# Private, local AI

- Page: https://scanovich.ai/research/private-local-ai/
- Language: en
- Updated: 2026-11-11

- Status: active

**Question.** How much speech, search and assistant work can run inside the owner's perimeter, on hardware they control, without giving up quality?

## Why it is uncertain

Some documents and recordings cannot leave the building. Local models are smaller than hosted ones, and the honest answer to "is local good enough" depends on the task, the language and the hardware. It has to be measured case by case.

## Method

- Build the local version as a real tool that people use daily, then measure it.
- Keep every service behind a standard interface, so a local model and a hosted one can be swapped and compared on the same inputs.
- Publish the code and the benchmark harness.

## What we measure

- Quality against a hosted baseline on the same inputs.
- Latency on ordinary hardware.
- What still has to leave the perimeter, if anything.

## Results so far

- Voice typing that runs entirely on a Mac, released as open source, with a local server that speaks the standard audio interface.
- Embedding and reranking services that serve several of our projects from one machine.
- A call-analytics reference stack that keeps audio and transcripts on the owner's disks.
- A private chat workspace that routes to models the owner chooses.

## Next

A one-command installation of the search stack on a client's Linux machine.

## Projects on this line

- [Services-BGE](https://scanovich.ai/projects/services-bge.md) — released, open source · MIT. Local embedding and reranking services for hybrid search, and a shared, measured memory for AI agents.
- [VoiceToText](https://scanovich.ai/projects/voicetotext.md) — released, open source · MIT. Private voice typing for Apple Silicon Macs. Hold a key, speak, let go, and the text appears at the cursor.
- [Call analytics](https://scanovich.ai/projects/call-analytics.md) — prototype, public code. Call recordings become transcripts and structured quality reviews on infrastructure the owner controls.
- [Private AI workspace](https://scanovich.ai/projects/private-ai-workspace.md) — prototype, public code. A chat workspace that routes to models the owner chooses and can ground answers in company documents.

## Evidence

- **92%** — of questions get the right passage ranked first by search with reranking. Plain text search manages 38%. Measured 2026-11-11. Source: https://github.com/FUYOH666/Services-BGE/blob/main/docs/MEMORY-BENCH.md
- **6 of 6** — times a planted false note wins without trust tiers. With reviewed files preferred, it never does. Measured 2026-11-11. Source: https://github.com/FUYOH666/Services-BGE/blob/main/docs/MEMORY-BENCH.md
- **0.861** — recall at rank one with the better of two rerankers on the Russian demo set. The other reaches 0.750. Measured 2026-11-11. Source: https://github.com/FUYOH666/Services-BGE/blob/main/eval/results/COMPARISON.md
- **v2.2.0 · MIT** — Public release that installs from source with one command. Measured 2026-11-11. Source: https://github.com/FUYOH666/VoiceToText
- **Reference stack** — A working pipeline behind a web interface, an HTTP interface and a command line. Measured 2026-07-17. Source: https://github.com/FUYOH666/Scanovich.ai-audio-call
- **Reference stack** — A working prototype with its architecture and feature documents in the repository. Measured 2026-07-17. Source: https://github.com/FUYOH666/scanovich-webUI