Skip to main content
You open Runable. You type what you want built. The agent builds it. This page gets you from sign-in to your first real output as fast as possible.

Sign in

Go to runable.com and sign in using one of four methods You land on the home dashboard with an input bar at the center.
Runable home dashboard showing the central input bar, artifact type shortcuts below it, and model selector

Pick what you want to create

Click an artifact type below the input bar before sending your message. Each type configures the agent with the right tools for the job.
ArtifactWhat it createsWhat you can set
SlidesPresentation decks in HTML or image formatSlide count, theme (Professional / Visual), format
ReportMulti-page formatted documentsPage count
SpreadsheetInteractive spreadsheet files
CarouselMulti-slide social media postsSlide count range
ImageAI-generated images on the canvasAspect ratio
VideoAI-generated video clipsAspect ratio, audio on/off
AudioSpeech, music, sound effects, voice contentVoice selection
CanvasFreeform visual workspace for images and videos
WebsiteFull-stack deployed web application
ChatText conversation with the Latest AI Models
You can also skip the artifact picker entirely. Type your prompt and the agent figures out which type to use.

Write a detailed prompt

The more context you give, the better the output. Here are prompts you can copy and use right now.

Investor pitch deck for a fintech startup

You do not need a perfect prompt. The agent reads what you wrote, identifies what is missing, and asks clarifying questions before it starts building. But more detail upfront means fewer back-and-forth rounds and better results on the first try.

Answer the agent’s questions

Before building, the agent asks a few short questions to make sure it gets things right. The questions depend on what you are creating, but typically cover:
  • Target audience and tone
  • Visual style or color preferences
  • Structure (how many pages, slides, or sections)
  • Specific features or integrations needed
You answer once. The agent builds with those answers baked in.
Chat interface showing the agent asking about audience, style preferences, and structure before generating content

Watch it build in real time

Once you answer, the agent starts working. Every agent chat gets its own isolated sandbox environment.

Refine by talking

When the output appears, keep the conversation going. Tell the agent what to change.

Restyle a section of your website

Restructure a slide in your presentation

Add a new feature to your website

The agent edits your existing output in place. You never leave the window.
Did not like a change? Click the rollback button on any previous message to restore that exact version. Or branch the conversation to try a different direction without losing your current work.

Know your input bar

ElementWhat it does
Text areaType your prompt here.
+ buttonAttach files: images, PDFs, spreadsheets, videos, code, anything. Up to 10 files, 25 MB each. You can also drag and drop files, paste from clipboard, or take a screenshot.
Chat modeSwitches to text-only conversation mode. The agent answers questions and analyzes attached images and PDFs, but it cannot execute code, generate media, or search the web.
Plan modeThe agent researches your request, asks questions, and writes a detailed plan for your approval before building anything.
Model selectorLite Fast and low cost. Pro Handles most tasks well. Max Deep reasoning for complex projects.
ConnectorsLink Google, Slack, Notion, GitHub, Shopify, and 3,000+ other apps so the agent can work with your existing tools.
Quick actionsShortcuts below the input bar: Create Slides, Generate Videos, Build Websites, Canvas, and more. Tap one to start with a focused prompt.

Two things that will save you time

Use Plan Mode for complex projects

If your project has more than one or two moving parts (a multi-page website with Stripe, a database, and user auth), turn on Plan Mode before sending your message.The agent researches your request, asks clarifying questions, and writes a structured plan showing exactly what it will build and in what order. You review, adjust, and approve. Only then does it start.This saves you from the “that is not what I meant” moment three steps into a build.

Save your preferences to memory

Tell the agent your defaults once and it remembers them across every future conversation.

Set your brand defaults once. The agent remembers this in every future chat.

Every website, deck, carousel, and image starts from these defaults. No repeating yourself.

Next steps

Plans and credits

See which plan fits your usage and how credits work.

How the agent works

Understand the agent loop, sandbox, tools, and real-time streaming.