Use ai to expand on topics and fill holes
t
trae267
When on a tag or card chat, if you ask a question that is not answered in the content, you have the option to allow the ai to find another source online that includes the info you need.
It could make suggestions for content that would take your understanding of the topic to the next step, or fill a hole into the info that your saved cards dont include.
So now when an issue comes up that you haven't seen before, you can still go Recall to search you for new content and help you expand on your topics. This also gets more people using your app more often. It becomes the default search engine for your heavy Recall users and changes the way they research.
Log In
P
Paul Richards
Merged in a post:
AI behavior
K
Kevin Van
Hello team! It's been really frustrating when the AI mentions that questions I ask that are related, but not directly found in the text, that they're unable to confirm in one form or another. It would be a nice QoL feature to have web search or something similar to cross-check/validate
P
Paul Richards
Merged in a post:
Opt‑in web search for Recall Chat with source‑cited summaries
s
samzheng2003
Hello Recall team—thank you for building such a thoughtful product. “Chatting with my knowledge base” has become a daily habit and meaningfully improves how I capture and retrieve knowledge.
Request: please add an opt‑in way for Recall Chat to access and search the broader web (per message or per thread). This would increase Recall’s utility while keeping me in the same focused workflow.
Why this matters
Up‑to‑date answers: Many questions need current info (news, docs changes, pricing, release notes) that isn’t in my KB.
Fewer context switches: Research directly in Recall reduces copy‑paste friction and preserves conversation context.
Trust and citations: Summaries with citations let me verify claims and save sources as new cards.
Stronger knowledge loop: One‑click to turn vetted results into cards (URL, timestamp, excerpt) keeps the KB fresh.
Competitive edge: Web‑connected assistants are becoming standard; bringing this to Recall’s excellent chat UX would differentiate it.
How it could work (suggestions)
Opt‑in controls: a “Search the web” toggle or slash command, with options like “this message only” or “entire thread.”
Source‑cited responses: Summaries include 3–5 links and quick actions to insert snippets or create cards tagged with the source.
Privacy and safety: Clear network indicator, domain allow/deny list, global disable, and “bring your own search key” (Bing/Brave/Google CSE).
Quality safeguards: Timeouts, rate limits, deduplication, and explicit “no strong sources found” messaging.
Offline/KB‑only mode: Easy fallback when web search is off.
Happy to share use cases, provide feedback on early designs, or join a beta.
Warm regards,
Zilong
P
Paul Richards
Merged in a post:
External AI/LLM support to complement the internal.
R
Richard Dias
Sometimes I do not have the info I want in my library/cards/notesbooks yet, so to be able to query Chat GPT or what ever to complement whatever I do have would be very helpful.
P
Paul Richards
Merged in a post:
A full chat with the web experience
Sankari Nair
Have all the capabilities of a traditional chatbot (ChatGPT / Perplexity) inside Recall, so you can choose to chat with either your own knowledge or the internet.
Problem Statement
Currently, Recall's chat feature is limited to saved content in my knowledge base with enhancements from the internet if information is not found in your knowledge base. While this is valuable for retrieving personal research, it creates gaps when I need:
- Current information (news, updated documentation, recent developments)
- Context that wasn't previously saved to my knowledge base
- Comprehensive answers that require both my notes and fresh web sources
This forces me to context-switch between Recall and web search tools, breaking my workflow and losing conversational continuity.
Proposed Solution
Intelligent chat routing that can dynamically decide when to search my knowledge base, the internet, or both, based on query context and available information.
This should enable me to have just one subscription for my core AI needs.
D
DB
+ 1 , would love this. Surprised more people aren't requesting it.
p
pilotlon
Seems like a good idea. I had discussed this issue with other AI chatbot, that my limited searching might not have found enough material to build an adequate knowledge base of my own within Recall. By being able to expand into the WWW for more answers, I might then easily find new sites/articles/vids to add to my Recall knowledge base. Otherwise, there is the chance that I might get no answer, or a less than adequate answer from chat, simply because my personal knowledge base was not expansive enough.
s
schroell
If I understand correctly, your wish is more about being able to access external information via RAG (Retrieve Information from the internet), because that's independent of which AI is used as the LLM in Recall. Or are there specific reasons why you explicitly want ChatGPT 5 or 4.5 or 4.1 as the LLM? Can you clarify that?
Sankari Nair
Merged in a post:
Chat can reach outside of content
t
trae267
Have the option of a chat ai that, when asked a question that the card doesnt have an answer for, it can search the internet for content on the topic that does answer the question and suggests content that you can save to your Recall.
If you are using Recall for research purposes, the ai could help you fill holes and get clearer information by helping you expand your saved content.
Sankari Nair
We do have this now, but it is not a fully developed chat with the web experience. I will merge this into a new feature request.
D
DB
Sankari Nair any eta on this? Is it being considered or merged with a different update?
D
DB
Sankari Nair any updates on this?
S
Sankari Nair
DB hi, it will be considered as part of the Agentic Rag project, but I'm hesitant to merge it in just now as we're still in the final stages of development. I will come back to you.
D
DB
Sankari Nair awesome, thanks for the update I appreciate it. I can honestly say that this is one of the few real blockers that prevents me from fully committing to your app. Keep up the great work!