FAQ

Is this another chatbot?

No. Havenware is a fleet of role-specialized agents (inbox, outreach, social, finance, research) doing sustained operational work under one governance layer. You do not sit in a chat window driving it; the fleet works, and consequential actions come to you as proposals in an approval queue.

What can an agent do without my approval?

By default, nothing consequential. Anything external, irreversible, or touching shared state, sending, publishing, invoicing, paying, requires an approval record to execute. Agents draft, read, organize, and prepare freely. Over time, an agent with a clean track record can graduate specific low-consequence action classes to autonomous execution; that graduation is earned through clean runs and can be revoked, and every autonomous action still lands in the audit trail.

What happens if I ignore a proposal?

It expires. Every proposal carries an expiry date, and an expired proposal executes nothing. Ignoring the queue is always safe; it just means the fleet's blocked work stays blocked.

Which model providers does it use?

Havenware runs on Claude today and is built against a provider-neutral interface, so agents are not tied to a single model vendor.

Where does my data live?

In your accounts. Agents work with your email, your CRM, your books, through your credentials, scoped per agent. Your data is not used to build or improve anyone's models.

Is this actually running anywhere?

Yes. Havenware runs the daily operations of a real business today: inbox, outreach, finance, and research. The product pages describe the system as it runs, not a roadmap.

How do I get access?

Book a call from the homepage if you're evaluating Havenware for a company, a team, or an institution. If you want to follow along first, get the dispatch there too: one short email a month on what the fleet shipped.