Fastly for AI

Product · Hardware / device · Engineering & development by Fastly

Fastly for AI is a specialized platform designed to govern, control, and protect AI workloads and agentic traffic. It features the AI Accelerator, which uses semantic caching to map queries to concepts, reducing redundant LLM calls and saving on token costs. The suite includes an AI Firewall to stop prompt injection attacks in real-time at the edge and AI Runtime Control for managing model usage across multiple vendors. These tools provide businesses with the visibility and security needed to deploy generative AI applications at scale while maintaining budget and performance.

Free AI Runtime Requests
100,000 per month
Free AI Accelerator Requests
20,000 per month
AI Traffic Growth Rate
6.5X faster than human traffic

Overview

Fastly for AI is a specialized platform designed to govern, control, and protect AI workloads and agentic traffic. It features the AI Accelerator, which uses semantic caching to map queries to concepts, reducing redundant LLM calls and saving on token costs. The suite includes an AI Firewall to stop prompt injection attacks in real-time at the edge and AI Runtime Control for managing model usage across multiple vendors. These tools provide businesses with the visibility and security needed to deploy generative AI applications at scale while maintaining budget and performance.

AI Governance and Control

  • AI Accelerator: Implements semantic caching for LLM APIs to reuse similar AI responses and reduce token costs.
  • AI Runtime Control: Provides an intelligent control plane to route AI traffic, set budget caps, and apply rate limits.
  • AI Firewall: Shields AI-enabled applications from LLM-specific threats like prompt injection at the edge.
  • AI Bot Management: Enables organizations to block, throttle, or monetize AI crawlers scraping their content.

AI Observability and Tooling

  • Unified AI Visibility: Offers searchable prompt and completion logs with token attribution for full visibility.
  • Fastly MCP Server: Allows management of Fastly infrastructure through natural language interactions using AI developer tools.
  • Fastly Agent Toolkit: Provides an open-source collection of skills to teach coding agents how to interact with Fastly.

Pricing

AI Runtime Control (per 10k Requests)
$2.50 (after first 100k free)
AI Accelerator (per 1k Requests)
$0.40 (after first 20k free)

Specifications

Free AI Runtime Requests
100,000 per month
Free AI Accelerator Requests
20,000 per month
AI Traffic Growth Rate
6.5X faster than human traffic

Added by AntSpace from public sources (fastly.com).

Related

More products by Fastly