# Clanker Cloud | AI Workspace, Projects & Agents > Clanker Cloud is a browser-based AI workspace for recurring project work. Projects keep files, instructions, and outputs together for reports, research, and Apps. Reuse Agents and configure hosted Loops or Schedules for repeatable work. Computer access and DevOps are additional workflows. Positioning reviewed: 2026-09-14 Canonical: https://clankercloud.ai/ ## Give your work a home. And a crew. Bring your ideas. Your AI crew researches, writes, and builds, learning from what you save. You stay in control. ## Different talents. Same home base. Scout, Secretary, and Patch are example custom Agents. Choose their instructions and tools in Workspace. The homepage walkthrough uses fictional inputs and does not run live AI work. ### Scout: Research Agent Curious by nature. Always checks the source. Give a research Agent your sources and a question. Get a comparison, a brief, and the evidence to check it. Example task: Compare these proposals. Show me the trade-offs. Example inputs: proposals.pdf, project-brief.md. - Read the proposals - Compare scope, terms, and missing details - Prepare a comparison with source references Result to review: vendor-comparison.md. [Explore research with an Agent](https://clankercloud.ai/learn/research-with-project-files) ### Secretary: Writing Agent A tidy mind for your wonderfully untidy notes. Give Secretary your notes, examples, and the way you like things written. Refine its instructions as you work, and get drafts you can shape, check, and keep. Example task: Turn these notes into a client update. Flag what’s missing. Example inputs: meeting-notes.md, client-brief.pdf. - Read your project notes - Pull out progress, blockers, and next steps - Prepare a client update for your review Result to review: weekly-update.md. [Explore writing with an Agent](https://clankercloud.ai/learn/turn-notes-into-a-report) ### Patch: App-building Agent Sees a spreadsheet. Imagines a little app. Give an app-building Agent your data and an idea. Get a private tracker or dashboard to try, refine, and share when ready. Example task: Make a simple task tracker from this project CSV. Example inputs: project-tasks.csv, app-brief.md. - Inspect the project data - Build a private App preview - Prepare the App for you to test and refine Result to review: Private App preview. [Explore building with an Agent](https://clankercloud.ai/learn/build-an-app-from-a-spreadsheet) ## Create your own Clanker. A detail detective? A deadline whisperer? A specialist in that very specific thing you do? Give your Clanker a name, a little personality, and a clear job. Make its expertise your own. ### Pick a name. Give it a job. Define what it should help with, how it should sound, and what a good result looks like. ### Pack its toolkit. Add your instructions, examples, and Skills. Choose the files, connections, and tools it can use. ### Try it. Teach it. Make it yours. Run a real task, review the result, and save useful feedback in its instructions. Keep it for yourself or share a copy. [Create your Clanker in Workspace](https://workspace.clankercloud.ai/). [See how custom Agents work](https://clankercloud.ai/make-your-own-ai-agent). ## A crew that learns how you work. Your examples, saved knowledge, and useful corrections give your Agents more to build on. Refine their instructions and keep the lessons that matter, so the next task starts with what you’ve already taught them. Example lesson: Put blockers first in client reports. Save it in Secretary’s instructions, then reuse it for the next update. ## Your data. Your call. No black boxes. With your own Agents, you can see the setup, follow recorded work, and open the results. Choose their access and decide what gets shared. ### Choose what comes in. Choose the files, connections, and tools your Agents can use. Review their access and set the approval mode that fits the job. ### See the work behind it. Inspect your own Agent’s instructions and selected tools. Review recorded activity, source links, and saved outputs, then ask for a correction. ### Keep the final say. Download your files, decide what to share, and disconnect services when you no longer need them. Your work stays reviewable and reusable. Hosted work uses cloud services and model providers. [Data processing, retention, and deletion](https://clankercloud.ai/privacy). [Service providers](https://clankercloud.ai/subprocessors). ## There is a method to the magic. ### Bring the messy bits. The brief. The notes. That spreadsheet. Keep them in a Project with your goal and instructions, so the work has somewhere to start. [Give your project a home](https://clankercloud.ai/learn/organize-project-files) ### Call in the right talent. Choose or create an Agent for the job. Give it the knowledge, Skills, and Tools it needs, then ask it to research, write, or build. [Put an Agent to work](https://clankercloud.ai/learn/build-a-custom-agent) ### Keep the good work going. Review the result, ask for a change, and save what works. Your files and instructions are there for the next job. Set up a Schedule when you want work to repeat. [Make the work repeatable](https://clankercloud.ai/learn/loops-and-schedules) The Secretary writing example uses fictional Northstar Studio notes and an illustrative report. It is not a verified customer run, a quality benchmark, or a measured cost. [Inspect the sample inputs, prompt, and review checklist](https://clankercloud.ai/templates/weekly-client-report). New accounts start with $20 in included credits. No credit card is required. AI work and active cloud environments consume credits; recurring workers have ongoing costs. Review the first task and its usage before expanding the workflow. [Understand credits and usage](https://clankercloud.ai/learn/workspace-credits-and-costs). Start in [Workspace](https://workspace.clankercloud.ai) from your