Fresh from the Build: Storytell's Latest on Chrome Extension, Data Guidance, and V3 Vision

In our recent engineering demo, May 30th, our engineering team walked us through enhancements to our Chrome Extension, clearer guidance when you're working with specific data scopes, and an exciting glimpse into the architecture of Storytell V3. 

Streamlining your workflow: new copy options in the Chrome Extension

Andi, our senior frontend engineer, showcased two new menu options in the Storytell Chrome Extension’s text pop-up, designed to make capturing and sharing information quicker and more efficient. We heard from users that our pop-up could sometimes interfere with the native browser menu, especially when simply trying to copy text. These new options aim to provide the functionality you need directly within our menu, saving you a step and enhancing your workflow.

The new options are:

• Copy text: This option allows you to copy the selected text directly. When you paste this text into an application that supports rich text (like Google Docs or Notion), it will retain its original formatting, such as lists, bolding, or embedded links. If you paste it into a plain text editor, you'll get just the clean text.
• Copy text with link: With this feature, you copy the selected text, and when you paste it into a rich text-supporting application, the copied text will automatically include the source page where you copied it from. This is incredibly handy for quickly referencing your sources. Again, pasting into a plain text environment will just provide the text itself.

These additions mean fewer clicks and less friction when you're gathering information using the Chrome Extension. We want to make it easy for you to grab what you need and integrate it seamlessly into your work, whether you're compiling notes, drafting documents, or sharing insights with your team.

Andi also shared a glimpse of a future enhancement: the ability to generate a summary of selected text directly via an LLM when you copy. Imagine selecting a long passage and, with one click, copying a concise summary to share—this could save several steps and is something we're looking to prioritize based on positive initial feedback.

Clearer error messages when data isn't found

Shelby, our principal software engineer, then demonstrated an important improvement in how Storytell communicates with you when there's no data available for your query's scope. We know it can be confusing if you ask a question expecting an answer from your specific documents, but the system doesn't find any relevant data to use. This can happen for a few reasons, such as assets still processing or being in a different Collection than anticipated.

Previously, this might have led to an unexpected or generic response from the LLM. Now, we’ll introduce a persistent interstitial message that will appear, clearly stating that no matching data could be found and offering potential reasons and solutions. This message prevents your query from unnecessarily going to the LLM, saving time and avoiding confusion.

This update is all about providing transparency and empowering you to resolve issues quickly. Instead of wondering why a response seems off, you'll get immediate, clear feedback and actionable links to check your Collection, asset status, or query scope. We're also working on making these links even more dynamic, potentially taking you directly to the relevant part of the app or opening a feedback modal.

A sneak peek at Storytell V3 architecture

DROdio, our CEO, provided a look at the upcoming architecture of Storytell V3, highlighting some fundamental changes designed to make working with your data more flexible and powerful.

In V2, our current version, your workflow typically involves uploading an asset, which gets split into chunks and manually placed into Collections. You then chat with these Collections, sometimes pulling in additional specific information like PDFs or other collections.

V3 introduces several key shifts:

• Assets as attachments for quick chatting: When you simply want to chat with a document, assets will be treated as attachments by default. This bypasses the full processing system, making interactions much faster, similar to how web pages from the Chrome Extension are handled in V2. You'll have to explicitly choose to permanently save attachments to your knowledge base for more extensive processing.

• Enhanced knowledge base processing: When assets are added to your knowledge base in V3, there's more extensive processing beyond just chunking. This includes the creation of "seed concepts" and a more sophisticated concept-based view of your data.
• Introducing data views: Instead of collections, V3 will feature "views" of your data. Think of a view as a flexible bucket or basket of information that isn't restricted by the old Collection hierarchy. You'll be able to create views based on criteria like:
  
  • All your knowledge
  • Assets you've manually labeled
  • Assets uploaded within a specific timeframe (e.g., the last 3 days)
  • Content related to a particular concept, drawing pieces from various assets.

You will then be able to chat directly with these tailored views. The aim is to simplify data interaction and provide more intuitive ways to talk to specific subsets of your information, whether it's a single document, a set of documents, or data connected by a concept across your entire knowledge base. V3 will also allow you to create actions around these data views.

V3 is being built to offer a more dynamic and less rigid way to organize and interact with your knowledge. The shift from collections to flexible views, combined with faster initial chatting and richer background processing, aims to give you greater control and more powerful ways to access the specific insights you need, exactly when you need them. This architecture lays the groundwork for future features that can proactively bring important information to your attention.

These updates from our latest engineering demo represent our ongoing commitment to refining the Storytell experience. From small but impactful workflow improvements in the Chrome Extension and clearer communication about your data to the foundational work being done on V3 to transform how you organize and access knowledge, we are focused on building a platform that empowers you. We are excited about these steps and look forward to sharing more progress in the future.

Become an Alpha or Beta Tester

Get early access to features as we release them by becoming an alpha or beta tester. Here's how to sign up: https://web.storytell.ai/early-access