Announcing the launch of Blue Morpho!
As a repeat entrepreneur, I understand the importance of selecting the right idea space for a startup. Much of my previous company’s success came from being perfectly positioned, in the right place at the right time. I must admit, with GitGuardian, the cybersecurity company I co-founded right after my engineering studies, our success in choosing the right topic to focus on owed more to luck than to meticulous planning. We had no idea that building in a space with so much urgency, ample room for growth, and compelling events that demanded a radically new approach would be such a game-changer. The market was in such a need, it would forgive us all the mistakes startups inevitably make, and reward ten times our agility and ability to innovate.
So when I decided to launch a new business earlier this year, exploring the Gen AI space seemed like a natural choice. However, it’s not without pitfalls. This sector is riddled with noise and delusions - many startups make false promises, and many companies don’t know who to trust or which technology they should leverage -. The old YC mantra, suggesting founders should be embarrassed by their first product version, might not hold up here. On the contrary, I believe it’s more crucial than ever to come up with a product that can rise above the fog of war. Secondly, there are large and rapidly growing players in the field already. It’s not wise to bet against the likes of OpenAI. Sure ChatGPT isn’t perfect. But rather than betting against OpenAI models, I believe that the most successful startups will find ways to capitalize on their continuous improvement (as I write this, I am eagerly waiting for GPT-5 to come out!). Lastly, while there are exceptions, engaging in pure research right from the start isn’t advisable most of the time, and is also a pitfall. Pure research activities differ from startups in skills required, time horizons, and financing needs. There’s plenty of work to do already, and significant engineering challenges to solve in assembling existing technologies intelligently and applying research papers to the real world. For example, at Blue Morpho, we believe in assembling the best of Knowledge Graph - which have been around for quite some time! - and LLM technologies. These are hard problems!
That might be quite a long introduction to announce the launch of Blue Morpho! So what is Blue Morpho doing exactly? We are building a critical piece of architecture that is currently missing in the Gen AI landscape: an intermediate layer between LLMs and organizations’ data, that takes the form of an enterprise knowledge graph. Knowledge graphs are used to categorize and link organizations’ data, enabling the integration of data from diverse sources, including databases and documents. LLMs can then navigate through the graph, a bit like we navigate from website to website using hyperlinks, in order to perform even the most complex tasks. The beauty of our approach is: we are using LLMs to build and curate the graph, which in turn makes LLMs a lot smarter! Now you know why I am eagerly waiting for GPT-5 to come out: when LLMs get better and better, so do we!
This approach has many benefits, here are some of them:
- Improved accuracy: LLMs navigate the graph to retrieve results from it, which yields incomparably better results, compared to traditional RAG techniques. Knowledge graphs improve search capabilities by providing context to LLMs through relationships between the nodes.
- Improved reliability & explainability: the graph is an auditable & explainable source of truth, and graphs facilitate the identification of inconsistencies in the data.
- Improved integration & interoperability: knowledge graphs categorize and link disparate elements in a coherent, unified whole.
This is huge: every organization wants to use AI on their proprietary data, but simply putting the latest AI models on top of their data does not cut it (and it is not going to be a source of competitive advantage as anyone can do it). Nvidia is selling the shovels but except on simple use cases or restricted perimeters, noone is digging gold at an enterprise scale yet! Remember AI = Data + Algorithms. Sure, algorithms are advancing rapidly and organizations can integrate the latest generalist models as they evolve. But when it comes to their data, they have to do the hard work themselves! And Blue Morpho helps them do exactly that!
Simply putting the latest AI models on top of organizations' data will not cut it and is not going to be a source of competitive advantage as anyone can dot it.
It is so much fun for me to have the opportunity to start building again from scratch! I know all the magic of great businesses comes from the team members. Being able to share adventures with great coworkers, build meaningful relationships, grow next to each other is a very large part of what makes building a business such a renewed pleasure. We are currently a team of 4, and have open senior positions in Data Science, ML Engineering, and Web App Development (front/back) (full remote opportunities). If you’re interested, let’s talk, don’t hesitate to drop me a line on LinkedIn.
Our current team:
To be updated about our progress, hit subscribe.
Let’s build!
