What Claude (Anthropic) with web search answered for feature flag platforms (agencies) on 2026-09-22

Claude (Anthropic) with web search claude-opus-5 asked 2026-09-22

Coverage: Claude (Anthropic) with web search. Other engines not included.

The question asked, verbatim: best feature flag platforms for agencies

Brands named, in the order the engine gave

1. ConfigCat recommended
   ConfigCat - my default recommendation for agencies.
2. Flagsmith mentioned
   Flagsmith - the pick when clients have data-residency or on-prem requirements, or when you want the handoff story to be clean.
3. AWS mentioned
   It's fully open source with three deployment models: managed cloud, single-tenant private cloud in the client's own AWS account, and self-hosted.
4. Unleash mentioned
   Unleash - if you want to self-host one instance and bill it as infrastructure across all clients, this is the most mature option.
5. GrowthBook mentioned
   GrowthBook - pick this when the engagement is "improve conversion," not just "ship safely." It's flags plus genuine warehouse-native A/B analysis, so you can point at a statistically defensible lift in the retainer review instead of a screenshot.
6. Flipt mentioned
   Flipt - the lightweight choice for small projects.
7. Statsig mentioned
   Statsig - worth knowing if a client is doing serious product experimentation.

Sources the engine cited

The answer, as recorded

For an agency the constraints are different from a product team: you run many small projects at once, traffic per client is low and lumpy, your headcount on each project rotates, and at some point you have to hand the thing over to the client without leaving them with a bill they didn't agree to. That pushes hard away from per-seat and per-MAU pricing. My ranking: **1. ConfigCat** - my default recommendation for agencies. It bills on config-fetch requests, not seats or MAUs, with unlimited team members on every plan. That's exactly the agency shape: you can add a freelancer for a two-week sprint, add a client's internal dev as a viewer, and nothing changes on the invoice. It supports multiple products/projects/environments under one org, so each client gets its own boundary. Free tier is real but capped at 10 flags, so you'll outgrow it per-project fairly fast. ([ConfigCat pricing/alternatives](https://configcat.com/blog/top-launchdarkly-alternatives/), [multi-tenant targeting](https://configcat.com/blog/how-to-target-features-by-tenants/)) **2. Flagsmith** - the pick when clients have data-residency or on-prem requirements, or when you want the handoff story to be clean. It's fully open source with three deployment models: managed cloud, single-tenant private cloud in the client's own AWS account, and self-hosted. That means you can run one shared instance for your small clients and stand up a dedicated instance for the regulated ones, using the same SDKs and the same team knowledge. Paid starts around $45/3 users/month, and the free tier is 1 user / 50k requests - the per-user pricing is the weak spot versus ConfigCat. ([Flagsmith vs LaunchDarkly](https://www.flagsmith.com/compare/flagsmith-vs-launchdarkly), [flagsmith.com](https://www.flagsmith.com/)) **3. Unleash** - if you want to self-host one instance and bill it as infrastructure across all clients, this is the most mature option. Founded 2015, very broad SDK coverage, and the enterprise tier has the governance features (change-request approvals, audit logging) you'll need the day a client asks who turned the flag on. Cloud is free to 1,000 MAU with unlimited seats, then Developer ($10-500) and Business ($500-10,000). The MAU model is the catch - fine for B2B clients, expensive for consumer ones. ([Unleash pricing comparison](https://www.getunleash.io/blog/feature-flag-tools-which-should-you-use-with-pricing), [open source comparison](https://flagshark.com/blog/open-source-feature-flag-tools-compared-2026/)) **4. GrowthBook** - pick this when the engagement is "improve conversion," not just "ship safely." It's flags plus genuine warehouse-native A/B analysis, so you can point at a statistically defensible lift in the retainer review instead of a screenshot. Self-hostable, so per-client cost is near zero. It's more tool than a brochure-site client needs. ([GrowthBook comparison](https://www.growthbook.io/blog/best-open-source-feature-flagging-tools-compared)) **5. Flipt** - the lightweight choice for small projects. v2 is Git-native: flag state lives in a Git repo with no database, so flag changes ride your existing PR/CI review. For an agency that means a client project's flags are just files in the repo you already hand over - the cleanest possible offboarding. Trade-off is no real targeting/experimentation depth and a cloud offering still in progress. ([Flipt in 2026 comparison](https://dev.to/sai_ram_0000/tombstone-vs-unleash-vs-flagsmith-vs-flipt-vs-growthbook-feature-flag-platforms-compared-2026-5cb3)) **6. Statsig** - worth knowing if a client is doing serious product experimentation. Unlimited flags at every tier including free, priced on events. Good product, but it's a fit for one client's product org rather than a platform you standardize your whole agency on. ([Kameleoon roundup](https://www.kameleoon.com/blog/top-feature-flag-management-tools)) **LaunchDarkly I'd skip for agency work** despite it being the category leader. Enterprise contracts start around $25K/year and MAU-based pricing at 50K MAU lands in the $100-150K/year range - that only makes sense when one client is big enough to buy their own seat, in which case they procure it, not you. ([pricing detail](https://rollgate.io/blog/feature-flags-pricing-comparison)) Practical note on handoff: whatever you choose, keep each client in a separate project/org from day one rather than separate environments inside one project. Untangling a shared project at handoff is far more painful than the small overhead of setting it up separately at the start. Sources: - [The 10 top feature flag management tools in 2026 | Kameleoon](https://www.kameleoon.com/blog/top-feature-flag-management-tools) - [Top LaunchDarkly Alternatives in 2026 | ConfigCat](https://configcat.com/blog/top-launchdarkly-alternatives/) - [Multi-Tenant Feature Flags | ConfigCat](https://configcat.com/blog/how-to-target-features-by-tenants/) - [Flagsmith vs LaunchDarkly](https://www.flagsmith.com/compare/flagsmith-vs-launchdarkly) - [Flagsmith](https://www.flagsmith.com/) - [Feature flag tools: Comparison and pricing | Unleash](https://www.getunleash.io/blog/feature-flag-tools-which-should-you-use-with-pricing) - [Open Source Feature Flag Tools Compared | FlagShark](https://flagshark.com/blog/open-source-feature-flag-tools-compared-2026/) - [8 Best Open-Source Feature Flagging Tools Compared | GrowthBook](https://www.growthbook.io/blog/best-open-source-feature-flagging-tools-compared) - [Tombstone vs Unleash vs Flagsmith vs Flipt vs GrowthBook (2026) | DEV](https://dev.to/sai_ram_0000/tombstone-vs-unleash-vs-flagsmith-vs-flipt-vs-growthbook-feature-flag-platforms-compared-2026-5cb3) - [Feature Flags Pricing Comparison 2026 | Rollgate](https://rollgate.io/blog/feature-flags-pricing-comparison)

Markdown edition · the question page, with every recording · Answer Watch · AEO Watch

Answer Watch records what one AI answer engine said on one date, nothing more. Each page names the engine and the model, carries the date the question was asked, and lists the brands that appeared in the answer in the order the engine gave them, with the sources it cited. It is a measurement, not a ranking, not a recommendation and not a claim about the world. Brands that did not appear in an answer are not named on any public page. Coverage: Claude (Anthropic) with web search. Other engines not included.