Workspace context
Start from the resources you already operate instead of explaining your platform from scratch.
Turn an idea into a full-stack application
Describe what you want to build — Shard AI plans the project, creates the app and database, and connects services. Start from an existing project instead when you need deeper context.
One task view for workspace context, execution, and review
Live
streaming activity
Workspace
workspace context
Approval
reviewable changes
Apps + data
connected resources
A Shard AI task
From idea to reviewed feature.
Shard AI brings together context from your resources, plans the implementation, and prepares every change for your decision.
Start with text, an image, or audio.
Add a feature to the application
Shard AI
Add a reports page to my application
Task activity
- Analyzed the application structure
- Checked the available data
- Planned the new reports page
Proposed change
Create the reports page and connect the required data
File to create
src/app/reports/page.tsx
The proposed implementation is ready for your approval before it changes the application.
Workspace operations
Turn a goal into an investigated and reviewable task.
Shard AI is built around the resources in your workspace. It can inspect the current state, work from an optional application context, and propose the next operation without leaving the dashboard.
Explore capabilitiesInvestigate workspace state
Ask Shard AI to inspect applications, databases, virtual machines, domains, status, logs, configuration, and connections before it recommends a next step.
Focus on one application when needed
Start from a project to carry its application context into the task, while still keeping the rest of the workspace available for connected work.
Coordinate changes with a clear review step
Shard AI can propose creating apps or databases, creating or restoring backups, managing files, updating resources, controlling services, and changing domains or DNS records.
Built to operate
Workspace-aware assistance with visible execution boundaries.
Shard AI streams its work in the task view and groups activity, proposed changes, and results so you can understand what it checked and decide what happens next.
Project-scoped tasks
Attach a task to one application for focused context and keep it organized with that project.
Live investigation
Inspect resource state, logs, configuration, files, and operational signals in one conversation.
Connected operations
Plan changes across apps, databases, virtual machines, files, and domains with their relationships in view.
Reviewable actions
Actions that are not read-only wait for human approval unless you enable auto mode for eligible actions.
Fix and evolve your application
Investigate deployment failures and application metrics, create new features, and execute changes after your approval.
What the agent can do
Build, inspect, and operate from one task.
Shard AI keeps proposed changes reviewable, while actions that require approval wait for yours before they run.
Design and create applications
Turn an idea into a project, then create an application or update its resources.
Inspect application health
Check resource metrics, application logs, and HTTP metrics and logs before deciding what to change.
Create and update databases
Provision a database or propose a configuration update for an existing one.
Run virtual machine commands
Execute commands on virtual machines through proposed, reviewable actions.
Create and restore backups
Create a backup or restore one when the requested change is approved.
Configure domains and DNS
Create, update, and remove DNS records; connect domains to applications or virtual machines, and adjust SSL settings.
Shard AI models
Choose the right model for the work ahead.
Choose based on task scope: a quick request, an advanced implementation, or a complete full-stack project.
01
Default
For simple, focused tasks where you need a quick answer or a small change.
Examples
- •Explain a log error
- •Write a Dockerfile
- •Create an environment variable
02
Pro
For advanced tasks that require deeper reasoning across several steps.
Examples
- •Debug a failed deployment
- •Plan a database migration
- •Improve an API endpoint
03
Max
For full-stack projects that span code, data, deployment, and operations.
Examples
- •Build a SaaS MVP
- •Create a dashboard with a database
- •Plan and deploy a production service
Every plan includes tokens for Shard AI; check the available quantity and models for each plan.
View tokens and models by planProgressive setup
Start with an outcome, then keep every decision in view.
Shard AI creates a task, investigates the available context, and waits for your decision before applying proposed operations.
Get started01
Start from the workspace or a project
Open a workspace task for broad work, or start from an application when the task needs that project as its initial context.
02
Describe the outcome
Ask Shard AI to investigate, create, connect, or change something. It can use tools and ask follow-up questions as the task becomes clearer.
03
Review the proposed operation
Use the activity, changes, and output views to review the work, then approve or reject each proposed operation.
From connection to the first decision, in one flow.
Frequently asked questions
Frequently asked questions about Shard AI
Yes. Every plan includes tokens for Shard AI. The number of tokens and available models vary by plan; see the details on the pricing page.
It can inspect the platform resources your account can access, including applications, databases, virtual machines, and domains. Its tools can retrieve details such as service state, logs, files, and connection information when relevant to the task.
Yes. You can buy more tokens whenever you need them from the dashboard.
Yes. It can investigate the application context, plan the implementation, and propose changes to files and related resources. Any execution follows the task approval rules.
Yes. Shard AI can check deployment status, application logs, resource metrics, and relevant HTTP signals before it recommends or proposes a change.
Depending on the request, Shard AI can propose creating applications or databases, creating or restoring backups, changing application, database, or virtual-machine resources, managing files, controlling service status, and managing domains, DNS records, and SSL settings.
Reads and investigations do not change your resources. Every action that creates, changes, restarts, removes, or executes something requires human approval; you can enable auto mode for eligible actions. Domain and DNS actions still require explicit confirmation.
Yes. You can enable auto-approve for eligible actions from the Shard AI interface. Domain and DNS actions always require an explicit confirmation, even when auto-approve is enabled.
Yes. Shard AI accepts text, image, and audio input, so you can include a screenshot, diagram, or spoken description with your task.
No. Shard AI is already integrated into the dashboard and uses the tokens included with your plan.
Ready to move forward
Operate the platform without rebuilding context every time.
Open Shard AI to start a workspace task with your resources already connected.
Guided setup with no commitment
Investigate before it proposes
Shard AI inspects real workspace state before suggesting a change.
Nothing happens without approval
Every proposed action waits in a review queue for your confirmation.
Open Shard AI
Describe what you want to build and let Shard AI plan the first step.
A plan for every stage of your project
Rated 4.9/5 on Trustpilot · Trusted by 50,000+ developers
Basic plan
Perfect for freelancers or individuals who need to host applications and bots.
1-2GB RAM
1-2 vCPU
4-8 projects
5-10GB CDN
Core plan
For small teams and startups that need infrastructure for multiple projects.
4-8GB RAM
4-5 vCPU
16-32 projects
35-100GB CDN
Connect your own custom domain
Pro plan
Built for scaling companies with high project volume and advanced support.
12-24GB RAM
6-10 vCPU
48-96 projects
150-250GB CDN
Load balancers
Automatic backups