Using Backend Data Sources in SmartForms

FormWise allows SmartForm owners to attach backend data sources to their AI-powered tools. These sources help the AI provide more accurate, context-aware responses by referencing pre-uploaded documents, scraped URLs, and other structured data.

✅ What Can Be Used as a Backend Data Source?

• 📂 Pre-Uploaded Files: PDFs, CSVs, TXT files

• 🌍 Web Scraping: Extracting content from URLs

• 🔍 Web Browsing: Searching the web for real-time metadata

• 🔗 Multiple Prompt Steps: Using different data sources across prompt chaining

By attaching backend data before user submissions, you ensure that AI responses are aligned with your industry, niche, or proprietary knowledge. However, too many files or combining multiple retrieval methods can cause timeouts or token issues—this guide will help you optimize performance.

---

1️⃣ Uploading Backend Data Sources in SmartForms

🔹 Where to Upload Backend Data Sources

📌 Before attaching them to a SmartForm, all backend data sources must be uploaded in the "Documents" tab inside the SmartForm builder.

🔹 Steps to Upload: 1️⃣ Go to the SmartForm builder → Navigate to the "Documents" tab. 2️⃣ Click "Upload Document" and select a PDF, CSV, or TXT file. 3️⃣ If you want to scrape a URL, enter the URL in the provided field. 4️⃣ Once uploaded, these documents & scraped URLs are now available to reference inside your SmartForm’s prompt instructions.

---

2️⃣ Attaching Backend Data Sources to Prompt Steps

Once documents are uploaded, you can attach them to specific prompt steps to guide how AI should use the data.

🔹 How It Works: ✅ Each SmartForm prompt step can have its own backend data sources attached. ✅ You can include pre-uploaded documents, scraped URLs, or enable web browsing within each step. ✅ If using multiple prompt steps (prompt chaining), you can attach different data sources at each step.

📌 Example Use Case: Step 1: AI pulls insights from a pre-uploaded industry report (PDF). Step 2: AI references a scraped competitor website (URL) for comparisons. Step 3: AI searches real-time trends via web browsing before generating the final response.

🚨 Too many data sources in one step can overload AI processing—spread them across multiple steps when needed.

---

3️⃣ Best Practices for Using Backend Data Sources

🔹 File Size Limits (Per Upload)

📌 Pro Plan: 2MB max per file (multiple files allowed). 📌 Agency Plan: 20MB max per file (multiple files allowed).

🚀 You can attach multiple files per SmartForm, but too many large files can cause token issues and slow performance.

🔹 No Hard Limit on File Uploads, But Be Cautious

✅ There’s no upper limit on the number of files you can attach. 🚨 However, AI only processes a certain amount of text at a time. If too much information is loaded, responses may be incomplete, slow, or cause a timeout.

🔹 Using Web Scraping & Web Browsing

🚨 Avoid combining all retrieval methods (file uploads, web scraping, and web browsing) in one step. 📌 Web Scraping (URLs): The AI extracts content from the webpage and treats it like an uploaded file. 📌 Web Browsing: AI searches for real-time metadata from search engine results and summarizes findings.

✅ Best Practice: If your SmartForm already has uploaded PDFs and CSVs, avoid adding unnecessary web browsing or web scraping unless the request specifically needs it.

---

4️⃣ How Retrieval Works (RAG Explained in Simple Terms)

When AI processes a backend data source, it does not "train" on the file but instead: 1️⃣ Extracts key insights from the uploaded file or scraped URL. 2️⃣ Retrieves relevant portions based on the SmartForm prompt. 3️⃣ Generates a response using both AI knowledge + retrieved data.

🚀 Think of it as a "smart document search" where the AI finds and summarizes what matters most.

---

5️⃣ Customizing How AI Uses the Data (Prompt Injection in Data Sources)

Each backend data source allows you to inject a short instruction that tells AI how to use the uploaded content.

🔹 Default Instruction: 📌 "Use this data to substantiate your response."

🔹 Best Practices for Custom Instructions: ✅ If analyzing financial reports, use: "Use this data to generate financial summaries." ✅ If comparing competitor websites, use: "Use this scraped data to compare industry trends." ✅ If referencing a policy document, use: "Use this PDF to answer questions about our policies."

📌 Customizing this instruction ensures AI retrieves only relevant data instead of processing everything at once.

---

6️⃣ Avoiding Timeouts & Optimizing Performance

If your SmartForm tool is timing out, try these fixes: ✅ Reduce file sizes (split large PDFs into sections). ✅ Limit the number of attached documents per prompt step. ✅ Avoid combining multiple retrieval methods in the same step. ✅ Use precise prompt instructions to make AI focus on the right data.

📌 Example Fix: ❌ Bad: "Analyze the entire 50-page report and summarize it." ✅ Better: "Find and summarize the top 3 insights from the report."

---

🚀 Conclusion

Backend data sources make SmartForms more powerful by allowing AI to reference custom documents, structured data, and real-time information.

✅ Upload your files first in the “Documents” tab before attaching them to SmartForm prompts. ✅ Be mindful of file size limits & avoid combining too many retrieval methods in a single step. ✅ Use clear prompt instructions to optimize how AI retrieves data.

By following these best practices, you’ll ensure that your SmartForms run efficiently and deliver high-quality AI-powered responses. 🚀

4o