Blog
Posts on local AI, privacy, GDPR and the EU AI Act, architecture and retrieval — grounded, with sources.
RSS feed-
Local AI with 4 GB VRAM: model fit and document workflows
Understand GPU memory, model loading and indexing on a small GPU. Check LokLM settings and evaluate speed and source quality with your own documents.
Read more → -
Ask AI about a PDF: a practical source-checking workflow
Check PDF text, ask focused questions, verify citations and keep conflicting versions separate. A practical workflow with a reusable evidence note.
Read more → -
Understanding citations: evidence and privacy
What a source reference lets you check, why it does not prove an answer correct, and which separate questions establish where your document data goes.
Read more → -
GDPR and cloud LLMs: roles, data flows and transfers
What to check before sending documents to a cloud LLM: lawful basis, processor contracts, international transfers and the limits of local processing.
Read more → -
On-device AI under the EU AI Act
How the EU AI Act applies to local AI: provider and deployer roles, use-case risk, transparency, AI literacy and the limits of open-source exceptions.
Read more → -
A taxonomy of "local AI": inference, retrieval, training
What can be "local" about an AI — three pipeline stages, three times the question "where does this actually run?". A reference piece for the rest of the series.
Read more → -
What "private" actually means for an AI assistant
Five practical checks for AI data handling: processing location, storage, telemetry, auditability and sync. With LokLM provider and storage caveats.
Read more →