Arcade Raises $12M to Improve AI Agent Quality

Arcade Secures $12 Million in Seed Funding

Arcade, a startup focused on AI agent infrastructure, has successfully raised $12 million in a seed funding round. The company was established by Alex Salazar, previously an executive at Okta, and Sam Partee, a former engineer from Redis.

Laude Ventures Leads the Investment

The funding was provided by Laude Ventures, a newly launched venture fund for 2024. This fund is spearheaded by Andy Konwinski, a computer scientist from UC Berkeley and also a co-founder of both Perplexity and Databricks.

This investment represents Laude Ventures’ inaugural publicly disclosed funding activity, as confirmed by co-founder and general partner Pete Sonsini to TechCrunch. Sonsini brings significant experience from his tenure at NEA, where he played a key role in early-stage investments in companies like Databricks, Anyscale, and Perplexity.

Founders’ Backgrounds and Inspiration

Alex Salazar is an experienced entrepreneur. Prior to Arcade, he founded Stormpath, an authentication API startup, which was acquired by Okta in 2017. Following the acquisition, he held a VP role at Okta, concentrating on product development.

Sam Partee has a strong background in LLM-based application development and has contributed to prominent open-source projects, including LangChain and LlamaIndex, as highlighted by Arcade.

The inception of Arcade can be traced back to February 2024, sparked by Salazar’s observation of the capabilities demonstrated by ChatGPT 3.5. He envisioned a future centered around AI agents and subsequently founded the company.

Challenges in AI Agent Development

Initial efforts to develop functional AI agents proved challenging. Salazar and Partee encountered significant obstacles in achieving reliable performance.

“We were attempting to create a site reliability agent to rival established players like Datadog,” Salazar explained. However, “the majority of agents exhibited limited functionality and effectiveness.”

The founders dedicated considerable effort to resolving issues related to agent connectivity and data access, essential for task completion.

The Issue of Data Access

A key impediment identified was the reliance of many agents on LLMs trained exclusively on publicly available data. This limitation prevented them from accessing and utilizing private data.

Consequently, while agents could discuss product features, they were unable to verify specific actions, such as confirming order delivery status.

A Platform Empowering AI Agents

The founders envisioned Arcade as a foundational element for AI agents, mirroring the role Okta previously established within the SaaS cloud services landscape.

Initially, they developed a tool-calling platform specifically for a site reliability agent. Demonstrations of this agent, however, sparked a different kind of interest.

“Feedback consistently centered on the functionality enabling the agent, rather than the agent itself,” explained Salazar. This realization prompted a pivotal shift in strategy.

“We collectively determined that focusing on and offering the underlying tool-calling platform would be the most effective path forward,” Salazar stated.

Thus, Arcade was born, designed to grant each agent the same level of access to applications and data as the human worker it supports, or the specific role it is designed to fulfill. Pricing for Arcade is structured around usage or subscription models.

Secure Integration and Authentication

Arcade leverages OAuth for seamless integration, managing authentication across a vast number of SaaS services and websites.

It functions as a secure intermediary, handling token management to prevent direct access to sensitive credentials by Large Language Models (LLMs), according to Salazar.

Early Investment and Strategic Focus

Sonsini, having previously invested in Salazar’s Stormpath venture, promptly expressed interest upon learning of the new startup.

“Our investment strategy prioritizes technically proficient founders, and we maintain strong ties with the research community through our limited partners,” Sonsini noted.

While many AI startups concentrate on the visible aspects of LLMs, such as agents, Sonsini’s focus lies on the essential infrastructure capable of supporting substantial business growth.

Arcade, he believes, “perfectly aligns with this strategic vision,” representing a critical component in the evolving AI ecosystem.