Internal Portal

Sign in with your Verne Power Google account.

Loading authentication…
Verne Power — Internal Tools

Good morning, team.

Access AI tools, document processing, and project resources — all in one place.

API checking…
documents indexed
RAG checking…

Ask the Bot

CPU —
RAM —
V
Hello! I'm the Verne RAG assistant. Ask me anything about your indexed projects, documents, or Slack threads.
All folders

Document Converter

Drop your file here

or click to browse

.docx .xlsx .pptx .pdf
Word Document
Georgia font · Navy headers
Excel Spreadsheet
Verne table styles
PowerPoint
Verne slide template
PDF Export
Final distribution
Processing…

Index Status

Documents indexed
Total chunks
Avg chunks / doc
TITLE / PATHSOURCECATEGORYMODELSIZETOKENSTIMEINDEXED AT

Connector Settings

All connectors are read-only — they only list and download files. No writing, no deleting.
To get a refresh token: run python3 /opt/verne-rag/scripts/get_dropbox_token.py on the server, or download the script and run it locally.
When creating the Dropbox app, enable only: files.metadata.read and files.content.read.
Create an integration at notion.so/my-integrations, copy the token, then share each page/database with the integration using the Share menu in Notion.
Use Haiku via the claude CLI (jules@verne-power.com subscription — no API key). On the last day before token reset, after 7 PM, queries auto-switch to Haiku to maximise remaining tokens.
Token usage this cycle Loading…
Set the day your Claude.ai subscription tokens reset. The auto-switch triggers on the day before, after 19:00.
This model reads each document and decides its category, chunk strategy, authority rank, and extracts metadata. It runs once per file at ingest time.
Model recommendations (best → worst)
# MODEL SIZE BADGE NOTES
Loading…

Document Search

Vector Map

Computing UMAP projection…