When you want Retool Assist (or any builder AI) to “leave no stone unturned”, your prompt must do four things clearly:
1. Assign a Strong Role
Instead of saying “help me build…”, you say:
“You are an expert full-stack Retool engineer…”
Why it matters:
It forces the AI to think in Retool-native concepts (queries, workflows, drawers, tables, AI actions).
It avoids vague UI or backend assumptions.
2. Separate Task from Context
Bad prompt:
“Build a client onboarding app for agencies.”
Good prompt:
Task = what to build
Context = why it exists
This helps the AI:
Make better architectural decisions
Choose the right components and workflows
3. Over-specify Structure (Databases, UI, Workflows)
Retool Assist performs best when you:
Explicitly list tables
Explicitly list fields
Explicitly list workflows
Why:
Retool is schema-driven
The AI maps your words → actual Retool objects
Think of it as “designing the database in English.”
4. Tell the AI Where to Use AI 
Instead of:
“Generate a summary”
You specified:
What data to read
What sections to output
Where to save the result
This avoids:
Generic summaries
Unusable blobs of text
5. Finish with “Polish & Guardrails”
This is where quality jumps:
Validation
Loading states
Notifications
Clean UI
This signals: “This is a production SaaS, not a demo.”
Final Optimized Prompt (Retool Assist–Ready)
You can paste this entire block directly into Retool Assist (Beta).
Detailed Prompt (Improved + Fully Structured)
Role:
You are an expert full-stack Retool engineer experienced in building internal tools, workflows, and automations for agencies and SaaS businesses.
Task:
Build a complete Client Onboarding & Project Summary Automation Tool inside Retool.
Context:
Agencies and freelancers waste hours collecting client details across emails, WhatsApp, Google Drive, and calls. Important information gets lost, onboarding is slow, and teams start projects without clarity. This tool centralizes intake, file uploads, and AI-powered summaries into one clean workflow.
- Database Structure
Create two tables:
clients
id (uuid, primary key)
business_name (text)
contact_name (text)
email (text)
phone (text)
business_description (long text)
services_needed (text)
goals (long text)
pain_points (long text)
ai_summary (long text)
created_at (timestamp, default now)
client_files
id (uuid)
client_id (uuid, foreign key → clients.id)
file_name (text)
file_url (text)
created_at (timestamp, default now) - Interface Requirements
Client Intake Modal
Business name
Contact name
Email
Phone
Business description
Services needed
Goals & objectives
Pain points & challenges
File upload (multiple files allowed)
Include instruction text explaining what clients should upload.
Clients Table
Display:
Business name
Contact name
Email
Services needed
Created date
Add actions:
View full details
Generate / regenerate AI summary
Client Details Drawer
Show:
All client fields
Uploaded files (clickable)
AI-generated summary
Button to regenerate summary - AI Summary Generation
Use Retool AI to generate a structured summary using:
Client form data
Uploaded file references
Output sections:
Business overview
Problems identified
Project goals
Recommended services
Technical considerations
High-level budget & timeline estimate
Save output to clients.ai_summary. - Workflows
Create Client
Insert client record
Upload files
Save file metadata
Trigger AI summary generation
Select Client
Fetch client data
Fetch related files
Display drawer
Regenerate Summary
Re-run AI action
Update summary field - Search & Filters
Search by name or email
Filter by service type - Finishing Requirements
Clean, modern UI
Loading states for all queries
Success and error notifications
Field validation
Smooth file uploadsTART HERE TO GET USEFUL FEEDBACK FASTER-->
- My goal: To build a smart client onboarding system that eliminates back-and-forth communication, organizes client files automatically, and instantly generates AI-powered project briefs so teams can start projects faster and with clarity.
- Issue: Agencies and freelancers currently collect client information across scattered tools like email, WhatsApp, Google Drive, and calls. This leads to lost context, unclear requirements, delayed project starts, and repeated clarification cycles.
- Steps I've taken to troubleshoot: Designed a structured client intake form inside Retool. Created a centralized database for client data and uploaded files. Implemented automated file uploads and storage. Integrated AI inside Retool to summarize client inputs into actionable project briefs. Built a searchable admin dashboard for managing all clients in one place
- Additional info:
Deployment**:** Cloud (Retool Cloud).
Features**:** Client intake, file uploads, AI summaries, admin dashboard.
Artifacts**:** Screenshots and demo video included.
Scalability**:** Can be sold as a SaaS to agencies, studios, and consultants.
Community Show & Tell