept uses a workflow builder (examples of their experiments can be viewed here or in reference #4) to complete tasks on software with quick instruction and 95% reliability. Some examples are using software for a corporate recruiter to advance their candidate to the next stage of the process and request their availability, open attached invoices for accounting managers and extract information and enter it into their accounts payable, and do things you don’t even know how to do such as create a new discount code on Shopify, all with a single click of a button! This illustrates the wide breadth of possible application of Adept’s AI tool but in very narrow roles for specific individuals.
With this specific application, I can think of applications that I used in my previous two roles before business school, both in insulin manufacturing and life sciences consulting, that would greatly benefit from this AI that can perform repetitive, specific tasks faster than I can and decrease my quantity of repetitive, mundane work. I can also think of an easy example prevalent across industries, such as entering vacation into workday, that would in and of itself be a huge timesaver because although it is not difficult, it does require a lot of clicks!
The reasons why I think companies like Adept AI will succeed are the following:
1. Adept has proved the integration of the product into existing systems is not only feasible but very accurate and robust, meaning that even if the underlying software is slightly altered (to a level an intern could still figure it out) the pixel-reading ability will still be able to function properly
2. Adept’s customer-centric approach is fantastic because they start building the workflow by understanding customer needs in excruciating detail and their tailored solution is a key factor in company differentiation
3. Their flexible solution is scalable to continuously learn new software and evolve with market demands
The consequences for the industry are that there are many forms of generative AI that are in the process of evolving and maturing, but this is a much more concrete application that has potential to be used now. I believe this to be the next step on the journey of a company that is attempting to transition to heavy use of AI technologies.
The possible competition is AI embedded on software itself in coming updates. This would not be applicable to all software tools and would be harder to be specified for the exact use case. However, this will likely be included in future software updates for a lesser cost than developing one’s own workflow using an outsider technology such as Adept. I do not see any reason why a company could also replicate Adept’s mission, so there is possibility of similar competitors popping up even if they do not currently exist. There is also a chance that the workflows currently utilized will be entirely replaced by generative AI, but I see that taking time (in the span of a 5–10 years) rather than happening today.
There is great concern that AI can remove human jobs, but in this case, Adept AI is working alongside humans as a partner to make their day-to-day tasks faster and open up more time for workers to focus on what really requires thought in their roles. This helps to increase efficiency by decreasing the time allotment on mundane tasks and increased time spent on thought-provoking work. However, if efficiencies are too great, I do see that this could possibly reduce the total workforce required.
Possible barriers are retaining a skilled workforce with top talent with the amount of cutting-edge AI companies that currently exist and keep popping up. There needs to be a large focus placed on employee retention to continuously enlarge Adept’s capabilities and keep them competitive. Additionally, continuous innovation pressure and ensuring updates if major software changes occur will be essential for successful continuous use of Adept AI.
In summary, with this innovative technology and specific use cases, I believe Adept AI has multiple perfect product-market fits and a huge sell at this point in time as companies begin to leverage AI technologies.
References:
2. https://www.cbinsights.com/research/generative-ai-funding-top-startups-investors/
3. ChatGPT 3.5: “How will companies such as Adept AI succeed”; 15NOV2023
