What is hybrid RAG for email?
Hybrid RAG combines vector search over indexed mail and documents, graph expansion across people, orgs, and topics, and taxonomy-aware query planning. Teams bind document sets to an Outlook category, Gmail label, or folder so HEIDI grounds classification and drafts for that mail type in the right knowledge, with sources shown.
Capabilities
What you can do with HEIDI Workspace
Per-target document binding
Support@ gets the support manual; sales@ gets the price list — not one muddy pool.
Vector + graph + taxonomy
Finds the right paragraph even when the customer uses different words than the manual.
Sources visible on drafts
First-line agents approve faster; compliance can audit what was used.
Retrieval guidance per target
Limit a category to specific documents — business rules without code.
Use cases
Where teams apply HEIDI Workspace
Real workflows that benefit from visual design, automation, and governance.
Technical Support category → Product X manual + escalation SOP
Drafts for that category retrieve only the bound technical knowledge.
Invoices folder → AP policy + vendor master guidance
Billing questions are grounded on finance policy, not product docs.
Customer inquiries answered from approved product docs
No hallucination — answers come from the bound sources with citations.
Policy exception routed with the right policy cited
HEIDI surfaces the governing policy document on the draft.
RFP folder → extract requirements into proposal-ready output
Mail plus attachments are mined for requirements with sources.
How it works
From chaos to clarity in 4 steps
Connect or upload documents
Manuals, SOPs, FAQs, price books, contracts, onboarding packs.
Bind documents to a mail target
Assign sets to an Outlook category, Gmail label, or folder.
Retrieve with hybrid RAG
Vector, graph, taxonomy, attachment search, and thread context.
Compose with visible sources
Drafts show which manual, SOP, or document was used.
Why hybrid retrieval beats generic RAG
Generic RAG misses domain-specific answers because it searches one big pool with one method. Hybrid retrieval combines semantic vector search with graph expansion across relationships and taxonomy-aware planning, scoped to the documents bound to a mail target. The result: L1 support and customer mail answered from the right manual, with citations a reviewer can verify before approving.