Simple Mode
Simple Mode is the fastest path from idea to architecture. You describe your SaaS in plain English — StackAlchemist handles the rest.
When to Use Simple Mode
Use Simple Mode when:
- Your idea is still at the "concept" stage
- You haven't thought through every entity and relationship yet
- You want StackAlchemist to interpret your requirements and make sensible modeling decisions
- You want the fastest path to a generated codebase
Use Advanced Mode instead when you have a precise data model in mind and want explicit control over every entity, field, and endpoint.
How It Works
- Enter your application description in the prompt field on the home page
- Click Synthesize or press
Ctrl + Enter - A free Spark build starts immediately and lands you on the delivery page with a demo app running in your browser. Spark is a fixed template, so nothing is interpreted at this step.
- Choose a paid tier from that page and complete checkout
- The generation pipeline reruns against your description — for Boilerplate and Infrastructure the model reads your prose directly and emits the entities, repositories, endpoints, migration, and typed frontend client in one pass
- Both halves are compiled, then your archive is packed and made available
There is no schema-review step in Simple Mode. Your description goes to the model as written. If you want to see and edit the entity model before you buy, use Advanced Mode — its wizard is where the ER canvas lives, and the schema you build there is what the paid run generates from.
Writing a Good Prompt
The quality of your output is directly tied to the clarity of your description. Here are guidelines for getting the best results:
Be Specific About Entities
Name the core "things" in your system. The model maps these to database tables and API controllers.
Vague:
"A todo app"
Better:
"A task management app with users, projects, tasks, and labels. Tasks belong to projects and can be assigned to users."
Best:
"A task management SaaS. Users belong to workspaces. Workspaces have projects. Projects have tasks with fields: title, description, status (todo/in-progress/done), priority (low/medium/high/urgent), due date, and assignee. Tasks can have multiple labels. Users can comment on tasks."
Describe Relationships
Explicitly state how entities connect to each other:
| Relationship Language | What It Maps To |
|---|---|
| "Users belong to organizations" | FK: users.organization_id → organizations.id |
| "Projects have many tasks" | FK: tasks.project_id → projects.id |
| "Tasks can have multiple tags" | Junction table: task_tags |
| "Each order has one customer" | FK: orders.customer_id → customers.id |
Mention Key Workflows
If there's a meaningful business process, describe it:
"When an order is placed, inventory is decremented. If inventory reaches zero, the product is marked as out-of-stock."
The model uses this context to generate service logic in the appropriate layer.
Field Types and Constraints
You don't need to be exhaustive, but hints help:
"Products have a price (decimal), stock count (integer), and a published flag (boolean)."
Example Prompts by Category
E-commerce
A multi-vendor e-commerce platform. Vendors have stores. Stores have products
with categories, variants (size, color), and inventory per variant.
Customers place orders containing order items. Orders have a status workflow:
pending → confirmed → shipped → delivered. Payments are tracked per order.
B2B SaaS
A CRM for small businesses. Organizations have team members with roles
(admin, member). Contacts belong to organizations. Deals are linked to contacts
and have stages (lead, qualified, proposal, won, lost) and a dollar value.
Activities (calls, emails, meetings) are logged against deals or contacts.
Marketplace
A freelancer marketplace. Clients post projects with a budget and description.
Freelancers submit proposals with a price and cover letter. Clients select a
proposal to create a contract. Contracts track milestones and payments.
Both parties can leave reviews after contract completion.
After Submission
After you submit a Simple Mode prompt:
- A free Spark build runs immediately (fixed demo app, no AI call, counts against your 5 builds per month)
- You land on the delivery page and pick a tier
- After checkout, the model (Claude Sonnet 5.5 by default) reads your prompt and generates the project
There is no interpreted-schema preview in Simple Mode. If you want to see and correct the entity model before paying, use Advanced Mode instead. Your progress shows on a status page that refreshes itself every few seconds.
Limitations
- Simple Mode relies on LLM interpretation — ambiguous prompts may produce a model of your domain that you need to adjust in the code afterwards
- Very complex systems with 15+ entities work better in Advanced Mode
- Domain-specific logic (complex validation rules, multi-step workflows) is better modeled explicitly in Advanced Mode