Platforms 1
The feature flag market can look like a long list of products with similar checkboxes. Most support Boolean flags, targeting and percentage rollout. The useful distinction is not who has a toggle. It is what system each product is trying to make feature flags part of.
Five different starting points
| Platform | Architectural starting point |
|---|---|
| GitLab | software delivery and DevOps workflow |
| Datadog | production telemetry and runtime release safety |
| LaunchDarkly | dedicated feature management |
| Harness FME | software delivery plus experimentation |
| Unleash | open source feature management and infrastructure control |
GitLab: feature flags inside the delivery platform
GitLab exposes an Unleash compatible feature flag API and supports common strategies such as all users, percentage rollout, user IDs and user lists. The attraction is consolidation: source, merge request, pipeline, deployment and feature control can live in one platform.
The tradeoff is that GitLab is not trying to be the deepest experimentation or telemetry driven feature management system.
Datadog: feature flags inside observability
Datadog Feature Flags is built on OpenFeature and connects flag decisions to Datadog telemetry. Its strongest story is not merely turning a feature on. It is progressively exposing a change while watching production signals, and using canary guardrails to pause or stop when regressions appear.
LaunchDarkly: feature management as the product
LaunchDarkly has spent years building around targeting, release controls, environments, governance, experimentation and SDK coverage. It remains an important benchmark because feature management is not an extension of another product category; it is the center of the platform.
Harness FME: feature management in a delivery system
Harness FME, which carries the Split heritage, combines flags with experimentation, release monitoring, approval workflows and Harness pipelines. It is particularly interesting when an organization wants feature changes to participate in deployment governance rather than live in a separate dashboard.
Unleash: control and self hosting
Unleash is attractive when open source, self hosting, data control and predictable deployment architecture matter. It also provides a useful counterweight to SaaS only control planes.
The wider market
Other credible products solve adjacent versions of the same problem:
- Flagsmith combines SaaS with open source and self hosted options.
- Statsig leans strongly toward product experimentation and analytics.
- GrowthBook is attractive for open source and warehouse native experimentation.
- AWS AppConfig treats feature flags as part of safe runtime configuration on AWS.
- Azure App Configuration brings feature management into the Azure application stack.
- Firebase Remote Config is especially relevant for mobile and consumer applications.
- ConfigCat emphasizes straightforward feature flag delivery and local evaluation.
- DevCycle has strongly aligned its developer experience with OpenFeature.
The market is expanding toward AI runtime control
The AI era is pushing some feature management vendors beyond conventional flags. LaunchDarkly's AgentControl manages prompts, model settings, targeting, evaluation and agent workflows. Harness AI Configs manages prompts, models and agent parameters with schemas, approvals and audit history. Statsig is extending its experimentation model to versioned prompts, model configuration, offline evaluation and production evaluation; its current documentation marks AI Experimentation as Early Access.
This does not mean every feature platform is becoming an AI control plane, or that every organization needs one. It does show that the category is broadening from feature exposure toward governed runtime behavior.
Do not start with the longest feature list
A platform with more features can still be the wrong choice if the organization only needs controlled release toggles. A simpler platform can be the wrong choice if production health, experimentation or compliance approvals are core requirements.
The next lessons look at the major platforms individually before the series compares them across cost, operations, adoption and architecture.
Feature Flags series
← Testing Applications That Use Feature Flags GitLab Feature Flags →