Last year, our team introduced Claygent, an AI web researcher, to the world. We have been amazed by the creative ways our community has used Claygent for a wide range of research applications, helping teams accomplish tasks that used to take hours of manual work in minutes.
With Claygent, our goal was to help users go beyond the limitations of traditional data providers, turning the entire internet into their canvas for prospecting and go-to-market research.
Today, we are thrilled to announce a new Claygent model, Neon—available to use only through Clay credits. Read more about it below, or watch this video for a live demo and walkthrough.
Robust capabilities for complex web research
Claygent Neon is our most capable model yet, able to handle a wider range of tasks than before with higher quality and more customizable answer formatting.
- Wider Range of Tasks: Claygent Neon can help you successfully accomplish a broader spectrum of research tasks than before. For example, we’ve seen significant capability jumps in parsing PDFs and harder-to-find information buried on the web. (E.g. “Given this financial 10k, find out if executive pay is tied to environmental goals. Answer Yes or No.")
- Higher Quality: In our internal testing, we’ve found that Neon performs with the highest accuracy (less null / inaccurate replies) relative to any other model option on a representative sample of Claygent queries.
- Reliable and Flexible Answer Formatting: Neon offers a better way to format Claygent answer outputs. Instead of getting outputs back in a single text cell, you can split the components of any given output across multiple columns (see photo below). You can use this improved data structuring to better segment information and trigger downstream workflows such as outbound messaging.
How to use Claygent Neon for your creative lead research
Use Claygent Neon in all the ways you’ve used it in the past: to fill in the gaps for unique web research tasks. It’s all about finding and creating data where data may not exist!
For example, you can ask Claygent to:
- Infer a company’s competitors
- Check how many pricing plans it offers
- Score prospects with unique criteria and web research
- Find recent initiatives, news, etc. and indicate its date
- Research industry trends
- Filter for SOC-II compliance
- Find if a company sells through a reseller or distributor
- Figure out the head of a firm or function (e.g., ideal buyer profile)
- Determine if they are an influencer agency
- General product descriptions based on social media/website
- Check whether a company is venture backed
- … and much more
How to select models and credits
To select a model, use the dropdown selector.
Claygent Neon will cost two credits. Other models are still available for use, including GPT-4o, GPT-3.5 Turbo, Claude Opus, Claude Sonnet, and Claude Haiku, each with an associated credit cost. You also have the option to use your own AI API key and not use any credits.
When you’re deciding which model to use, we recommend playing with several options. In general, Neon has the highest accuracy whereas GPT3.5 and Haiku are the most cost effective. The best thing to do is try a few out and see works best for your specific type of query.
