Portkey.ai has announced a successful US$3 million seed funding round as it builds out its Large Language Model Ops (LLMOps) solution. Lightspeed led the funding round which also included a number of angel investors. Those investors included AWS, OpenAI, Cloudflare, Postman, and Asana.
Rohit Agarwal, Co-founder, Portkey.ai, commented: “Tech chiefs are facing a rush of demand from teams for AI apps that will save money without too much delay. But they cannot say yes to all their requests. There’s so much work to be done that there are often competing priorities and Portkey wants to help solve these dilemmas for tech teams. Our vision for Portkey has been to enable teams and companies to deploy Gen AI apps and features with confidence.”
What is the money to be used for?
To understand a little more about Portkey and what this money is to be used for, Enterprise Times spoke with Agarwal. We asked him where the money would be spent.
Agarwal said, “We want to use the money specifically for product development and research. We are getting a bunch of feedback from our customers. We’ve got about 30 customers and we do millions of requests a day. And all of these customers definitely add up to amazing feedback on the direction of the entire LLMOps. So product development will be a really key area for us.
“The second area will be partnerships with players like the LLM providers, vector databases, the ecosystem. How well are we able to embed ourselves through partnerships and integrations in the ecosystem?”
What is interesting is that Agarwal is not planning to spend much of the money on increasing staff numbers. Given the current salary demands in the market, that’s a good move because the burn rate would be high. One reason for not going on a recruitment drive is that Portkey is not building LLMs. They are helping customers train their own models, Portkey will not train models for its own business. That means it doesn’t need a large staff of LLM engineers.
Agarwal also believes that this money will get Portkey through the next 18 months. At that point, he believes that the company will either be cash positive or be able to go back and do a Series A funding round.
What does Portkey do?
Portkey sees itself in two roles. The first is enabling customers to build applications that can point to several LLMs. These might be internal LLMs or they might be public LLMs. One of the selling points that Agarwal talked about is the ability to restrict sensitive information from being sent to public LLMs. This is a major concern of many organisations and something that Portkey has invested in for its product.
The second part is to make it easy for customers to better and more easily manage their use of LLMs. Agarwal commented, “The biggest challenge is what I like to call forward compatibility. Technology is changing every month. You might build your entire application on one system today, and then next month, it’s all obsolete. This is giving enterprises a nightmare, saying we can’t keep changing our systems every month.”
Agarwal went on to Portkey is a full stack, LLMOps solution. It gives customers observability, delivers lower costs and latency, and improves the success rate for building and managing LLMs.
Enterprise Times: What does this mean?
It seems that everyone wants to use either public LLMs, build their own or, increasingly, operate a hybrid model. That latter is likely to be the way many organisations will go. They will use open-source LLMs against their own private data and want to mix that with responses from public LLMs. But to do that, they need to ensure there is no leak of sensitive data, something that is already challenging many organisations as users rush to try generative AI solutions.
Portkey wants to change that and seemingly has a plan and product that will appeal to IT teams. Those teams are already familiar with DevOps solutions and will see capabilities in Portkey that map to the way they currently work.
What will be interesting is how quickly Portkey builds out the capability of its solution. If the requests from its customers continue at the current pace, it is hard to see how $3 million will provide enough money for its R&D plans. However, those early customers are already seeing a responsive partner and are likely to invest more in Portkey. The question is, can it become profitable and bootstrap itself before it has to start seeking Series A funding?
