Product Updates | September 2023

At Context.ai, we’ve been working hard to build high-impact features to help you even better understand how users are interacting with your AI products. To keep track of it all, here is the first set of product updates in what will be a regular series.

In this post, we’ll highlight several significant improvements, including the new Workspace and Users pages, expanded analytics capabilities, and PII filtering for protecting sensitive data.

Workspace

We listened to your feedback and are excited to launch the Workspace to give you more control over data visualizations. Create custom graphs, breakdown your data to compare segments, and filter according to topics or other metadata. This is just v1, so there’s much more to come!

Users page

The Users page surfaces insights about who your users are and breaks them down into cohorts—new users, returning users, and veterans. You can track DAUs, frequently discussed topics, and even the most engaged users. Act on this data to attract new users and retain existing ones. To get set up, make sure to enable user engagement monitoring.

Transcript backfilling

We now support transcript backfilling so you can analyze your entire corpus of conversations from when you first start using Context.ai. Log transcripts as you normally would, and just set the event_timestamp field within at least one message to accurately record when a conversation happened:

New analytics features

We’re always investing in our analytics fundamentals. Here are three things we added recently.

Cost estimation 💰

LLMs can be expensive, and nobody likes a surprise bill. We implemented conversation cost estimation so you can measure your spend and compare different models. We support models by OpenAI, Anthropic, and others. Check out our integration instructions for more information.

Foreign language support

LLMs let you interact with users in nearly any natural language, so it’s key that your analytics reflect that. Context.ai now automatically detects messages in over 100 foreign languages and translates them into English.

Go to the Overview page to see the most common languages your users are conversing with.

Global filters

To speed up your analyses, we added global filters that persist between pages. You can filter by any metadata you’ve attached to transcripts, whether that’s model type, user ID, or something else. As you navigate the app, all transcripts and visualizations you see will be limited to that filter.

PII filtering

By default, Context.ai now detects and redacts personally identifiable information at the time of transcript ingestion. Sensitive user details like credit card numbers and email addresses do not get stored and are replaced with dummy strings instead.

Stay tuned for more

We’re already hard at work on the next set of product updates including improved reporting functionalities and as always, even better analytics. To stay in the loop, scroll down and subscribe to our newsletter using the form below.

Have a feature request for the team? Let us know! Feedback is welcome at context.ai/contact.