mirror of
https://github.com/BookStackApp/BookStack.git
synced 2026-02-05 16:49:47 +03:00
Feature Request: Native Semantic Search + Optional Natural-Language Q&A Layer (“Talk to Your Docs”) #5512
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @ivano-buffa on GitHub (Nov 20, 2025).
Describe the feature you'd like
Summary
I would like to request consideration for a built-in, fully optional feature that enables BookStack users to perform semantic search and - where appropriate - natural language Q&A across their documentation.
This would allow users to “ask a question in plain English” and retrieve the most relevant answers or page references from across their BookStack instance. In other words: an optional layer that allows users to “talk to their documentation”.
This is not a request for a chatbot UI or any heavy AI-first redesign - just a request to explore modern search and retrieval enhancements that remain aligned with BookStack’s guiding principles.
Why this feature is important
1. Documentation repositories are growing in size and complexity
Many organizations now use BookStack as a full internal knowledge base.
As content grows into hundreds or thousands of pages, traditional keyword search — while excellent — becomes insufficient for:
Semantic search and natural-language Q&A help bridge this gap in a way that keyword search cannot.
2. The software landscape around us is rapidly adopting semantic search
Most modern documentation platforms — from GitHub to Atlassian Confluence competitors — are exploring or already offering:
This is quickly becoming a baseline expectation for internal tooling.
Adding this feature would help BookStack remain competitive as an open-source alternative to corporate knowledge platforms.
3. Use cases and productivity gains for teams
Users increasingly ask questions such as:
With semantic search / natural-language Q&A, the system can point them to:
This saves time across:
This aligns beautifully with BookStack’s mission: help people find and organize knowledge easily.
4. The BookStack structure is uniquely well-suited for this
Unlike unstructured wikis, BookStack has:
This structured data makes BookStack an excellent candidate for:
The clean hierarchy makes implementation more reliable than in many competitors.
5. A privacy-respecting, optional design is absolutely feasible
This feature does not need to compromise BookStack’s simplicity.
A realistic approach could be:
Implementation Option A – Built-in semantic search layer (no external AI calls)
Implementation Option B – Pluggable Integration Layer
Provide a standardized extension point allowing community plugins to add:
This avoids burdening the core while enabling a community-driven ecosystem.
This keeps BookStack lightweight — a value many users and sysadmins care deeply about.
There are multiple scattered issues, comments, and external blog posts from BookStack users asking how to achieve:
The interest is real, but fragmented.
Providing a formal roadmap acknowledgment (even long-term) would unify the discussion.
Proposed Scope (Realistic, minimal, and aligned with BookStack philosophy)
Here is a pragmatic version that fits BookStack’s ethos:
Why This Fits the BookStack Roadmap
BookStack has:
Semantic search is a natural evolution of its existing search system — not a departure from its identity.
It enhances discoverability without complicating editing or navigation.
This improvement would multiply BookStack’s adoption across:
Closing Thoughts
This feature request is meant as a long-term, optional enhancement — not a disruption to BookStack’s clean foundation.
The goal is to:
If implemented, even partially (semantic search alone), it would significantly upgrade BookStack’s value for thousands of users.
Describe the benefits this would bring to existing BookStack users
Implementing optional semantic search and natural-language Q&A would bring substantial benefits to BookStack’s existing user base:
Users can find relevant information even when they do not know:
This is increasingly important as BookStack libraries grow into hundreds or thousands of documents.
Semantic search helps:
Users can ask questions in everyday language and immediately receive relevant pages or excerpts.
This reduces time spent searching, navigating, and re-reading.
Many companies adopt BookStack as their “single source of truth.”
Semantic search strengthens this role by:
Tools like Confluence, Notion, and GitBook are marketing semantic search heavily.
Offering optional semantic capabilities helps BookStack remain a strong open-source competitor.
New team members often don’t know:
Natural-language querying solves this elegantly.
There is already community interest for:
A native or pluggable solution would address an active user demand.
Can the goal of this request already be achieved via other means?
Only partially, and with significant limitations.
What can be done today:
But the limitations are significant:
Conclusion: The goal cannot be fully or securely achieved with current capabilities.
A native optional feature or official plugin interface would be far superior.
Have you searched for an existing open/closed issue?
How long have you been using BookStack?
Not using yet, just scoping
Additional context
This request is made with full respect for BookStack’s philosophy:
The suggestion is intentionally modular and optional so that:
No complexity would be imposed on installations that do not opt in.
I am confident that even a minimal implementation (Phase 1 semantic search) would provide tremendous value to both new and long-time BookStack users.
Thanks for considering it - and sincere appreciation for the outstanding work on BookStack so far.
~Ivano
@virtadpt commented on GitHub (Nov 21, 2025):
Having semantic and vector search would be really handy.
@ssddanbrown commented on GitHub (Nov 21, 2025):
Thanks for the request @trendpx but I'd probably consider this to have the same fundamental desire as existing issue #5611.
I did start on something along these lines earlier this year which can be seen in https://github.com/BookStackApp/BookStack/pull/5552, but as stated there I havn't been too keen on the vector results at scale so far in a generic manner.
I know it's possible to get good results, but those systems often work best with scenario/context/language specific tweaking, which is a bit tricker to do when supplying a more generic solution like BookStack which may be used in many various scenarios.
I need to revisit that PR though and play with some other options.
@ssddanbrown commented on GitHub (Dec 24, 2025):
I'll go ahead and close this off as per the above, since #5611 is already open.