Browse the web — live
Your employee can read the open web and control a real browser in the cloud — the same way you would.- Search the web for fresh information and pull back highlights with citations.
- Read any page — articles, docs, product pages, forum threads — and summarise, extract, or compare.
- Live browsing — navigate to a site, click buttons, fill in forms, scroll, and watch what happens. Your employee can sign into your accounts (with your permission), download files, and operate apps that don’t have an API.
- Watch it work — when your employee opens a live browser, you get a link to peek at the session in real time.
Generate and edit images
Your employee can make images from a text prompt, edit images you give it, or combine multiple images into a new one. Results come back as a clean URL you can drop into emails, decks, or your app. Examples of what you can ask:- “Make a clean illustration for the top of my newsletter — bold flat colors, no text.”
- “Take this product photo and put it on a beach background.”
- “Combine this logo and this hero shot into one social card.”
Generate videos
Your employee can produce short videos from a text prompt or an image — useful for social posts, product demos, intros, and B-roll. Generation runs in the background and your employee tells you when it’s ready. Like images, finished videos return as a URL you can hand off, share, or embed.Voice — speak and listen
- Text-to-speech — turn a script into an expressive voiceover with a chosen voice (you can pick from named voices like “Rachel” or “Adam”, or supply a custom voice). Supports expressive cues like whispers, laughs, and emphasis.
- Transcription — drop in an audio file (mp3, wav, m4a, webm) and your employee transcribes it. Optionally returns timestamps for each segment or word — great for captions, meeting notes, or repurposing recordings.
A real workspace
Every employee runs in its own private, isolated sandbox — a small Linux workspace where it can:- Create, read, edit, and organize files of any kind.
- Run scripts and command-line tools to crunch data, generate reports, or process files.
- Keep working notes, drafts, and intermediate output between conversations.
- Search across its own files for facts it has saved.
Talk to your apps and services
Once you authorize a Connector, your employee can read and write through that service from inside chat — sending emails, booking meetings, posting to Slack, updating a CRM, pulling reports, and so on. Don’t see a service you need? Your employee can talk to any HTTP API through a Custom integration.Build and run your BuildAI apps
Your employee can build brand new BuildAI apps from chat — a CRM, a content calendar, a daily dashboard — and then run them for you, populating records, updating statuses, and pulling live numbers without leaving chat. Existing apps you’ve already built work the same way: just point your employee at one and it takes over. Open any connected app from the APPS strip above the chat to see what your employee has been doing. See Connected apps for the full walkthrough.Learn new skills
Your employee can be taught skills from a curated catalog — drop-in playbooks for specific jobs like writing newsletters, vetting leads, building decks, or running QA. Open the Skills tab, pick a skill, and click Teach — your employee instantly knows the playbook, no extra prompting required. See Skills for the full list.Remember things and follow up
- Memory — your employee remembers facts about you and your work between conversations. See the Memory tab.
- Scheduled tasks — hand off recurring work and your employee runs it in the background, daily, weekly, or on any cadence. See the Tasks tab.
Putting it together
These capabilities compose. A single request can chain several of them without you having to say “now do this, now do that”:“Find me three recent articles about home espresso machines under $500, pull the key specs into a sheet, generate a hero image for the roundup, and draft a Tuesday newsletter using my usual voice.”That’s a web search, multiple page reads, a Google Sheets write (Connector), image generation, and a drafted email — one prompt. Your employee figures out the steps and uses the right capability for each one.