LaunchDarkly

Founded by Edith Harbaugh in 2014

Manage feature flags and control releases and AI agents safely in production

Pricing model

Free Tier, Paid

Rating

★ People love it

Time to value

Quick Setup (< 1 hour)

You can use LaunchDarkly to control how software features and AI models are released to your users. Instead of bundling code into risky all-or-nothing deployments, you can toggle features on or off instantly with runtime feature flags. You can roll out updates to targeted user segments, conduct continuous experiments like A/B tests, and automate rollbacks if errors spike. The system also gives you control over autonomous AI agents in production, letting you monitor agent behavior, evaluate outputs, and adjust model configurations without pushing new code.

What LaunchDarkly does

Create a boolean or multivariate feature flagDefine custom user targeting rules based on server attributesConfigure automated rollback thresholds based on error ratesRoute queries between different language modelsEvaluate AI agent responses with custom judgesRun multi-armed bandit experiments in productionAudit user access controls and permissions across environmentsControl feature flags dynamically across production environmentsTarget feature rollouts to specific user segments and custom attributesTrigger automated rollbacks when system errors or anomalies occurRun production experiments including A/B/n testing and multi-armed banditsMonitor and steer AI agent outputs and behavior in real timeDynamically route AI traffic between different large language modelsManage prompt changes without deploying new code

Pricing breakdown

Pricing model: Free tier, Paid

PlanList price

DeveloperFree

Foundation (per Service Connection) (annual billing)$10 / mo, billed annually

Tutorials & Demos

Frequently asked

AWS CloudWatch, Datadog, Slack, Jira, GitHub, GitLab, Bitbucket, Azure DevOps, CircleCI, PagerDuty, New Relic, Sentry, Segment, Amplitude, Google Analytics, Microsoft Teams, Trello, Zendesk, Snowflake, Splunk, Jira Software, Slack / Microsoft Teams, Datadog / New Relic, Terraform, Snowflake / BigQuery, Segment / Amplitude, GitHub / GitLab, Jira/Trello, Dynatrace, ServiceNow, Jira (Feature-to-ticket linking), Slack (Notifications and controls), Datadog (Event overlays), GitHub Actions (CI/CD triggers), Terraform (Infrastructure as Code), GitHub Actions