Second Layer Spotlight, Edition #1: AI Infrastructure
For the first edition of Second Layer Spotlight, I'll be highlighting promising companies in the AI Infrastructure vertical: those developing solutions to optimize LLMs and agentic workflows.
Every week, I want to highlight a few companies I’ve been particularly interested in, all of which have demonstrated strong early success signals within my 9-factor startup evaluation framework. Additionally, all of these companies align nicely with my investment thesis: the Second Layer Approach to Venture Capital.
This week, the second-layer vertical of focus is AI infrastructure. A lot of investor and media attention remains on developing a more powerful and capable LLM, or finding new ways to implement AI into existing workflows for operational improvement.
However, AI systems are only as effective as the infrastructure in place to support them: that means enabling integration into systems, understanding and addressing bottlenecks, and digesting unstructured data across a variety of sources.
It won’t be an absence of innovation that will inhibit the development of new AI-focused technologies - it will be limitations in the infrastructure meant to support this growth in innovation at scale.
Here are three early-stage companies within the AI infrastructure vertical with promising early signals:
1. Arcade.dev
$12M, Seed Stage. Based out of San Francisco. Investors include Flybridge, Hanabi Capital (Mike Volpi), Neotribe Ventures, Garuda Ventures, and angel Andy Rachleff
Founded in 2024 by Alex Salazar and Sam Partee, Arcade AI is a tool-calling and authentication infrastructure platform for AI agents. AI agents need to call external tools, whether it’s sending emails, querying databases, reading calendars, or hitting APIs.
However, each one of these activities requires secure authentication, permission scoping, and token management - AI agents constantly run into bottlenecks across these areas, inhibiting workflows from running smoothly. Arcade AI builds the layer that handles this, giving agents pre-built connections to dozens of services alongside a Software Development Kit.
Arcade.AI scored the highest across Founder-Market Fit (previous founder exit within the space), TAM (agentic software market is massive and growing), Timing (enterprise agentic AI adoption is scaling exponentially), and Investor Signal (funding partners have an extensive, successful track record).
2. Trellis AI
$500k Y-Combinator Raise. Based out of San Francisco. Funded by General Catalyst (undisclosed amount)
Trellis AI automates healthcare paperwork and prior authorization workflows for healthcare providers. It builds AI agents that process unstructured clinical documents, including referrals, medical charts, insurance guidelines, and appeals, converting them into structured data directly integrated into Electronic Health Record systems.
By enhancing these processes, the time-to-treatment bottleneck is significantly reduced, allowing patients to be treated more quickly while allowing health centers to treat a higher volume of patients.
The U.S. spends over a trillion dollars a year on administrative healthcare costs, delaying care, draining provider revenue, and driving staff burnout. By reducing time-to-treatment by over 90%, Trellis is tackling one of the most substantial problems facing American families today.
Trellis AI performed exceptionally well across a few areas: backing from General Catalyst demonstrates high investor quality, Product-Market Fit (billions in annual therapy processing, nationwide scale across hundreds of locations), and Technical Execution (clinical training on millions of data points, along with integration into EHR platforms).
Though Trellis AI is early stage, they have already experienced quite significant, noteworthy traction. It is a company I’d keep an eye on.
3. Maxim AI
$3M, Seed Stage. Based out of San Francisco. Seed round led by Elevation Capital, with Investment from various Angels: Founders of Postman (Abhinav Asthana), Chargebee (Krish Subramanian), Groww, Razorpay, and Media.net
Maxim AI is an end-to-end evaluation and observability platform for enterprise AI agents and LLM applications. Most enterprises managing LLM applications use an observability tool or an evaluation framework separately - Maxim was developed as a unified solution, targeting teams that have adopted AI implementation at scale and need their systems evaluated for quality assurance.
Maxim AI covers Pre-deployment (testing LLMs before production), Post-deployment (live monitoring, tracing, and observation), and Quality assurance (evaluation of outputs to detect failures throughout the complete lifecycle). By bridging the gap between observability and evaluation, Maxim AI eliminates the need to toggle between multiple tools.
Where Maxim performs exceptionally well in my startup analysis is in the Market (every enterprise AI deployment needs quality assurance infrastructure), Timing (LLM experimentation to production is happening fast), and Investor quality (Elevation Capital is a well-established fund).
Closing Thoughts
I’m always happy to share more specific details about these startups and why they are on my radar. I plan to share high-quality early-stage startups across a new vertical each week - if there’s a vertical that excites you, please let me know.
As always, if you have any questions, feedback, or recommendations, I’d love to hear them. Please don’t hesitate to reach out to me via email or LinkedIn.
Note: The views expressed in this article are my own and are intended for informational purposes only. Nothing here constitutes financial or investment advice. Venture capital involves significant risk, and all investment decisions should be made in consultation with a qualified financial professional.
Relevant Links
Email: bryanhanleyvc@gmail.com
LinkedIn: https://www.linkedin.com/in/bryan-stanley-hanley/
Lovable Website: https://bryan-hanley.lovable.app/