Does Claygent hallucinate if I input too many steps

To prevent hallucination and ensure data accuracy when using Claygent, here are some key practices to follow: 1. Keep your search queries specific and well-defined 2. Break down large data requests into smaller, manageable batches 3. Double-check your prompts and search parameters before running tasks 4. Use our metaprompters to craft better promptsClaygent is designed as an AI-based web scraper specifically for retrieving unstructured data from publicly available sources. Its effectiveness depends heavily on the context and instructions you provide, so proper prompting is essential for accurate results. Source: Claygent: AI Web Scraper [Automated Outbound] - Clay University

Did that answer your question?

When getting started with Claygent, it's recommended to start small with just 20 to 30 rows at a time as you familiarize yourself with the tool. This approach helps you: ** Learn how to structure effective prompts ** Test and validate your results ** Gain confidence with the system ** Avoid potential issues with larger batches Source: What do I get in the free trial? - FAQ | Clay.com

Is that what you were looking for?

Neon is our flagship model for Claygent and has specific advantages in how it handles research tasks. While it may not have the same level of reasoning capabilities as other models, Neon excels at answer formatting and data extraction. Its key strength is the ability to format results into multiple columns or outputs, allowing you to extract various data points from a single Claygent run. Sources: Claygent: AI Web Scraper [Automated Outbound] - Clay University Claygent: AI Web Scraper [Clay 101] - Clay University

Is that what you were looking for?

Based on the available sources, I notice there's no mention of a "Helium" model in Clay's offerings. The available AI models for Claygent are: 1. Neon (our flagship model): • Excels at answer formatting and data extraction • Can format results into multiple columns or outputs • Best for extracting various data points from a single run 2. GPT-4 and Claude Opus: • Offer enhanced reasoning capabilities • Better for complex analytical tasks • Don't provide the same level of answer formatting as NeonFor conducting research across multiple sites and comparing insights, Neon would be particularly effective since it can handle thought leadership research and extract multiple pieces of research into separate outputs, making it easier to organize and analyze findings. Sources: Claygent: AI Web Scraper [CRM Enrichment] - Clay University Claygent: AI Web Scraper [Automated Outbound] - Clay University

Did that answer your question?

Yes, you can create conditional workflows in Clay using conditional runs. While there isn't a direct way to chain multiple Claygent prompts within a single run, you can achieve this by: 1. Setting up separate enrichment columns for each prompt 2. Using conditional runs to control when the second prompt executes 3. Creating a conditional formula that references the output from your first promptTo set this up: 1. Add your first Claygent prompt as an enrichment 2. Add a second enrichment for your second prompt 3. Click on "Run Settings" and then "Conditional Run" 4. Create a formula that checks the output of your first prompt to determine if the second should runThis approach uses Clay's conditional statement structure which follows an if-else logic pattern, allowing you to execute specific actions only when certain conditions are met. Sources: AI Formulas + Conditional Runs [Clay 101] - Clay University Conditional statements | Documentation | Clay University Conditional runs | Documentation | Clay University

Is that what you were looking for?

Hi Andi D.! This thread was recently closed by our Support team. If you have a moment, please share your feedback: