We cover the AI agents and services built on top of the big models — the tools you'd actually pick. Here's where the week landed.
Cursor leans all-in on parallel agents
On April 2, Cursor launched Cursor 3, rebuilt around running many coding agents at once — across local checkouts, git worktrees, the cloud, and remote SSH — with every agent (including those started from mobile, web, Slack, GitHub, or Linear) surfaced in one sidebar. The bet: "supervise a fleet of agents" is becoming how people code, not one completion at a time.
Genspark raises for its 'AI employee'
Around April 4, AI "super-agent" startup Genspark extended its Series B to $385M at roughly a $1.6B valuation, after reportedly passing ~$200M annualized run rate in 11 months — and launched Claw, an "AI employee" agent that executes multi-step tasks across software.
ElevenLabs widens its toolkit
ElevenLabs shipped an April 1 update adding speech-to-text from hosted URLs (YouTube/TikTok links), video-to-music, and new ElevenAgents controls — pushing further from "voices" toward a full voice-agent platform.
What it means if you're picking tools
The coding-agent question is shifting from "best autocomplete" to "best at orchestrating multiple agents." See the community's current picks for coding, automation, and voice & audio.
Sources: Cursor 3 · Genspark Series B (Dealroom) · ElevenLabs changelog