During this week’s engineering demo, our engineering team shared updates on several core areas: the newly released dynamic views for data organization, improvements in concept discovery and scoring, and a unified agentic chat experience.

Dynamic views transform how you organize and access your data

Andi Rosca demonstrated the new Data Views feature which act as dynamic buckets that automatically collect and filter your data by labels. Users can now create personalized workspaces for any subset of information (like project, topic, client, or data type), making it much easier to focus research, or review.

• Create as-you-go: Whenever you add or label new content, corresponding dynamic views update in real time, always showing the latest relevant materials.
• Slice your knowledge: For example, users can set up views for each client, all Zoom recordings, or ongoing research topics, with every view always up to date and ready for deeper exploration.
• Concept-first or record-first perspectives: Each view surfaces not only the knowledge assets associated with those labels but also the concepts they contain, powering both broad overview and detailed inquiry.

Because all views are dynamic for now, there’s no manual curation. Whenever you label new documents, articles, or meeting notes, the appropriate view expands, so you’re always working with the most current collection of insights. In the future, Storytell will build on this core capability to bring automated actions—such as automatic summarization or notifications—triggered whenever new information appears in a view.

Concept clustering and scoring improvements

Accompanying dynamic views, Storytell introduces smarter concept clustering and scoring. Previously, all extracted concepts appeared as one interconnected web, even if they were unrelated. Now, Storytell separates concept clusters based on semantic distance, so unrelated ideas form distinct groups—making it easier to navigate and analyze diverse datasets. For instance, technical documentation and market strategy notes will display as their own clusters, rather than being artificially linked.

Storytell also now scores each concept on occurrence and contextual relevance (how many times a concept is cited and how central it is). Users can filter key concepts using a slider control, quickly surfacing the most important themes in either a specific view or the full knowledge base. This helps pinpoint critical insights without sifting through unrelated information.

Improvements to the agentic workflow

Alex shared progress on his goal of creating a single, unified workflow that merges “chat” and “agentic” modes. Historically, users toggled between chat and agentic flows, with each relying on a different backend. This duality could create limitations. The latest update replaces Storytell’s older “temporal tools” with a streamlined “LLM tool house” approach: now every intelligent workflow, including AI-assisted actions like web search, happens in a seamlessly consolidated mode.

For users, this means clearer interactions—there’s no longer a choice between chat or agentic mode. Instead, all intelligent actions, such as generating insights or triggering web searches for recent information, are managed in one consistent place. Although the interface remains familiar, the underlying improvements make it easier to introduce new AI abilities, like sending emails or integrating further automated actions, all of which will become available as new tool calls in upcoming releases.

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://docs.storytell.ai/about/early-access