The useful detail is in the repository
The product description may be in Confluence, while the current behavior or exception is already documented in a README or another Markdown file.
Answers from repository documentation
Ask the question. Open the source behind the answer.
The useful detail may be in a repository while the page in Confluence is already out of date. Krynia indexes the material you add, answers questions about it and links back to the source.
Krynia workspace preview
Ask Krynia
What is Andocs and which sources explain the product?
Andocs manages templates, approvals and signing.
Krynia cites the product docs first.
Access checks before answersConnected company memory
Knowledge ready
88k
Stored originals
0.1 GB
Public product page
Krynia Product · 12 min ago
Public features page
Krynia Product · 31 min ago
Public security page
Krynia Trust · 7 min ago
Public use cases page
Krynia Product · 18 min ago
One public GitHub repository Markdown, MDX, text and CSV Answers with source links No charge per teammate Runs on Cloudflare infrastructure
The problem
The answer exists. Finding it is the slow part.
A wiki search can return a long page without the detail you need. The current explanation may be in a repository, while the page everyone knows has not been updated.
Documentation falls behind the work
The familiar page stays the first place people search even after the useful explanation has been updated somewhere else.
Search returns a page, not the answer
Even the right long document still has to be read. What you need is a short result and a source link you can check.
Chat-first company memory
Answers stay tied to source context.
Krynia keeps chat simple for the team while admins can review the material used, the relationship map, the connected sources and monthly knowledge usage behind every answer.
Source summaries keep answers concise without exposing raw internal documents.
Admins can review sources and graph links before trusting or correcting an answer.
Pricing stays tied to retained indexed knowledge, not seats.
Chat-first company memory
Cited answer in context
Short answer, sources ready to review.
Product
One question over the material you added.
Krynia currently works with one public GitHub repository and bounded Markdown, text or CSV uploads. It returns an answer with citations instead of asking you to scan every matching file.
Ask instead of scanning every file
Write the question in your own words. Krynia searches the indexed material and returns a short answer.
Open the exact GitHub version
Repository citations point to the file at the commit Krynia indexed, so a later edit does not change the evidence behind an old answer.
Upload a small set of documents
Add bounded Markdown, text or simple CSV files when the useful material is not in the connected repository.
Keep the previous index during refresh
A new indexing run is prepared separately. The last successful version stays searchable until the replacement is ready.
Who it fits
For teams with knowledge spread everywhere.
Small teams that keep product or project documentation in a public GitHub repository.
People who know the answer is in a long Markdown document but do not know the page or wording.
Teams that want to verify each answer before adding more sources and colleagues.
Use cases
Chat is the main UI, but every answer stays tied to sources.
Product, support, engineering and onboarding teams can ask normal questions and still get answers they can verify.
Product and support
Find product context
"What is this product and where is the source?"
Short answer with citations to docs, repositories and decisions.
Engineering
Trace dependencies
"What depends on this service, package or feature?"
Evidence from repositories, tickets, docs and graph links.
People and team leads
Onboard new people
"What should a new teammate read first?"
A scoped reading path based on real company sources.
Founders and managers
Recover decisions
"When did we decide this and why?"
Timeline-style answer grounded in PRs, issues, docs and chats.
Workspace admins
Admin visibility
"What are people searching for and what is missing?"
Topic groups, failed searches, source gaps and traceable citations.
Everyone
Meeting prep
"What context do I need before this discussion?"
KB-first briefing now; calendar-aware preparation can follow later.
Workflow
Add the material, let it index, then ask.
The first setup is deliberately narrow. You can see when a source is queued, indexing, ready or needs attention before anyone relies on its answers.
01
Add the first source
Connect one public GitHub repository, or upload Markdown, text or CSV documents from the workspace.
02
Wait for indexing
Krynia reads the supported files and prepares them for search. A larger source can take a few minutes.
03
Ask the question
Search in your own words instead of guessing the title or exact phrase used in the document.
04
Open the citation and invite colleagues
Check the source behind the answer. When the workspace is useful, invite colleagues by their verified email address.
Integrations
What you can add today
Connect one public GitHub repository, or upload bounded Markdown, text and CSV files. GitLab, Google Drive, Slack, Confluence and private repositories are planned, not available today.
Available connector
1
Indexed data in this preview
0
GitHub repositories
manual initial indexing
Public repository Markdown and text documents for the first production connector.
Scope: one public repository
GitLab projects
not available
Planned connector for repository documents and project context.
Scope: planned
Google Drive
not available
Planned connector for selected documents and shared drives.
Scope: planned
Slack channels
not available
Planned connector for selected channel context.
Scope: planned
Admin controls
See what is indexed and what needs attention.
The source list shows the current indexing state, the last successful version and failures that need action. Removing a source also removes its indexed material.
Krynia's managed AI route is used only to prepare answers. Customer prompts, cited source context, embeddings and answers are not used to train AI models.
Confluence search works when the right page exists, is current and uses the words you searched for.
Krynia is useful when the answer is inside the repository documentation you added and you want the supporting source, not another file list.
Workspace health
Access, answer controls and knowledge usage in one place.
Passwordless access
Google, Microsoft, passkeys and domain-based workspace join.
Workspace membership
Verified workspace members can search the sources available in that workspace.
Managed AI route
Default answer path, training policy and future external routes stay admin-visible.
Clear pricing
10-page policy: included in EUR 47 base.
Pricing
Pricing people can understand before they connect a single source.
Krynia bills the knowledge still kept at billing close. Regular sync, storage, citations and normal chat are already part of the base product.
Meter
€0.47
per 1M indexed tokens
Minimum
€47.00
product plus 5M indexed tokens
Included
5M
about 7.7k A4 pages
Why the minimum is EUR 47
The monthly base includes product access, sync, citations and the first 5M indexed tokens. Overage starts only after that included volume.
No seat-based tax for quiet users.
Deleted or stale chunks removed before billing close do not count.