Changelog

We’ve just added remote feature configuration to Bucket. A dynamic, flexible approach to configuring feature behavior outside of your app – without needing to re-deploy it.

For example, you can use configs to configure AI models for your customers at runtime. With remote config, you can safely upgrade customers to the latest AI model or you can try different models for different segments of customers. If a new model is causing quality issues, you can easily roll back to a previous version.

You can also use configs to manage things like in-app copy. This means non-engineers can update the application on their own without touching the codebase.

How it works

Let’s say you’re using GPT-4o in your application and want the option to easily upgrade all or some customers to the latest version as soon as it arrives. Furthermore, you want to be able to give certain customers access to a competing model, Claude 3.7, to see if it outperforms GPT-4o.

Remote Config is based on JSON and is therefore highly flexible. You can decide which JSON payload to send to your application.

Here’s an AI configuration example:

All customers get gpt-4o from OpenAI as default except the two customers – Apex and Blaze – which get claude-3-7-sonnet from Anthropic.

When gpt-4o gets upgraded to the next version by OpenAI, you can simply go here to update the payload, or you can add a new, small beta segment to test the new OpenAI version, like so:

How to get started

To get started with Remote Config, you need to first upgrade to the latest version of our SDKs.

In the React SDK you can now use config when using useFeature:

const { config } = useFeature("sent-chat-message");

‍

As per our example in the screenshot above, if you’re authenticated as Adrian Borer or any user in the Logix or Hightrix companies, you’ll get this config:

Get user feedback on models

If you want to get user feedback on the models, you use Bucket’s requestFeedback, like so:

All feedback will be sent to the Feedback tab, right next to Remote Config, as well as to Slack, if you choose to.

To learn more about Remote Config, check out our documentation.

Happy shipping!

By default, when you create a new feature on Bucket, we'll suggest a feature key derived from the name. The format of the key is kebab-case by default. However, you may prefer to use a different naming convention. In some cases, it's just that you're used to SNAKE_UPPER_CASE from previous feature flagging providers. In other cases, a specific format might work better for the programming language of your choice.

Now you can define your feature key naming convention when creating a new feature. You can always customize the name of the key, but when picking a naming convention, Bucket will then enforce that new keys adhere to it.

When you create a feature today, the feature becomes available for use in your browser application through Bucket SDKs. That means the feature is retrievable through our public facing API using a publishable key. For most features, this is desirable because it lets you toggle features in your app UI. It also lets you build integrations right in your app.

However, once in a while you're working on something you need to keep secret from everyone. For this use-case, we've made it possible to mark features as secret on Bucket. Secret features are not retrievable with a publishable key — only with a secret key. That means you can use a feature on your backend and ensure that the feature key is never revealed to any users.

‍

When browsing companies or users, it’s much more fun to see company logos or user avatars, rather than randomly generated avatars.

So, we’ve shipped support for it!

‍

If you send the company logo URL as the avatar attribute on a company, we’ll display it in the UI. Similarly with avatars for users.

More updates soon!

‍

We’re always trying to make things simpler. As such, we’ve renamed the feature tabs and split them into tabs that each represent a clear use case. We’ve also improved the onboarding instructions per use case.

Clearer terminology

First, we’ve renamed  the “Targeting” tab to “Access”, which is easier to understand.his tab is all about controlling feature access.

Secondly, we’ve split “Analyze” into “Adoption” and “Feedback” to provide both of those use cases with more real estate. 

Here’s the new feature tabs – simple!

The new Feedback tab

The feedback tab includes a list of the latest feedback and a new satisfaction timeline that lets you easily track feature satisfaction trends over time. 

The satisfaction timeline gives you a rolling average of satisfaction scores for a feature alongside qualitative feedback. It’s a simple way to visualize overall feature satisfaction, evaluate whether a feature needs an iteration, and then track an iteration’s impact on satisfaction.

You can select different evaluation periods and timing windows to track the impact of feature iterations on satisfaction.

Use case-specific code examples

We’ve also added specific instructions and code examples to each tab so they’re relevant to each stage of the release process. 

When creating a feature, you're usually early in the development process and are only looking to implement a feature flag. That’s why the code example in the Access tab only contains the feature key. 

When you’ve reached the beta or general availability rollout stage, you’ll likely add adoption tracking and feedback collection. 

The code example in the Feedback tab looks like this:

Happy shipping!

You can now toggle features access directly on the company pages. This makes it even simpler to grant individual companies access to features in the Bucket UI.

In B2B, customers get feature access based on the subscription plan. However, it’s not unusual for them to be given access to features outside of their set subscription.

You can now grant individual companies access to features by flipping a toggle on the company page.

Another practical use case for toggles is enrolling early adopters into the Beta stage of a new feature.

Feature toggles don’t alter your existing feature targeting rules. They simply add additional companies to the targeting rules, which is useful for granting ad-hoc access without changing the core targeting rules of a feature.

Note: You cannot use feature toggles to disable feature access granted by targeting rules.

Happy toggling!

You can now use our adoption metrics and automated feedback surveys through OpenFeature.

OpenFeature is a small code layer that acts as an intermediary between your code and a feature flagging provider, letting you experiment with different providers with minimal changes to your codebase.  

