wrok Bot is an AI employee and digital teammate platform.
We’re building AI colleagues that take work from a goal to a result.
wrok BotYour AI colleagues, ready to work.
Explore the mobile App and desktop workspace with the same sample team and deliverables.
Your lead organizes AI colleagues with clear roles
The lead delegates; colleagues work and return useful results
Confirm how your AI colleague should work with you
These examples are readable without JavaScript. Interactive controls appear when the demo is ready. Product development progress
Set up a product launch team: a writer for the announcement, a developer for the sample page, and a lead to coordinate and bring the results back for my review.
Follow the lead, the handoffs and the result in the mobile interface. Switch between the conversation and task without starting a different story.
This is a website demonstration using our mobile UI styles. Native sign-in, device pairing and release acceptance remain incomplete.
Your lead organizes AI colleagues with clear roles
The task comes first. The result takes shape here.
Mobile UI sample · Tasks / Approvals / Chat
Give the goal · 1 / 4
Prepared examples of independent work, teamwork and deliverables. No model, device or team messaging service is connected; the product is in development. See development progress
Start with a goal in the workspace, bring in context and tools, then hand the result back to the team. Open a node to see its current scope.
wrok BotYour AI colleagueOrganize the work around a goal
Channels, teammates, skills, memory and conversation views are implemented. Complete macOS use and accessibility still need acceptance.
Connect the capabilities a task needs
Custom connections, a persistent catalog and parts of the runtime are implemented. SDK authorization has bounded deliveries; readiness across all three model sources is still open.
Engines, screen transport, control permissions and input have individual implementations. Real use, takeover, stopping and recovery still need complete validation.
Planned for a later phase, with no public product execution entry point today.
Keep the working agreement in context
Confirmation, explicit saving and management have individual implementations. Native authentication and complete product workflows still need combined validation.
Bring the work together in a useful result
Lists, records and charts have rendering foundations, with bounded artifact-reference integration. Complete persistent storage and recovery have separate acceptance.
This map explains how the product fits together. Node status uses the same capability record as the progress page. In development and planned do not mean released.
Give your AI colleague the goal, material and requirements. Keep the work connected from reviewing information to preparing a recommendation, with progress, open questions and deliverables together.
Prepare a showcase page and announcement for our new product. Have the lead coordinate, the developer build the sample page and the writer prepare the copy. Bring both results back for my review.
The developer works on the page while the writer prepares the copy
This example uses a prepared product brief to show team handoffs and delivery. It does not connect to a model or publish anything.
The sample page and announcement are prepared for human review and publication. This demo does not publish anything.
“Lead with the conclusion, then the evidence.”
This example saves one explicit preference. You can review, edit or delete it in the sample memory list.
Let your AI colleague prepare material and recommendations around the team’s goal, then bring open decisions to the people responsible. You confirm important actions and explicitly save working preferences, so handoffs keep their context.
From a team brief to an action plan or a recommendation, give each task a useful deliverable. Keep the reasoning, next steps and open questions with it, ready for the team to use.
Complete customer interviews, then revise the proposal using the feedback.
Decisions, evidence and open questions, together.
Make the team’s next actions clear.
Keep deliverables and open items ready for a handoff.
These deliverables are prepared examples. Complete task execution, teamwork and persistent downloadable files remain subject to capability and release validation.
No public installer is available yet. The project is still in development, with macOS as the first target. You can explore the source, view example demos and follow progress. The getting-started page will list files and system requirements when a real version is released.
wrok Bot is an AI employee and digital teammate platform, designed for AI colleagues that work independently. Its Rust core brings assistants, tools and task output together around a goal. It remains in development and validation, with macOS as the first target; complete independent execution depends on verified capabilities and actual releases.
Custom connections, a persistent catalog and parts of the runtime are implemented; SDK authorization has bounded deliveries too. Complete readiness across gateways, account bridges and custom models is still open. Real conversations, tools, cancellation and recovery need combined validation. See the model guide for scope.
Browser and native computer capabilities are within the product’s scope, with some foundations in place. Complete integration, authorization and real-device validation are still in progress. Website demos do not connect to or control your devices, and no online computer-control service is offered here.
The current approach uses explicit memory: you choose what to save, then review and correct it. Confirmation, saving and management have individual implementations; complete product workflows still need validation. Website samples do not train a model from visitor input and reset on reload.
The product database manages conversations, tasks and memory; its location depends on the desktop or self-hosted deployment. External model and tool requests go to their configured services. Displaying text, checklists or charts does not establish a persistent file artifact. A download is not currently promised for every output.
Meet wrok Bot and the AI employees and digital teammates we’re building.