Quickstart

Welcome to Gosper. Gosper is an AI agent that takes a goal written in plain language, figures out the steps on its own, and carries them out for you — writing and running code, using external tools, working through long jobs in the background, and handling files you upload. You describe what you want; Gosper does the work and shows you the result.

This guide walks you through your first task from start to finish. It takes about five minutes, and you don’t need to know how to code.

Before you start

You’ll need:

A Gosper account (your sign-in email and password, or whatever method your team set up).
A goal in mind — anything from “summarize this spreadsheet” to “build and run a small script for me.” Don’t overthink it; you can start small.

That’s it. There’s nothing to install.

Open Gosper in your browser and sign in with your account. Once you’re in, you land in the console — your single, always-on workspace where you and the agent talk back and forth.

You have one console, and it stays with you. You don’t create a new one for each task; you just keep the conversation going. Gosper remembers the context of what you’ve been working on, so you can build on earlier requests without repeating yourself.

Step 2: Open the console

The console looks like a chat. There’s a message box at the bottom where you type, and the space above it is where the conversation and the agent’s work appear.

Take a quick look around:

The message box is where you type your goal in plain language.
The conversation area shows your messages and the agent’s replies, including a running view of what it’s doing as it works.
The upload control (a paperclip or “attach” button) lets you add files or videos for the agent to work with.

You don’t need to configure anything. Just start typing.

Step 3: Type your first task

Click into the message box and describe what you want — the way you’d ask a capable colleague. Be specific about the outcome you want, and let Gosper handle the “how.”

A great first task is something small and concrete that you can verify at a glance. For example:

Build me a hello-world Python script and run it.

Send that, and Gosper will write the script, run it in a secure sandbox, and show you the output. It’s a satisfying way to see the whole loop work end to end.

Here are a few more first-task ideas, depending on what you’re curious about:

Try the code sandbox: “Write a Python script that lists the first 20 prime numbers and run it.”
Try data work: “Generate a CSV of 10 sample customers with names and emails, then show me the file.”
Try a quick research-and-write task: “Draft a short, friendly welcome email for new users and save it as a text file.”

Tips for a good first request:

Say what you want, not how to do it. “Summarize this report into five bullet points” works better than spelling out every step.
Include any specifics that matter — a format, a length, a filename, a tone.
One clear goal at a time is easiest to follow while you’re getting the feel for it.

When you’re ready, send the message.

Step 4: Watch the agent work

This is the fun part. Gosper doesn’t just reply — it breaks your goal into steps and carries them out, and you can watch it happen live in the console.

As it works, you’ll see Gosper:

Plan its approach — figuring out what needs to happen and in what order.
Run code in a secure sandbox — a private, isolated environment where it can write files, execute scripts, and check the results. Nothing it runs touches your computer.
Use external tools when needed — for example, connecting to a service like YouTube or Gmail to get something done. The first time Gosper needs access to one of these, it will ask you to authorize the connection; after that, it can use it on your behalf.

The progress streams in as it goes, so you’re never left wondering what’s happening. If you realize you want to adjust something mid-task, you can just send another message — Gosper takes your input into account as it works.

Step 5: Read the result

When Gosper finishes, it gives you a clear, final answer: the output of the script it ran, a summary of what it did, a file it produced, or whatever your goal called for.

If your task created files, they’re saved in your workspace, and you can ask Gosper to show them, change them, or do more with them. The work doesn’t disappear when the task ends — you can keep building on it in the same console.

If the result isn’t quite what you wanted, just say so in plain language: “Make it shorter,” “Use a different format,” or “Try again, but sort the list alphabetically.” Gosper picks up right where you left off.

Working with files and videos

Some goals start with material you already have. Use the attach control to upload a file or a video, then tell Gosper what to do with it — for example:

Here’s a video. Pull out the audio and give me a transcript.

Or:

Here’s a spreadsheet of sales. Find the top five products and chart them.

Gosper works directly with what you’ve uploaded, all inside its secure workspace.

Handing off the long jobs

Not everything finishes in a few seconds. Some goals — processing a large batch, running a multi-step project, moving a big video between services — take a while.

For those, you can hand Gosper a goal as a background task. Gosper keeps a running plan, works through it on its own, and reports back when it’s done. You don’t have to sit and watch; you can come back later and check on it. The console shows you the status of anything that’s still in progress.

This is what makes Gosper feel less like a chatbot and more like a teammate: you can give it something substantial, walk away, and find it finished (or well underway) when you return.

What to try next

Once your first task clicks, stretch a little:

Chain a few steps: “Generate some sample data, analyze it, and write up the findings.”
Connect a tool: ask Gosper to do something that involves an outside service, and authorize the connection when it asks.
Upload and transform: bring in a file or video and have Gosper turn it into something new.
Kick off a longer job as a background task and check back on it later.

Tips for great results

Describe the outcome. Focus on what “done” looks like, and let Gosper choose the steps.
Add the details that matter to you — format, length, tone, filenames — and leave the rest to Gosper.
Iterate out loud. The fastest way to refine a result is to react to it in plain language.
Lean on it for the heavy lifting. Long, multi-step, or repetitive work is exactly what Gosper is built for — that’s when handing it a background task pays off most.

You’re ready. Sign in, open the console, and give Gosper its first goal. Start small, watch it work, and you’ll quickly get a feel for just how much you can hand off.