Changelog

This week is all about performance.

Enhanced Content Ranking

Our AI-driven generative search is now more powerful! With our latest improvements, Brainfish can interpret and rank more and more documents in parallel when performing a search. This advanced feature enhances the accuracy, and relevance of generated answers, providing your customers with the most relevant information swiftly.

Widget Configurator Performance Fixes

We've addressed and fixed some performance issues with the widget configurator. Users can now enjoy a smoother and faster experience when customising their in-product widgets.

Improved Health Checks Loading Times

To bring you an even better user experience, we've significantly upgraded the load times of our system's health checks.

We heard that customer support teams spend weeks, sometimes months every year trying to figure out which content is out of date, broken or irrelevant.

So we simplified it.

Introducing Health Checks:

Health Checks enables you to instantly see a dashboard of everything that needs an objective review based on Brainfish's understanding of your documentation. Within Health Checks, you can immediately see:

• Broken article links: Hyperlinks that no longer work or navigate to a valid URL. Brainfish will even tell you what the error is, and where you need to fix it in the article
• Broken images: Similar to hyperlinks, broken images show up when the original place where the image was hosted no longer exists, and this image isn't displaying to customers anymore
• Stale content: Content that hasn't been updated in a period of time (by default 3 months) that needs an objective review. We can configure the stale time definition if you reach out to us at hello@brainfi.sh
• Inappropriate content: Publishing content that has swear words or profanity is, in almost every case, not desirable. So Brainfish will now easily highlight for you any content that exists that might be inappropriate for customers to read.

Content Health Checks has been designed based on feedback from all of our customers and industry leaders, that not only do they want to write better quality, customer centric content, but they also want to know which content needs to be reviewed and/or deleted.

We hope you enjoy health checks as much as we do!

Widget upgrades

As part of the latest release, we've also exposed widget answers via Javascript, so that engineers in your team can see what questions and answers a customer asked/received in a session + use this information to pass it to a ticketing system, live chat, or fill out forms.

Dark theme contrast improvements

Dark theme now works better on the Help Center! We've improved styling of hyperlinks (in tune with checking out broken links from Health Checks) so that they display at higher contrast on a dark Help Center theme.

Now, you can conveniently configure and preview our in-product widget right from the settings page. This provides you with optimum control over the widget via our UI. This means no need for incessant code changes or the intervention of your development team every time you need to update a suggested question or next best action. Take your customer service to a notch higher with personalised prompts and actions, or even have different widgets per page/user.

Performance Improvements & Bug Fixes

Steered by our commitment to delivering smooth performance, we've fixed some bugs and made other performance enhancements to make your Brainfish experience even better.

We're ushering in the new season with some major updates to keep your Brainfish experience secure, robust, and smart!

Significant Infrastructure Upgrade

We have executed a substantial infrastructure upgrade for the Brainfish platform. This upgrade not only enhances the capability and compatibility of our system but also fortifies it against potential vulnerabilities. We'll continue investing in and prioritising the backend to ensure optimal performance and security for all users.

More Contextual Answers

Understanding context is key to any conversation and it's no different for AI. We've improved our search engine's ability to process and understand more context, allowing it to generate increasingly accurate and relevant answers. This means fewer misses and more direct hits, resulting in speedy and efficient customer service.

ML Guardrails Implementation

In our ongoing pursuit to refine and enhance our AI capabilities, we've initiated the implementation of guardrails for our machine learning algorithms. These guardrails serve to guide the AI’s learning process to avoid potential pitfalls and improve its understanding and response to customer queries. This brings us one step closer to our goal of delivering an intuitive, reliable, and efficient AI service to our users.

Stay tuned for even more enhancements in our upcoming updates!