browser. Projects keep files and instructions together; Sessions are where work happens. Connect a computer or choose a cloud environment when the task needs it. ## Work you can try ### A decision, with the evidence. Compare these three proposals. Give me the trade-offs and a recommendation with sources. Result to review: Vendor comparison. ### A file you can actually use. Turn these project notes into a weekly update with decisions, blockers, and next steps. Result to review: Weekly project update. ### Your idea, in a working interface. Build a dashboard from this project CSV. Show open items, owners, and a filterable table. Keep it private. Result to review: Operations dashboard. ### Repeat the work. Keep the context. Every weekday at 9, check these public sources. Update my brief and notify me when something meaningful changes. Result to review: Morning briefing. ## Common questions ### What is Clanker Cloud? Clanker Cloud is a browser-based AI workspace where Projects hold your files, instructions, and outputs, and Agents bring selected knowledge and tools. Use it to research, create reports and other files, build Apps, and configure recurring work. Keep the context and results together for the next task. ### Can I create my own Clanker? Yes. A Clanker is your custom AI Agent in Workspace. Give it a name, define its role and tone in its instructions, add relevant knowledge and Skills, and choose the tools and connections it can use. Test it on a task and refine the setup. Keep it for yourself, share a private copy, or submit a version for Marketplace review. The illustrated crew on this page shows example personalities, not preinstalled Agents. ### How do Agents learn from my work? Learning here means building on the context, examples, knowledge, and instructions you save and make available. Keep useful findings with their sources, turn a correction into an Agent instruction or Skill, and reuse the setup on the next task. Review and update saved context as your work changes. A conversation does not automatically rewrite every Agent or retrain its underlying model. ### What control do I have over my data and Agent work? Choose which files, connections, and tools your Agents can use, set Session approvals, review recorded activity and outputs, and decide what to share. You can download your files, disconnect services, and use the account deletion and privacy-request paths described in the Privacy Policy. Hosted tasks process data through the applicable cloud services and model providers; retention and deletion follow the published policy. ### What does “no black boxes” mean here? It means visibility and control around your workflow: inspect your own Agent’s instructions and selected tools, review recorded activity and sources, and open saved outputs. Model internals and private publisher instructions for managed Marketplace Agents remain outside that view. Review the Privacy Policy and Subprocessor Register for how hosted data is handled. ### How can I make a weekly client report with AI? Add your meeting notes, project brief, and reporting instructions to a Project. Ask an Agent for progress, decisions, blockers, and next steps, with missing information clearly marked. Review the draft, request corrections, and save the report and sources for the next update. ### What can I reuse next week? Keep the source files, reporting instructions, previous outputs, and related Sessions in the same Project. Add fresh inputs and tell the Agent which material is current. You can also reuse an Agent with saved instructions, Skills, and selected Tools. A saved Project does not automatically schedule future runs. ### Are these Agents working live in the example? The banner is an illustrative walkthrough with fictional inputs and example custom Agent names. Scout shows research, Secretary shows writing, and Patch shows app building. Create or choose Agents and configure their instructions and tools in Workspace. The walkthrough does not run AI or measure real task quality, time, or cost. ### What does an AI crew mean? Your crew is the collection of Agents you use for different jobs. Each Agent can have its own instructions, reference knowledge, Skills, and selected Tools. You choose the Agent for a Session, decide its access, and review the work. Projects keep related files and outputs together. ### Do I need to install anything? You can start in Workspace from your browser. Working with your own computer requires the Clanker desktop app and an approved connection to that device. Cloud browsers, cloud desktops, and hosted compute run separately from your computer. ### What is the difference between a Project, a Session, and an Agent? A Project keeps related files, instructions, and work together. A Session is a conversation where work happens and progress is recorded. An Agent is the assistant doing the work, with its own instructions, skills, and selected tools. ### Can it work while my laptop is closed? Hosted Loops and Schedules can run recurring work in the cloud. Tasks using a connected computer still need that computer to be online and available. Simply starting a browser conversation does not turn it into a recurring background task. ### How much does it cost to try? Every new Clanker Cloud account starts with $20 in included credits. AI work and cloud environments use credits at their applicable rates. You can buy more credits or choose a monthly plan; see Pricing for current rates and allowances. ## Read more - [Workspace overview](https://clankercloud.ai/workspace.md) - [Guides](https://clankercloud.ai/learn.md) - [Templates](https://clankercloud.ai/templates.md) - [Comparisons](https://clankercloud.ai/compare.md) - [FAQ](https://clankercloud.ai/faq.md) - [Guide for external agents](https://clankercloud.ai/agent.md) - [Pricing](https://clankercloud.ai/pricing.md) - [Security](https://clankercloud.ai/security.md)