Memraiq

Built for grounded team workflows

Hosted knowledge workspace in private beta: text documents today, graph-augmented answers with citations, and a clear roadmap for richer files and developer tools.

01 / Chat

Grounded AI chat

Every answer is tied to your indexed documents. The model can't make things up — if it can't find supporting context, it says so. Each response includes the specific sources it drew from, so anyone on your team can verify the answer.

Source citations with every response
Context window built from your documents
Explicit fallback when context is insufficient
Chat history scoped to your workspace

Chat trace

You

What does our onboarding doc say about access provisioning?

Memraiq

According to your onboarding guide, access is provisioned within 24 hours of account creation. New members receive an email invite with a temporary password...

↳ onboarding-guide.md · §3.2 (0.94)

↳ access-policy.md · §1 (0.81)

Pipeline run

scan12 documents found

chunk847 chunks created

embed100% coverage · 847/847

graph1,204 entities extracted

indexVectors stored · Qdrant

run complete · 43s · 12 text sources

02 / Indexing

Document indexing pipeline

PDF / DOCX / image · coming soon

Currently supports Markdown and plain text uploads. A structured pipeline chunks, embeds, and extracts entities for the knowledge graph. Background workers keep ingestion from blocking chat. Re-index when documents change.

Configurable chunk strategies for text sources
Background workers — ingestion stays non-blocking
Visibility into ingestion progress for uploads
Re-index on document update

03 / Retrieval

Hybrid retrieval paths

Vector similarity finds chunks that look like your question. Graph traversal finds related entities and their connections — people, projects, dates, concepts. Both paths run on every query and their results are merged for richer context.

Qdrant vector storage with semantic search
LightRAG knowledge graph traversal
Entity relationship mapping
Better answers for multi-hop questions

Retrieval paths

vector path

query → embeddings → top-k chunks

4 chunks · avg relevance 0.87

graph path

entities extracted → graph traversal

related nodes → additional context

+3 nodes · 2 hop depth

merged → 7 context chunks → answer

Org structure

organisation acme-corp

owner alice@acme.com

admin bob@acme.com

member carol@acme.com

member dave@acme.com

12 documents · 847 chunks indexed

last activity · runbook.md uploaded · 2h ago

04 / Workspace

Workspace operations

Organisations, members, and roles. Each workspace is isolated — documents, chat history, and settings belong to the organisation, not individual users. Invite members, manage access, and control who can upload or delete documents.

Owner, admin, and member roles
Invite members by email
Document-level access control
Organisation-scoped billing and limits

05 / Trust

Answer traceability & access

Every answer shows which passages supported it so teams can verify claims. Workspace roles and invites keep knowledge inside the right group; document uploads are scoped to your organisation.

Citations and source snippets with each answer
Owner, admin, and member roles at workspace level
Invite flow and membership boundaries per organisation
Activity you can reason about — who uploaded what and when

Answer trace

question → SLA for enterprise support?

↳ support-policy.md · §2 (cited in answer)

↳ customer-handbook.md · §4.1 (cited in answer)

recent uploads

alice@acme.com · playbook.md · indexed

bob@acme.com · faq.txt · embedding…

Workspace snapshot

documents12 indexed

chunks in index847

ingestion queue

policy-update.md · chunking

onboarding.md · complete

plan limits apply · see Pricing for allowances

06 / Activity

Documents, progress, and plans

See how many sources are in your workspace, follow ingestion as new text lands, and understand how plan limits apply. Grounded answers stay tied to what you've actually indexed.

Document counts and indexing status at a glance
Ingestion progress as files move through the pipeline
Plan-based limits (tokens, seats) as described on Pricing
Answers remain traceable to uploaded sources

Roadmap

Upcoming capabilities

Rich document types

In development

PDF, DOCX, and image ingestion with OCR so slides, contracts, and scans join your knowledge base without manual paste.

Embeddable widgets

In development

Drop search and chat into internal portals and tools so teams stay inside the systems they already use.

API keys & developer docs

In development

Automate uploads, queries, and integrations with documented APIs and scoped credentials for builders.

Ready to see these features in action?

Request access Compare plans