Add the ability to create detailed lengthy reports based on a prompt across your knowledge base
s
schroell
Feature Request: Add a comprehensive report generation feature to Recall that creates structured, long-form documents from filtered knowledge base subsets.
The current AI chat functionality limits output to approximately 7000 characters including references, yielding only 500 words of actual content, which is insufficient for professional use cases. The requested feature should enable users to generate markdown-formatted reports of configurable length specified by page count or word target, similar to NotebookLM's report function.
Users should be able to filter source material through tags or prompts, with the system proposing document structure and synthesizing information across multiple stored resources. The generated reports should serve as fact-based foundations for slide decks, customer presentations, or essays, maintaining proper source citations while delivering substantially more content than current chat outputs. Implementation should appear as a dedicated option in the sidebar alongside existing Chat, Graph, Review, and Settings functions, providing direct access to this document generation capability without relying on chat interface limitations.
Log In
s
schroell
Another idea just came to me in this context, looking at my own use cases and how I'm literally abusing Recall to achieve what I actually want to do with it. Also keeping in mind the ideas floating around right now about having a notebook that can be based on multiple input documents.
What I realized is that I very often use Recall to collect information in a structured way for client-facing stuff — like a training session or a detailed report on a topic. And before I start writing, I always gather facts, information, and foundational material first.
Here's how I normally go about it:
- I either start an extensive prompt in the AI chat against my entire curated Recall database to collect the information
- If I know it only concerns a specific area or I want to narrow it down, I restrict it further using a tag
- The output I get from a good LLM — which is now pretty high quality — I then use to start a new card
- So the result from the AI chat gets assigned to a specific card, and that card becomes my collection card for all the information I need on that topic
Then I usually read through everything. That's when I notice that a certain aspect isn't covered well enough, or something is missing that I forgot. So I go into the chat of that card and add additional aspects to it — refining, expanding, and supplementing the topic collection for my overall report.
And if I've also restricted those additions to very specific other card sets via tags etc., I add those to the actual notebook of the card.
This is basically where my Recall use case ends
— because from all that information, the only way to turn it into a consistent, structured report inside Recall is manually and very tediously.s
schroell
To add to this idea for this feature request: it would obviously be cool if there was an additional function in the notebooks. Not just
Concise Summary
or Detail Summary
, but a "Structured Report"
function — where you can optionally provide a prompt if you have specific requirements, like:- the order of topics
- how the topics relate to each other
- the focus
- the length
Basically anything you'd want to feed into the report — and then a clean report gets generated based on the contents of that card. That would obviously be great.
And if you could also say "this can be enriched externally via Rack and the internet to fill gaps or add precision" — hey, that would be a totally awesome extension.
Think about whether that wouldn't be a solid approach to turn Recall into a proper research tool.