OpenFeature recently added the Tracking API, which allows developers to send tracking events to providers whenever a user does something noteworthy. 

We’ve updated our OpenFeature Node.js SDK and OpenFeature Browser SDK to support this new API directly, making it easier to leverage our adoption metrics and feedback collection features if you use OpenFeature.

Happy shipping!

We’ve added support for deep linking to the Bucket app.

Simply replace the environment name and ID in the URL with “/current/”. 

For example, https://app.bucket.co/envs/current/features will direct you to the feature tab in Bucket.

This has also allowed us to directly link to the Bucket app in our documentation, making it easier to find the tabs or pages you need.

Happy shipping! 

We’re into the 4th week of polishing our UI, SDKs, and backend as part of our commitment to making a world-class product. 

We’re back after taking last week off to put together our roadmap for 2025 at our offsite in Berlin.

Here’s what we polished this week:

App improvements

• Let users create new features on Bucket by using flag.new 
• Added a tooltip to the targeting UI to clarify the “Some” targeting rule conditions
• Put a “Give feedback” button on the app sidebar
• Added ID columns to the Users and Companies tables
• Polished the Slack alerts for access rules and release stage changes

SDKs

• Added SDKVersion and ServiceName to feature flag events to make them simpler to debug
• The React SDK and Browser SDKs now use new-style exports
• Improved the automated survey documentation to explain how to position the modal when using requestFeedback

Bug fixes

• Improved the positioning of the automated survey popover widget so it always stays in the viewport
• Cleaned up legacy code and feature flags, which included making flag targeting validation stricter
• Fixed an occasional glitch where the “Hide feature sidebar” button didn’t appear in the Feature tab

Recap

With polish season coming to an end, this everything we've shipped over the past 4 weeks to make Bucket a robust, reliable, and intuitive product.

App improvements

• Overhauled the code instructions block. It’s simpler and actively listens for real-time checks or usage tracking
• Drastically improved internal data processing service start up times (reduced by 98.4%!) by initializing from stored aggregated values
• Included a feature view selector on the individual company screen under the Features tab
• New card design for the Analyze widgets.
• Simplified headers across the app.
• New card design for the Targeting environment rules.
• A rounded border-radius was added to buttons, inputs, and cards.
• Simplified the app and environment picker in the sidebar
• Polished the overall design of the feature sidebar
• Redesigned the main content area as a card
• Added a guide for feature entitlements
• Improved load times and stopped spinners from blocking the feature UI
• Updated the free trial indicator in the left sidebar for better clarity
• Added a new environment activity indicator to the feature sidebar
• Refactored some of the internal segment size estimation endpoints for increased simplicity
• Added a new profile menu to the bottom of the app sidebar

SDKs

React/Browser SDK

• Send feedback using the feature key instead of the feature ID.
• Added the requestFeedback function to allow users to implement the “Get feedback” button

Node.js SDK

• Now, it automatically sends user and company details to Bucket when either bindClient or getFeature(s) calls are made. This ensures that up-to-date information is pushed to Bucket so companies can be used immediately. Previously, you had to call updateCompany and updateUser manually.
• Introduced the enableTracking option to disable the sending of track, user, company, and flag events. This is useful when you’re impersonating other users, for example.
• ‍Updated and expanded the SDK documentation
• Introduced offline mode to improve the behavior of the SDK in tests and on localhost
• Added options to set local feature overrides for localhost development
• Improved logging
• Made improvements to the installation and getting started process
• Introduced file and environment variable-based configuration
• Fixed a rate limiter issue where it would keep accumulating data

Bug fixes

• Streamlined the STARS funnel step icons.
• Filters no longer show errors when you open the popover.
• Updated autocomplete selectors so they all work the same
• Reduced loading flicker in autocomplete selectors when a user is typing
• Changed how rules were stored to stop some flag rules from being misrepresented as OR instead of AND.
• Fixed an ingest issue where remote contexts overrode the given context
• Solved an occasional bug where users were prompted to create a new organization after receiving an invite link

Happy shipping!

The targeting UI is one of the most important screens in feature management tools, our ambition is to make the world's best targeting UI for B2B SaaS.

In speaking to users, it became clear that almost all B2B targeting rules focus on either specific companies or a segment of companies. Bucket company-level aggregation and segments make this simple, so we doubled down on these common use cases.

We’ve simplified the targeting UI to make it more intuitive. We’ve added three presets: 

• No one
• Some
• Everyone

These presets let you set targeting rules for features in the click of a button.

We’ve made defining the “Some” targeting rules simple while optimizing for the most common B2B use cases. You can add segments, individual companies, and individual users by simply clicking an “Add” button.

The whole team can be alerted to targeting rules updates through the Slack integration. A summary of all the targeting rule updates can be shared in the Slack channel of your choice. 

You can also still create advanced targeting rules when you need them with the "+ Add Rule" button.

Advanced rules let you specify rollout percentages and create targeting rules using company attributes, user attributes, feature targeting, or other contexts.

This is a great example of how Bucket’s focus on B2B lets us create a simpler product without sacrificing functionality.

Happy shipping!