top of page
Writer's pictureLipie Souza

Latest Open-AI announcements powering no-code tools for GenAI Product Management

Updated: Nov 13, 2023

There is widespread talk about the importance of experimenting with no-code and low-code tools before launching a completely proprietary product, saving time and effort. This trend is now expected to strengthen with the focus on new Generative AI tools.



A few months ago, I started some analyses on this blog in the hope of trying to predict the impacts of Generative AI on the daily lives of people involved in the development of digital products, in order to prepare ourselves for such changes. I conducted analyses on the impact on the architectural ecosystem of large companies, on the role and processes of a product team, but also on the way in which natural language may gain ground against the classic development languages that have been used for decades to build algorithms, and ultimately software, interfaces, and finally, products. I was wrong in the timing of some of these predictions, some of which I still believe are yet to happen, but in a more gradual and slower process than I originally imagined. 🤷‍♂️



The good news is that I continue to find evidence of these changes, and indeed, they are driven by some trends that are now strengthening: increasingly considering and including Open Source frameworks and no- or low-code tools in the pipeline of new solutions, especially when there is an opportunity to rethink digital journeys. Proof of this were the announcements made by OpenAI at its first DevDay, for those who didn't check out the announcements in full, they are available here. With dozens of no-code feature launches that are directed and triggered by natural language, reinforcing the union between the roles of product people and developers in tool construction, as the NLP evolved by LLMs (Large Language Models) opens doors for everyone to think about business and development strategy with very little code, since all that expertise in a specific programming language won't be necessary, increasingly hybridizing the roles that make up a product team.


Just as back in 2008 when Apple launched the iPhone and the entire software industry had to adapt to creating 'more mobile' experiences, now we have to adapt to more conversational experiences, widely supported by GPTs that will serve our customers and perform internal operational tasks of companies much more efficiently both in feasibility and cost, compared to the current digital ecosystem. A point that reinforces this trend was the launch of GPT-4 Turbo, which saw a 3x reduction in price compared to the original GPT-4, with an updated cutoff date to April 2023, increased feasibility, and understanding a much larger context window (input) than previous models. These advances of the 'base' models provided by OpenAI, as well as the work of the entire OpenSource community that has been improving public LLMs, open up space for new no-code tool startups to receive vast investments from YCombinator, which historically indicate trends that emerge in Silicon Valley and quickly gain global adherence. In this post, I analyze the new business models that were previously included in YCombinator's investments for the first half of 2024. Indeed, their investment announcements were advanced, reinforcing the market's bet on these technologies.



Let's start with Use Collate, the most nascent of all, still without a launched product, but already with a waiting list and obviously, being accelerated by YCombinator. The solution has a well-defined purpose: to provide a tool for GoToMarket teams to analyze behaviors and trends of their target audience and customers using AI, promising to easily integrate with social network databases of their business, internal CRM databases, and external public databases about consumer behavior, investment profiles, etc. The idea is to solve that old problem I've mentioned here a few times where piles of data do not generate any relevant insights for product evolutions and repositioning. It's a significant problem, and I will follow Collate's progress, as if they get the implementation right, soon we will have an excellent ally to diversify and modernize the proposal and the portfolio of services and physical and digital products offered by companies


Another venture I am closely following is OmniAI, which promises to easily integrate with all your structured and unstructured databases of different types and sources and transform them into language intelligible and consumable by LLMs (the so-called Vector Search, a recurring topic on this blog). From there, you'll be able to interact with a customized and self-updating model of databases that you consider important for decision-making in various domains in your company, and then, converse with the AI. The model will help you make inferences and base your response on practically all the data ingested from your company's databases, avoiding the work of data scientists who often have to manually extract and treat data, as well as having to establish training strategies often through trial and error that take time and generate rework whenever new databases or dimensions are added. It’s a big bet! 🎲🎲🎲


On the other hand, Artisan and AgentHub want to revolutionize no-code process automation tools like Zapier by associating such solutions with the famous prompt engineering, that is, you will be able to tell in natural language what you expect from each input and output to an agent that will act as an intermediary adding cognitive intelligence to these flows. Basically, instead of the process following a series of pre-established logical steps by you, you can add more complex analyses like: 'If document X contains any legal risk associated with law Y, please add a flag Z in field T,' and the model, based on its training, will make the decision. This goes far beyond OCRs for data extraction and treemaps for decisions, it is more comprehensive and generalist, and the output will become more sophisticated as you update and refine the prompt and the LLM to incorporate your domain of information relevant for decision-making in a certain business process. On one hand, Artisan has focused on creating agents that perform a certain role and connect with tools generally associated with this role, for example, a design agent performs prototyping tasks and therefore connects to Canva, Photoshop, etc. On the other hand, AgentHub focuses on the creation of customized agents that can be tuned by you to perform any task involving integrations with N no-code applications, without focusing on specific roles.



Finally, there's this Danish startup called Leya that will also be boosted by YCombinator in 2024, and will cater to lawyers who want to enhance their corporate work by helping in the monitoring of public procedural databases for updates on the progress of cases, in drafting legal opinions, and in referencing jurisprudence. There are already established companies, including in Brazil, working close to this scope, but they stop at the simple management of legal contracts and almost static composition of legal documents. With this, in the field of legal analysis and writing, Generative AI, it seems, will be of great help. And oh, by the way, for those who don't know yet, check out the Vade Mecum AI, an experimental product for lawyers and law students to consult about the law through conversations launched by me. Guys, that's it, I'll continue to follow the development of these ideas, and as there's a lot of money behind them and therefore a lot of speculation, we'll have cases of great success and sensational flops (there always are, right?). Let's keep going. See you. 🍻



30 views0 comments

Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page