Launching today

Zelvara
Deploy AI agents your clients can trust
1 follower
Deploy AI agents your clients can trust
1 follower
Build an agent for a client in an afternoon. Give each client its own workspace and branding, test changes in draft, approve sensitive actions before the agent acts, and see what it did. Works on your website, WhatsApp and Telegram.









Free Options
Launch Team

Render Sandboxes: a computer for every agentGive every agent its own machine. Starts in under a second
Promoted
Hi Product Hunt, I'm Arun, the maker of Zelvara.
If you build AI agents for other people's businesses, building the agent is no longer the hard part. Trusting it is: keeping clients apart, testing changes safely, and knowing what the agent did.
What Zelvara does today:
A workspace per client. Agents, connected apps and chats stay with the client that owns them, and the chat widget carries that client's name, color and greeting.
Draft and live stay apart. Chat with the draft while customers keep using the version you released. Every release is a numbered version.
Ask first, then act. Sensitive actions wait for a person to approve, and the answer is recorded.
Safety limits per agent, such as actions per reply and masking of sensitive values.
A record of every conversation with the steps the agent took, and a queue for chats that need a person.
Website widget, official WhatsApp (Cloud API) and Telegram.
Start from one of 27 starting points, or describe the agent in a sentence.
We built this in public: 400 commits since July.
One honest limit: end users of an agent share that connection's token for now. Per-user tokens are not done yet.
I would love feedback from agency owners: what would you need to see before trusting an agent with a client's bookings?