Live data from Hacker News

Launch HN: AgentMail (YC S25) – An API that gives agents their own email inboxes

news.ycombinator.com

171–180 of 183 posts

Re: Launch HN: AgentMail (YC S25) – An API that gives agents their own email inboxes

#173
Congrats on the launch. Sign-up was super smooth and the product is compelling.

Will be testing it with some of the agent work I'm doing on agent.ai.

If you're ever looking for a super low-maintenance indie investor, please reach out. Have lots of YC folks that will vouch for me.

Re: Launch HN: AgentMail (YC S25) – An API that gives agents their own email inboxes

#174

The idea here is great and I think it has a lot of legs but I think inherently people will have initial concerns with security. Do you have plans to only allow reading mail from specific senders, and giving fine-grained control on the API side? I would like to give this to an AI agent but only initially where I and my team can forward mail to the agent, and not give it the ability to draft and send emails on its own.

Yes we are adding for email/domain allow/blocklists. Same for fine-grained permissions.

Re: Launch HN: AgentMail (YC S25) – An API that gives agents their own email inboxes

#178

I'm concerned that this fits in "using today's innovation to solve outdated paradigms". Google has A2A: An Agent-to-Agent Protocol. SaaS is plumetting in value. Arbitrary semantics made sense when communications were human-dominated. If agents dominate these fields, why wouldn't they simply set their own protocols and methods to communicate both text, binary, and agreed data structures? There's an assumption that ema…

And you probably don't want to dump the emails first thing into the agent's context. You should insert a cup of coffee and the morning newspaper into the context, and only then the emails.

Re: Launch HN: AgentMail (YC S25) – An API that gives agents their own email inboxes

#179

If you know agents email address, it can still be Prompt Injected.. what prevention exists there ?

we have a few things in place, allowlists and permissions act as a layer. also beginning some work on prompt isolation within api soon. but having an isolated identity + data within a separate agentic inbox also puts less risk of your personal email data being injected - which is most people's main concern

>prompt isolation

You mean isolating data from instructions? As far as I know this is not a solved problem, or is it?

Post reply on HN