Ohio-based startup Mantium has today announced closing $12.75 million in seed funding, as well as the launch of a cloud-based AI platform -- which allows users to build with large language models.
The seed round, co-led by venture funds Drive Capital and Top Harvest, will be used to source for more talent, to add more features to Mantium’s AI platform and in driving awareness around what is achievable with large language models, especially across Africa, the firm's CEO and co-founder Ryan Sevey told TechCrunch.
It is looking to expand its team of 33, which is currently spread across nine countries, including Ghana, Nigeria and Kenya. Having a globally distributed team, Sevey said, helps in the generation of unique insights and varying problem-solving approaches around AI.
“We'll also continue with more activities within Africa. This (AI) is an evolving topic, and we will get more advanced with the talks, but right now, it's mostly an introduction to what is capable with large language models.”
Mantium’s AI platform, which starting today has moved from alpha to general availability beta, enables both technical and non-technical persons to build applications using resources of providers that include OpenAI, EleutherAI, AI21 and Cohere, and soon to be added, Hugging Face. Together with the platform’s comprehensive documentation and API integrations, with the compatible platforms like OpenAI, users are able to mix and match capabilities on Mantium without requiring expertise in each platform.
The launch of the cloud-based platform is designed to make it easy and possible for users to publish live prototype applications with the click of two buttons, a capability that does not require a user to have prior coding experience. It has also integrated logging and automated filtering, ensuring that users remain compliant with safety and security requirements.
“There is massive potential for innovation with AI, but we need to make it easier and faster for developers and creatives to get to deployment,” said Sevey.
“With Mantium, we are making it possible not only for anyone to use large language models and AI to build something, but to get to prototype faster, and share it with their communities, all while maintaining the most stringent safety and security measures,” he said.
Sevey got to finally build Mantium in January this year, after he attended an OpenAI beta program where most of the participants, who were from non-technical fields, were using OpenAI resources to build products but could not share with friends or launch. The attendees' pain points inspired him to create a product that makes it easy for people to build with AI and deploy.
This, he added, is the realization of his long-term dream of democratizing AI in a meaningful way -- something he has wanted to do throughout involvement with machine learning.
“About 80% (of the attendees to the OpenAI program) were not technical. They had no prior coding, knowledge or experience. And they were building these really cool things. These would be things like maybe a poem generator, or some people were doing really interesting stuff with music and helping them write lyrics, all kinds of more creative focused stuff,” said Sevey.
AI can be used to execute various actions, including helping users generate music by parsing terabytes of data from millions of songs. It can also be used in business to do classification on text to figure out the intent, can be deployed to extract information from patient medical records and to generate synthetic data.
“They had no idea how to actually share that with the rest of the world. So, we created Mantium. And one of the features that's within Mantium is what we call one-click deploy, which essentially means that once you have your AI creation, you click two buttons and it's live. So, you get a URL, it spins up a web application and you can interact with your creation, which you can share with your friends,” he said.
Such features, he said, push the boundaries of what is possible with AI, while hastening the pace of new developments -- from proof of concept to production -- by eliminating barriers. This in turn ensures that users or developers have more time to fine tune and build better models.
“And that is a big burden lifted off, quite frankly, everyone including the technical people, who don't want to spend hours or days to set up something like Heroku or Digital Ocean or whatever the case might be, just to share a prototype out there.”
“We’re going to see the next wave of technological breakthroughs come from artists and creatives who, with the power of Mantium, can now play and build with AI.”
Sevey says the company is currently focused on improving its user experience with plans to make it a consumption-based business later on. However, he noted that Mantium will always have a free version “that lets people get started and lets them get their prototype out there.”
Sevey said Mantium is also preparing to enter a phase where it will create impactful meaning applications beginning next year -- as it moves to connect “things with AI”.
“So by quarter one of next year, we will have a lot more integrations with our platform so that it becomes even more of a single pane of glass so that you have your large language model, it's generating something, or it's classifying something and then it connects to your chatbot or to your internal accounting system,” he said.
“And as we think about the future, I think this notion of connecting things with AI becomes increasingly important.”