There's a product decision we made recently.
We allowed people to start using our product without ever signing up.
In fact, they never even have to visit our website.
They just have to send an email to agent@gooseworks.sh
Here's the reasoning, and it starts with a question we kept asking ourselves while building Goose:
Goose is an AI coworker that is really good at GTM work like lead generation, outbound campaigns, brand monitoring, SEO / AEO, and more.
How can we make this feel more familiar?
Think about how you work with a contractor. You don't "onboard" yourself onto them. There's no dashboard to configure, no settings to toggle, no tutorial to complete before you can make your first request. You just communicate. You email them what you need. Maybe you have a couple of calls. You exchange context via documents. And then they do the work, and you get the result back.
That's it.
The entire onboarding industry — activation metrics, aha moments, time-to-value optimization, product-led growth funnels — is built on one foundational assumption: the user has to learn the tool.
That assumption breaks if the tool is smart enough to communicate.
The History of Software Interfaces
Every generation of software has had an interface that matched what users could do at the time.
Command lines required users who could write code. GUIs democratized software by hiding commands behind menus and buttons. Mobile made software location-aware and always-available. Chat interfaces (Slack, iMessage, WhatsApp) made software feel more like talking to a person.
Each transition shrank the gap between "what the user wants" and "how they express it to the machine."
We think we're at the next inflection point. And the next interface isn't a better chat UI or a smarter dashboard.
It's no interface at all.
It's whatever interface or surface you're already used to.
Email, Slack, iMessage, WhatsApp, Telegram, Phone call, Alexa, whatever.
What matters is not the channel but how you communicate the task.
What "No Interface" Actually Looks Like in Practice
We gave Goose its own email address: agent@gooseworks.sh
You send it a task in plain English. It does the work. It emails you back the result.
Last week I sent it an email: "Find me leads – scrape all the people who engaged with this LinkedIn post: [LINK]."
15 minutes later, I had a spreadsheet in my inbox with 77 names, companies, titles, LinkedIn URLs, and emails.
No account. No configuration. No dashboard I needed to learn.
Just an email describing my work and a result.
The Objection
"But users need an interface eventually. They need to manage results, configure preferences, see history."
Maybe.
Probably.
But notice how that's a different claim than "users need an interface to get their first value." Those are two separate problems. The onboarding problem and the power-user problem are not the same problem.
Every SaaS company conflates them because they've always been solved by the same artifact: a dashboard. But if an AI agent can give someone their first tangible result in 12 minutes via email — before they have an account, before they've committed to anything — then you've separated those problems. And you've made the first one almost trivially easy.
The interface can come later, for the users who want it. The value comes first.
What We're Building Toward
We think the end state looks something like this: you interact with AI agents the same way you interact with other professionals in your life. You message them when you have something to do. They do it. You don't maintain a relationship with their "platform" — you maintain a relationship with them.
Goose is an early version of this. Very early. But the core mechanic — email a task, get a result — feels like it's pointing in the right direction.
The interface isn't disappearing because we got lazy about building one. It's disappearing because it was never the point.
Try it: email agent@gooseworks.sh — send any task in plain English and see what comes back.
It's particularly good at lead generation tasks.