Promptfoo is a leading provider of AI security solutions, helping developers and enterprises build secure, reliable AI applications. Based in San Francisco, California, Promptfoo is backed by Insight Partners, Andreessen Horowitz, and industry leaders.
Our core product is an open-source pentesting and evaluation framework used by 300,000+ developers. Promptfoo is among the most popular evaluation frameworks and is the first product to adapt AI-specific pentesting techniques to your application.
Ian Webster breaks down real-world agent risk: the "lethal trifecta" (untrusted input + sensitive data + an exfiltration channel), and how automated red teaming helps catch issues before production.
Listen to episode →Ian talks with swyx and Alessio about shipping Discord's Clyde, why Promptfoo shifted from evals to AI security, and what red teaming looks like for agents and RAG applications.
Listen to episode →Ian Webster joins TBPN to discuss Promptfoo's mission and what it takes to secure LLM apps and integrations, including automated red teaming to uncover data leaks and unintended behaviors.
Listen to episode →In this episode, a16z partner Joel de la Garza speaks with a trio of security experts, including Promptfoo founder Ian Webster, about the security implications of the DeepSeek reasoning model. Ian's segment focuses on vulnerabilities within DeepSeek itself, and how users can protect themselves against backdoors, jailbreaks, and censorship.
Listen to episode →A deep dive into Ian's evolution from shipping Gen AI products as an engineer to launching a cybersecurity company, the fascinating origin of Promptfoo, and key insights on the latest AI security trends.
Listen on Spotify →a16z General Partner Anjney Midha speaks with Promptfoo founder and CEO Ian Webster about the importance of red-teaming for AI safety and security, and how bringing those capabilities to more organizations will lead to safer, more predictable generative AI applications.
Listen to episode →Our groundbreaking research on AI censorship and content filtering in DeepSeek models has been widely covered by major technology and news publications.
Leading AI platforms have integrated Promptfoo into their official educational materials, recognizing it as an essential tool for LLM application development, evaluation, and security. These courses and workshops, developed in partnership with industry leaders, provide comprehensive training on building reliable AI applications.
Featured in OpenAI's Build Hour series, where they highlight that "Promptfoo is really powerful because you can iterate on prompts, configure tests in YAML, and view everything locally... it's faster and more straightforward."
A comprehensive nine-lesson course covering everything from basic evaluations to advanced model-graded techniques. Anthropic notes that "Promptfoo offers a streamlined, out-of-the-box solution that can significantly reduce the time and effort required for comprehensive prompt testing."
A comprehensive workshop designed to equip you with the knowledge and practical skills needed to effectively evaluate and improve Large Language Model (LLM) applications using Amazon Bedrock and Promptfoo. The course covers everything from basic setup to advanced evaluation techniques.
A hands-on guided project that teaches developers how to master model selection using Promptfoo. Learn to adapt to new models, handle pricing changes effectively, and perform regression testing through practical scenarios.
A detailed blog post by Semgrep's AI team explains their approach to evaluating LLM features and why they adopted Promptfoo as part of their workflow.
Read article →Download official Promptfoo logos and brand assets for media use.
For press inquiries, please contact our media relations team:
For urgent inquiries, please include "URGENT" in the subject line.
2024
San Francisco, California
Ian Webster, CEO & Co-founder
Michael D'Angelo, CTO & Co-founder
Insight Partners, Andreessen Horowitz, and industry leaders
