San Francisco continues to attract founders building companies around one of the biggest shifts in technology: the move from AI that assists people to AI that can perform work, operate software and interact with the physical world.
The latest generation of startups is targeting problems that range from accounting and customer operations to developer testing and location-based data. Among the founders building in this space are five entrepreneurs whose 2026 companies are part of Y Combinator’s current San Francisco startup ecosystem.
1. Jibril Moinuddin — Founder and CEO, Rational
Jibril Moinuddin is the founder and CEO of Rational, a San Francisco-based startup building AI employees for accounting. Y Combinator lists Rational as a Summer 2026 company founded in San Francisco, with Moinuddin as its founder and CEO.
Rational is taking an ambitious approach to accounting automation. The company describes its product as an AI-powered accounting firm designed to handle work ranging from client intake and bookkeeping to tax filing. Its broader thesis is that AI agents could eventually take on responsibilities in industries where human judgment and trust have traditionally limited automation.
Moinuddin previously worked as a founding engineer at Buckeye AI and TrueFoundry and founded STEMplicity Education. Rational’s July 2026 launch outlined the company’s focus on using AI employees to address the capacity constraints facing accounting teams and firms.
2. Aidan Tiruvan — Founder and CEO, Archal
Aidan Tiruvan founded Archal to address a problem emerging alongside coding agents: how developers can reliably test AI-generated integrations with the software systems they are designed to use.
Y Combinator lists Archal as a Summer 2026 company based in San Francisco. The startup provides API sandboxes that allow coding agents and development teams to start controlled environments, connect integrations, run tests and inspect state changes before resetting the environment for another test.
Tiruvan’s background includes machine-learning research at Scale AI and NASA. In Archal’s launch post, he explains that he previously built agent environments at Scale AI and became interested in the testing problem as coding agents became increasingly capable of writing integrations.
The company’s premise is that conventional testing infrastructure was not designed for autonomous coding agents. Archal is positioning its sandboxes as an environment where agents can execute tasks against services such as Linear, Slack and Datadog while developers can observe what actually happened.
3. Ansh Chokshi — Founder and CEO, Mireye
Ansh Chokshi is building Mireye around another emerging challenge for AI agents: giving them reliable information about the physical world.
Mireye, a Summer 2026 Y Combinator company based in San Francisco, describes itself as infrastructure for agents that need to make decisions about physical locations. Its platform combines data, enrichment, tools and signals through an API and MCP interface.
The idea is to make information about places more accessible to AI systems. Data about land, utilities, demographics, regulations and other location-specific factors is often spread across different databases, portals and documents. Mireye is building a layer that allows agents to query and work with that information in a more unified way.
Chokshi previously led AI/ML engineering at Seismic. Before that, he co-founded Poker Pit, which Y Combinator says grew from zero to $250,000 in annual recurring revenue in four months before being acquired. Mireye’s early customers have used its infrastructure for applications including sourcing land for data centers.
4. Aniket Deshpande — Co-founder and CEO, Agent FM
Aniket Deshpande is the co-founder and CEO of Agent FM, a San Francisco startup focused on helping developers manage multiple coding agents from a single workspace.
Y Combinator lists Agent FM as a Summer 2026 company with a two-person team. The company’s desktop application gives engineers a group-chat-style interface for monitoring and steering coding agents, with updates covering progress, blockers, assumptions and decisions.
The product reflects a shift in how software development is being organized. As developers run multiple AI coding sessions simultaneously, the challenge is no longer simply getting an agent to write code. It is also keeping track of what different agents are doing and deciding when and how to intervene.
Deshpande’s background includes applied AI work at Meta Superintelligence Labs. Agent FM’s approach is to consolidate those interactions into one interface, including live text and audio updates, rather than requiring developers to move between separate coding-agent sessions.
5. Jeffrey Byun — Founder and CEO, Akkari
Jeffrey Byun is the founder and CEO of Akkari, a San Francisco startup developing autonomous customer operations software.
Y Combinator lists Akkari as a Spring 2026 company. The platform is designed to capture commitments, issues, requests and opportunities across customer channels such as Slack, calls, meetings, email and Discord, then help execute the work required to close those loops.
Byun brings experience from multiple startups. He previously scaled OrderAhead to millions of users before its acquisition by Square and helped scale BLADE to more than $1.2 billion in monthly trading volume, according to Y Combinator. He is also a two-time YC founder.
Akkari’s product is built around the idea that customer operations often involve a large amount of fragmented follow-up. A bug reported in Slack, a request made during a sales call or an onboarding blocker can all require action across different teams and systems. Akkari is attempting to turn those scattered signals into an ongoing operational workflow that can be tracked and executed by AI.
A Broader Shift in San Francisco’s Startup Scene
These five companies operate in different markets, but they share a common direction. Rather than treating AI simply as another software feature, their founders are building systems intended to perform work, coordinate processes or provide infrastructure for increasingly autonomous agents.
Rational is applying that model to accounting, Archal to software testing, Mireye to physical-world intelligence, Agent FM to AI-assisted software development and Akkari to customer operations. Together, they offer a snapshot of how San Francisco’s newest AI startups are expanding the definition of what an AI system can actually do.
