AI Developer Tools & SDKs
-
Salus is a runtime guardrail that checks every action your AI agent takes
An AI agent that can process refunds or change customer records is useful right up until it does the wrong one. Salus, a YC W2026 startup, is built for exactly that risk: a runtime guardrail layer that checks every action an agent takes before it executes. ## How Salus works Salus is a policy-aware proxy.… Continue reading
-
Orthogonal gives AI agents one MCP server to buy and call 50+ paid APIs
APIs were built for developers with accounts and contracts, not for autonomous agents that need data mid-task. Orthogonal, a YC W2026 startup, rebuilds that layer for agents: it gives them instant access to 50+ premium APIs through a single MCP server with pay-per-call credits. ## What Orthogonal does Instead of managing dozens of API keys… Continue reading
-
Zenbu is an extensible IDE built for orchestrating your coding agents
As coding agents move from novelty to daily driver, the editor around them matters more. Zenbu, a Y Combinator Spring 2026 project, is an extensible IDE built specifically for orchestrating coding agents rather than bolting them onto a traditional editor. ## What Zenbu does The pitch is that you shape Zenbu around exactly how you… Continue reading
-
Omnara runs Claude Code and Codex from your phone with voice control
Coding agents like Claude Code and Codex normally live in a terminal, which means you are chained to your desk while they run. Omnara, a Y Combinator S25 company, breaks that tie by turning your phone into a command center for those agents. ## What Omnara does Omnara lets you start a coding session locally… Continue reading
-
Microsoft Work IQ APIs give agents a governed view of your Microsoft 365 work
An enterprise AI agent is only as useful as the context it can see. Microsoft’s Work IQ APIs, going generally available on June 16, 2026, expose the intelligence layer behind Microsoft 365 Copilot so any agent can tap the same understanding of how work actually happens. ## What Work IQ provides Work IQ continuously processes… Continue reading
-
openmio gives every AI agent a permanent inbox at openmio.com/
Agents can call tools and browse the web, but they can’t message each other. openmio fixes that with a dead-simple primitive: a permanent address at openmio.com/<name> that any AI agent can send to and read from — across sessions, machines, and runtimes. How it works Claim a name like openmio.com/sam with one Google sign-in, then… Continue reading
-
Firecrawl Prometheus writes the web-scraping code for the data your AI agent needs
Pulling clean, current data off the web is one of the messiest jobs in building an AI product — pages change, scrapers break, and someone has to keep fixing them. Prometheus, a new release from Firecrawl, hands that whole job to an agent. ## What Prometheus does Firecrawl calls Prometheus a Forward Deployed Agent for… Continue reading
-
Qursor lets you point at any UI element to hand your AI exact context
Anyone who codes with an AI agent knows the frustration: you ask it to tweak “that blue button on the pricing page,” and it confidently edits the wrong element. Qursor, a developer tool that launched on Product Hunt this week, fixes that by letting you point at any UI element and hand the agent exact… Continue reading
-
Browserbase gives AI agents cloud headless browsers to act on the web at scale
An AI agent is only as useful as its ability to actually do things on the web — and that means running a real browser somewhere reliable. Browserbase handles that layer, giving developers cloud-hosted headless browsers built specifically for AI web agents. ## What Browserbase provides Instead of maintaining your own fleet of browsers, you… Continue reading
-
KugelAudio is a self-hostable text-to-speech model with 39ms time-to-first-audio
Voice agents in regulated industries face a problem most TTS vendors ignore: you cannot ship customer data to a third-party cloud. KugelAudio, built by a four-person Berlin team and accepted into Y Combinator’s Spring 2026 batch, is a real-time text-to-speech model designed to run on your own infrastructure. ## Low latency, self-hosted KugelAudio reports a… Continue reading
