LaunchDarkly announced
multiple platform innovations at its annual Galaxy user conference to help
engineering and product teams deliver with both high velocity and lower risk.
With the rise of AI-generated code, development teams are no longer just navigating
faster development cycles, they're facing an unprecedented surge in code volume
that dramatically expands the surface area for bugs, broken experiences, and
application outages.
The latest
capabilities at LaunchDarkly give teams the tools they need to innovate
boldly-without exposing customers or businesses to unnecessary risk. By
bringing observability, AI controls, and analytics directly into the release
process, LaunchDarkly is enabling engineering and product teams to ship with
confidence, respond to application issues, and continuously improve the user
experience.
"Software
used to evolve quarterly. Today, it changes by the hour. And with AI systems
adapting in production, often unpredictably, release management at feature
level granularity has become mission-critical," said Dan Rogers, CEO of
LaunchDarkly. "Teams need the ability to ship with precision, respond in real
time, and continuously optimize what's live. That's what LaunchDarkly delivers:
a safer, smarter way to build and release software in an AI-powered world."
Platform
Updates Introduced at Galaxy '25:
Guarded
Releases - Observability at the Point of Release
Guarded
Releases pair progressive rollouts with real-time monitoring, automated
rollback, and feature-level observability. Teams can now identify regressions
instantly and correlate them directly to specific changes, preventing incidents
before they impact users. With the recent integration of Highlight.io,
LaunchDarkly extends observability to include telemetry data like metrics, logs
and traces at the point of release.
AI Configs -
Runtime Control Plane for Model and Prompt Management
AI Configs
give teams a centralized control plane to manage prompt and model
configurations for AI-powered applications. Teams can safely iterate in
production, monitor key metrics like cost and latency, and deploy fallback
strategies when things go wrong without any code changes. This reduces risk
while accelerating the development of AI features.
Warehouse-Native
Experimentation & Product Analytics
LaunchDarkly
now gives teams real-time insights into user behavior and feature engagement,
powered directly by their data warehouse. With warehouse-native experimentation
and product analytics, teams can quickly understand what's working, what's not,
and how every feature impacts business outcomes. The recent integration of
Houseware strengthens these capabilities by making it easier to run
experiments, analyze results, and iterate faster, all within the existing data
ecosystem.
"Generative AI is
fundamentally changing the relationship between the code we build, the code we
deploy, and the code we maintain in production. Experimentation, understanding
user behaviour, is now a necessity, not a luxury," said James Governor, RedMonk
co-founder. "LaunchDarkly is building observability into its core offerings,
deepening its focus on analytics, and doubling down on release management to
create an integrated platform for progressive delivery in the AI era."
Availability
Guarded
Releases, AI Configs, and Warehouse-Native Experimentation & Product
Analytics are generally available today. Advanced observability features within
Guarded Releases, including error monitoring, session replay, and telemetry
integrations, are available in early access.