v1.8.0 — A deeper bench
Our biggest roster release yet — plus, for the first time, you can bring your own tools and connect more than one account per service.
70+ new specialists, 7 new teams
The bench is now 330+ specialist agents across nine model families. Seven new teams join the lineup — Science Lab, Cybersecurity, Startup Founders, Smart Shopping, Poets & Sages, Wild Companions, and Health Conditions & Medical Info — for 47 built-in teams in all.
Bring your own tools — and connect more than one account
- Connect your own MCP servers. Point Nuveki at any Model-Context-Protocol server and its tools show up for your agents inline, mid-conversation — with a secure connect flow and OAuth where the server supports it.
- Multiple accounts per service. Two Google Workspaces, or work and personal GitHub — connect them all, set a default, and pick (or let an agent chain) the right one at runtime.
A reworked home page
A chat-first welcome with a living constellation of agents that reconfigures around whatever you ask — a solo specialist, a team, a pipeline, or several @mentioned agents. Quick-look panels let you size up any agent or team without leaving the page.
Sharper models, truer costs
Generalist and coding agents route to Claude Sonnet 5; the GPT-5.6 family is live; Claude vision is back; and billing now reflects long-context tiering and cached-input discounts, so your costs match what the provider actually charged.
Editing files in place
Inline live-edit gained rendered previews for more formats and a before/after curtain so you can see exactly what changed before you accept it.
One note: code execution is temporarily unavailable while we move it to a new internal sandbox — it'll be back soon. The full release note lives in the in-app What's New drawer.
Keep the feedback coming in Feature Requests — a lot of this shipped because you asked.