Gen-AI is an open canvas. The future belongs to those who treat it as such!
February 26, 2024
The GenAI FOMO (fear of missing out) is real. Everyone talks about it but very few understand it (including myself). But if history of any transformative Technology is any guide, the magic of Gen-AI is no different than a magician swapping cards and making it look easy.
Creating viable and scalable Business models not only requires one to first identify what business problems they want to solve with GenAI, but also reimagine how Business is done in the first place. This was true for the Internet revolution and then the cloud revolution. Amazon did it twice first with eCommerce and then with Amazon Web Services. Salesforce did it with Salesforce Apps solely developed for the cloud.
What’s also true is that Large Language Models which are at the heart of Gen-AI, require TLC (Tender, Love & Care) to get them to a point where they are of real value to the business. There is no such thing is a “one-size-fits-all” out of the box Gen-AI to solve a business problem. But if you start by reimagining how a particular business process gets done, you may have a better shot at using Gen-AI to create a better way of doing things.
Technologies like GPT, make it simple to pass some prompts and use an API to get interesting answers to generic queries but unless until you analyze those results in the context of a particular problem statement, you won’t get much further. To make Gen-AI work, one must do some old-fashioned hard work and grinding. Maybe Gen-AI gives an accurate answer to your question a few times, but chances are it may not do so all the time. But if you narrow down your use case and iteratively utilize a combination of Gen-AI fine tuning, prompt engineering and traditional Machine Learning models, you will eventually get to a point where you have a level of consistency that is not only acceptable but might change the way you do business.
Think driverless cars. We were promised such cars in mass production by end of last decade, but we are still far off. All the Driver Assistance technology that exists today may be useful but from a customer’s standpoint, a driverless car by its design needs to be a binary thing. Either it’s fully autonomous or it’s not. We will get there eventually but just like I said earlier, it’s a grind to perfect the AI and those with the patience, focus and staying power will eventually crack the code and by doing so will reinvent how the world does things. For now, at least I don’t want to be a nervous “human in the loop” driver where I must be 10 times more alert while behind the wheel of a “semi” autonomous car. Call me old fashioned but that’s just me.
OpenBots set out to reinvent how Business processes Enterprise Documents using Gen-AI without the need for training unlimited Machine Learning models. What we found out in the process that even though Gen-AI does wonders, the output must be fine-tuned through multiple iterations of prompts, other fine tuning techniques and custom Machine Learning models. What we are realizing is that such iterations will continue as we get through multiple industry type documents and use cases. In short, team started out with few Gen-AI miracles, realized the limitations of those miracles, and focused on building a product and a framework which is designed to iteratively make the output better and better each time. There is nothing like this in the market not because our technology is revolutionary but because we realize the potential of Gen-AI is maximum when its combined with a consistent and iterative approach to make the output better and better each time.
In doing so, OpenBots is inventing a new category of Enterprise Document Transformation by progressively and iteratively removing the line between Enterprise Documents and Enterprise Data. Try it out at https://openbots.ai