LastMile AI closes $10M seed round to 'operationalize' AI models

Image Credits: Andrey Suslov / Getty Images

LastMile AI, a platform designed to help software engineers develop and integrate generative AI models into their apps, has raised $10 million in a seed funding round led by Gradient, Google’s AI-focused venture fund.

AME Cloud Ventures, Vercel’s Guillermo Rauch, 10x Founders and Exceptional Capital also participated in the round, which LastMile co-founder and CEO Sarmad Qadri says will be put toward building out the startup's products and services and expanding its seven-person team.

"Machine learning, and the broader field of AI, has gone through a few AI winters -- oftentimes due to a constraint on computing resources, a constraint on expertise or a constraint on high-quality training data," Qadri told TechCrunch in an email interview. "We plan to democratize generative AI by streamlining the tooling and disparate workflows and simplifying the need for deep technical expertise."

Qadri, along with LastMile's other co-founders Andrew Hoh and Suyog Sonwalkar, were members of Meta's product engineering team prior to launching LastMile. While at Meta, they built tooling, including AI model management, experimentation, benchmarking, comparison and monitoring tools, geared toward machine learning engineers and data scientists.

Qadri says that these tools served as the inspiration for LastMile.

"The recent wave of interest and adoption of AI is being driven by software developers and product teams that are using generative AI as a new part of their toolkit. Yet machine learning developer tooling is still mostly geared towards researchers and core machine learning practitioners," Qadri said. "We want to empower builders by providing a new class of AI developer tools built for software engineers, not machine learning research scientists."

Qadri has a point. Some companies, faced with the immense logistical challenges of adopting AI from scratch, aren't clear on how to leverage all that the tech has to offer.

According to a recent S&P Global survey, around half of IT leaders say that their organizations aren't ready to implement AI -- and suggest that it may take five years or more to fully build AI into their company's workflows. Meanwhile, about a third say that they're still in the pilot or proof-of-concept stage, outnumbering those who've reached "enterprise scale" with an AI project.

At the same time, business leaders aren't fatalistic about their opportunities to embrace AI. In a 2022 Gartner survey, 80% of executives said that they think automation can be applied to any business decision. Model management was cited as a top roadblock -- 40% of organizations had "thousands" of models to keep tabs on, respondents said -- but they indicated that other factors, including AI talent, weren't as big an issue as might be assumed.

LastMile allows customers to create generative AI apps leveraging text- and image-generating models from both open- and closed-source model providers. Developers can personalize these models with their proprietary data, and then incorporate them into their new -- or existing -- apps, products and services.

Using LastMile's AI Workbooks module, users can experiment with different models from a single pane of glass. The AI Workflows tool, meanwhile, can chain together different models to build more complex workflows, like an app that transcribes audio to text and then translates that text before applying a synthetic voiceover. And the AI Templates module, the last module in LastMile's AI dev suite, creates reusable development setups that can be shared with team members or the wider LastMile community.

"Our goal with LastMile is to provide a single developer platform that encompasses the entire lifecycle of AI app development," Qadri said. "Today, the AI developer journey is fragmented and requires stitching together a number of different tools and providers, and nuanced understanding of every step -- which increases the barrier to entry. We're focused on building a platform that non-machine learning software engineers can use to develop AI-powered apps and workflows, from experimentation and prompt engineering to evaluation, deployment and integration."

Now, LastMile isn't the only company tackling these challenges in the AI tooling, measurement and deployment space.

When asked who he sees as competitors, Qadri mentioned LlamaIndex, a startup offering a framework to assist developers in leveraging the capabilities of LLMs on top of their personal or organizational data. LangChain is another rival in Qadri's eyes -- the open source toolkit to simplify the creation of apps that use large language models along the lines of GPT-4.

But competition or no, Qadri sees a massive opportunity for New York City-based LastMile, which is pre-revenue, to make waves in a nascent -- but fast-growing -- space. With the market for AI model operations set to grow to $16.61 billion by 2030, according to one report, he might not be too far off base.

"Enterprises are investigating how to revamp their businesses to incorporate AI in their applications and workflows, but they’re encountering last mile issues that prevent them from getting things into production -- for example, how many ChatGPT-based chatbots have you seen incorporated into corporate websites?," Qadri said. "These blockers can be largely solved by better AI developer tools that enable rapid experimentation and evaluation, provide orchestration infrastructure, and deliver monitoring and observability for confidence in production. LastMile AI provides the tooling and platform to assist businesses in confidently incorporating AI in their applications."