ALPHA · WINDOWS DESKTOP APP

Private AI that answers from your documents.

Logmark runs small open models on your own computer, builds assistants over your files, and shows the passage behind every answer. When the answer isn't in your documents, it tells you so instead of guessing.

Runs on your machine No vector database Cited answers Says "not found"
8/8questions had the right passage in the top 5 results
4/4off-topic questions correctly refused
~17sto index a 150,000-word PDF
<1sper question, including reranking

Measured on a laptop with a 150,000-word PDF of India's Labour Codes and 12 test questions. A small test, shown so you can see how we measure.

How it works

From a folder of files to an answer you can check.

Keep scrolling. The shape on screen follows your documents through each stage.

01 · ADD

Add your documents

Drop in PDFs, Word files, spreadsheets, CSVs or plain text. Logmark extracts the text page by page and sheet by sheet, and everything stays on your machine.

  • PDF pages, Word headings and tables, every Excel sheet
  • Identical files are skipped; changed files replace the old version
  • Only extracted text and vectors are kept, not copies of the files
02 · SPLIT

Split by meaning, not by character count

Semantic chunking cuts on whole sentences into passages of about 150 to 225 words, with a little overlap so nothing falls between the cracks.

  • Never mixes two chapters in one passage
  • Each passage carries its heading path and page range
  • Repeated page headers are removed
03 · FIND

Find the right passage

Hybrid search blends meaning-based embeddings with BM25 keyword matching, then a cross-encoder reranker rescores the best 30 candidates.

  • Near-duplicate passages are dropped and results kept varied
  • Each hit is merged with its neighbouring text for context
  • Passages are never cut off mid-sentence to fit the model
04 · ANSWER

Answer with citations, or say "not found"

The model answers only from the numbered passages and cites them. If nothing relevant clears the bar, the model isn't called at all.

EXAMPLE
› What does Japanese law say about maternity leave?
I couldn't find this in the attached documents.
Closest passage: Code on Social Security, Ch. VI, p. 41
What's in the app today

Everything you need to build an assistant you can trust.

Logmark is the product; SLM Forge is the engine room behind it. Here's what ships in the current build.

Small models, on your own machine

Run open GGUF models such as Qwen 2.5 and Llama 3.2 locally through llama.cpp. No API key, and your documents never leave the computer.

llama.cpp · GGUF · CPU

Assistants over your documents

Give an assistant a set of documents and instructions. It answers from them using retrieval, so the model's weights never change and there's no training run to wait for.

RAG · no fine-tuning

Cited answers and honest refusals

Every answer lists the files and pages it used. Answers without citations get a warning, and off-topic questions get a clear "not found".

sources line · [n] citations

Datasets and evaluation

Generate question-and-answer sets from your own documents, then score your assistants with AI-assisted (RLAIF) and human (RLHF) feedback before you rely on them.

Q&A generation · RLAIF · RLHF

Share an assistant

Share an assistant you've built with a link, or drop a small chat widget onto a web page, like the one in the corner of this site.

share link · embeddable widget

Optional cloud compute

When a job is too heavy for a laptop, you can choose to run it on Google Cloud compute. Day-to-day use stays on your own machine.

optional · GCP
Under the hood

A small, fast engine. No database to run.

Retrieval runs in logmark-core, a Rust engine that keeps every document collection as a folder of plain files.

STORAGE

A manifest, a passages file and a vectors file per collection. Easy to back up, move or delete.

SAFETY

The manifest is written last, so a half-finished write is ignored. A lock stops two indexing runs from colliding.

LOCAL ONLY

The engine's server listens on 127.0.0.1 and only accepts requests from the app itself.

FALLBACK

If the embedding model is missing, search falls back to keywords and says so instead of crashing.

Good fits

For teams whose answers live in long documents.

HR

Policy and benefits questions

Leave, pay and benefits answered from your own handbook, with the page to check.

LEGAL & COMPLIANCE

Regulations and contracts

Ask across long codes and agreements and get the exact clause back.

FINANCE

Controls and procedures

Find the rule behind a process without searching a dozen PDFs.

IT

Internal support guides

Answer common how-to questions from your own runbooks.

SUPPORT

Product documentation

Answers drawn only from approved material, with a clear "not found" otherwise.

RESEARCH

Reports and papers

Keep sensitive reading material on one machine and still search it by meaning.

Where we are

Early, and honest about it.

Logmark is in alpha. Here's what works today and what doesn't exist yet, so you can decide whether it fits.

Works today

  • ✓
    Windows desktop app with local models through llama.cpp
  • ✓
    Assistants over your documents with cited answers and "not found" refusals
  • ✓
    Hybrid search and reranking in the Rust engine, with no database
  • ✓
    Q&A datasets and evaluation with AI and human feedback
  • ✓
    Shareable assistants and optional cloud compute

Not built yet

  • ○
    Single sign-on (SSO)
  • ○
    Audit logs
  • ○
    SOC 2 or other certifications
  • ○
    macOS and Linux apps
  • ○
    Large-model quality : for document chat, 1.5B to 3B models work best today
Pilot programme

Help us build Logmark with you.

We're looking for a small number of teams with real documents and real questions to try Logmark, tell us what's missing, and shape what we build next.

Tell us about your team and what you're trying to solve. We review every application personally and reply within five working days.

By submitting you agree to our privacy policy. We will not share your details with third parties. Pilot places are limited.

What pilot partners get

→Full access to the app for the length of the pilot
→An onboarding session with the people who build it
→A direct say in the roadmap through monthly feedback sessions
→Early access to every new feature as it ships
→Preferred terms when Logmark is generally available

Try it now on Windows

Windows installer · alpha build 9.0.6

Download

Not ready to apply? Let's talk first.

We're happy to have an informal conversation about whether Logmark fits your team. No commitment needed.