A schema for the real workflow
Turn spreadsheet columns, handoffs and statuses into relational records with clear ownership.
直接回答
什么是AI 内部工具构建器？
AI 内部工具构建器将书面需求转化为可编辑的 Web 应用。Marky 可在一个项目中创建界面、后端、数据库、身份验证和部署配置。
围绕真实任务构建
Marky 可以实现的功能
These are functional product layers tied to the workflow—not decorative sections copied from a template.
Practical CRUD interfaces
Create, review, update and archive records through forms, tables, filters and focused detail views.
Role-aware workspaces
Design different views and actions for operators, managers, reviewers or external participants.
Reports grounded in source data
Summarize operational records with real queries instead of decorative dashboard metrics.
Secure service connections
Use secrets for APIs such as email, SMS, AI, payments or cloud services without exposing keys in the client.
Versioned, portable code
Preview each change, roll back checkpoints, publish to Cloud Run and sync the repository to GitHub.
建议的数据模型
首个版本可能需要的数据
- Records
- The orders, cases, assets or requests being managed
- States
- Workflow status, transitions and validation rules
- Assignments
- Owners, teams, queues and due dates
- Activity log
- Changes, comments, imports and integration events
提示词示例
向 Marky 提供具体需求，而不是空泛概念
“Build an inventory operations tool for a specialty retailer. Track items, warehouses, suppliers, stock movements and purchase orders. Add low-stock filters, receive-order actions, manager and operator views, CSV import, an audit trail and practical empty/error states.”以此为起点
实用构建步骤
如何使用 Marky 构建
- 01
Start with the current process
Describe the spreadsheet, inbox or manual handoff and the failure it creates for the team.
- 02
Model records and state changes
Marky converts fields, permissions and transitions into a Postgres schema and typed APIs.
- 03
Test with realistic edge cases
Check empty states, invalid transitions, duplicate records, long values and integration failures.
- 04
Roll out deliberately
Approve the preview, publish it, add secrets and keep a checkpoint before every larger workflow change.
通用生产技术栈
标准代码与托管上线流程
React + TypeScript
An editable frontend, not a proprietary page format.
Postgres + typed APIs
Persistent records and protected server-side logic.
Accounts + secrets
Authentication and protected credentials for integrations.
Built-in publishing targets Google Cloud Run. AWS services can be integrated through protected credentials, and the GitHub codebase can be adapted for an AWS deployment by a developer.
具体问题
内部工具常见问题
Can Marky replace a spreadsheet workflow?
Often, yes—when the spreadsheet represents structured records and repeatable actions. Marky can model those records in Postgres and build validated forms, tables, filters and status changes. The original process still needs to be described accurately before it is automated.
Can an internal tool connect to our existing services?
Marky can integrate services that expose suitable APIs, webhooks or database connections. Credentials belong in project secrets, and each integration should include failure handling, permission boundaries and a clear source of truth.
Can different team roles have different actions?
Yes. Define the roles and allowed transitions in the brief so Marky can enforce them in server-side actions as well as in the interface. Hiding a button alone is not a sufficient permission control.
How do we recover if a change breaks the tool?
Successful previews create version checkpoints, and Marky can also save named checkpoints before risky changes. The project can be rolled back to an earlier working version; eligible published versions can also be restored to the live site.
构建第一个可用版本