CoreChat
A RAG-based internal knowledge OS. It indexes organizational and personal documents along your permission boundaries and answers with chunk-level sources.
It reads PDF, Word, Excel, HWP, and images with OCR — preserving merged cells and complex layouts — then splits and embeds them into chunks. Answers carry document–page–paragraph sources with a confidence score, and organizational and personal documents stay fully separated under role-based access control. Bulk uploads index asynchronously without downtime, and specialist agents can be called from the same screen to carry the work forward.
Only org documents within your permissions are searched
Start by picking documents
Choose the document context in the left sidebar
Type your message.
Agent store
Beyond document answers, these specialist agents are available too
Turns a recording into complete minutes — summary, transcript, and action items.
Writes a report from any file, document, or idea — outline through body copy.
Point it at a template and it pulls just the fields you need, even from handwriting.
Classifies the optimal HS code from a product name and computes the lowest duty.
2026_employee_handbook.pdf
File info
DownloadMeta keywords
File description
The 2026 revision of the employee handbook. Covers working hours, PTO and remote-work rules, family-event leave, travel and expense procedures, and the performance review cycle.
On upload it is split into 11 chunks and embedded, with meta keywords and a description generated automatically. Answers find their evidence in this index.
Key features
Permission-aware document context
Pick the answer scope by folder across personal and organizational documents. Only documents within your permissions are searched, so existing access boundaries are preserved.
RAG indexing on upload
Upload a document and it is split into chunks and embedded immediately, with meta keywords and a description generated automatically — searchable and answerable the same day, with no preprocessing work.
Grounded citations
Every claim carries a footnote such as [1]; expanding the references shows the cited source chunk with its relevance score, so you can verify exactly which document an answer came from.
Specialist AI agents
Beyond document answers, subscribe to specialist agents — meeting notes, reports, document extraction, customs — and call them from the same screen.
Workflow integration
Connect approvals, scheduling, and task management so work is completed inside the chat, then share the output with your team through the artifact library.
Enterprise security
Built for sensitive internal documents — on-premise deployment is supported, with an API for integrating external systems.
See how CoreChat fits into your workflow.
Tell us how the work runs today, and we'll map out where and how it would apply.