Distillate

Zotero papers → reMarkable highlights → Obsidian notes

Last verified:

Visit Distillate

What is Distillate?

Distillate is an open-source research platform that orchestrates autonomous ML research agents. It turns your research questions into agents that run ML experiments automatically, track results, and report findings while keeping your paper library organized with highlights and AI summaries. The tool is built for ML researchers and data scientists who want to accelerate their research workflow.

The core research loop consists of four stages: add papers to Zotero and highlight on any device (reMarkable, iPad, desktop), conjure experiments by describing a hypothesis, distill insights as experiments run with metrics and diffs tracked, and transmute findings by connecting paper insights to experiment results. Nicolas, the research alchemist, handles the automation - setting up git repos, drafting prompts, and spawning autonomous agents with time budgets.

Key features include CLI commands like /conjure to summon research agents, /steer to adjust running agents, /assay for deep analysis comparing runs, /distill to extract insights from local logs, /survey for experiment overview, and /transmute to turn paper insights into experiment ideas. The Library section includes /brew to sync papers from Zotero, /forage to discover trending research from HuggingFace, and /tincture for deep extraction from individual papers.

Distillate is designed for ML researchers, data scientists, and academic researchers who use Zotero for literature management, reMarkable tablets for reading/highlighting, and Obsidian for notes. It bridges these tools seamlessly, generating rich lab notebooks with run cards and metric sparklines that save directly to your Obsidian vault.

Distillate pricing

Pricing model: Freemium

Free and open-source. Install via pip: uv pip install distillate. The CLI is completely free with no paid tiers. Cloud & Email features require a free account with no credit card needed. Email notifications (experiment reports, daily paper suggestions, weekly digest) are free. AI features require your own ANTHROPIC_API_KEY - without it, AI features are skipped and papers use abstract as fallback summary. Storage depends on Zotero tier (free tier has 300 MB). GitHub Actions automation uses your own GitHub account.

Distillate pros

  • Open-source with CLI powering everything
  • Autonomous ML实验 agents that run themselves
  • Seamless Zotero integration for paper management
  • reMarkable tablet highlight extraction works flawlessly
  • AI-generated summaries for every paper
  • Obsidian vault integration with auto-generated lab notebooks
  • Metrics tracking with diffs and decisions logged
  • /transmute command turns papers into experiment ideas
  • /steer lets you adjust running agent goals mid-experiment
  • Zero API calls for /distill - works from local logs
  • Engagement scores (0-100) for reading depth measurement
  • Automatic citekey naming compatible with Better BibTeX
  • GitHub Actions automation for email digests without laptop running
  • State backup with .bak files prevents data loss
  • Chart export as PNG with --chart flag for sharing
  • Semantic Scholar metadata enrichment auto-fills gaps
  • Highlight back-propagation to Zotero native annotations

Distillate cons

  • Requires Zotero installed as dependency
  • Requires reMarkable tablet for full workflow
  • AI features require ANTHROPIC_API_KEY setup
  • Email notifications need Resend API key configuration
  • GitHub Actions automation requires classic PAT (not built-in token)
  • Desktop app is just a window into the CLI
  • State sync across machines needs manual --sync-state
  • Backfill highlights requires document on reMarkable
  • No native GUI - heavily CLI-focused
  • Custom AI models require .env file configuration

Frequently asked questions about Distillate

What is Distillate?

Distillate is an open-source research platform that orchestrates autonomous ML research agents. It turns research questions into agents that run ML experiments, track results, and report findings while keeping your paper library organized with highlights and AI summaries. Your research alchemist Nicolas handles the automation.

How do I install Distillate?

Run uv pip install distillate, then run distillate --init to start the setup wizard which prompts for your Obsidian vault path and configures Zotero, email, Obsidian, and reMarkable connectors individually.

What tools does Distillate integrate with?

Distillate bridges Zotero for literature management, reMarkable tablets for reading and highlighting, and Obsidian for notes. It also integrates with Semantic Scholar for metadata enrichment, HuggingFace for paper suggestions, and GitHub Actions for automation.

What is the /conjure command?

/conjure summons a research agent. You describe a research question, Nicolas sets up a git repo, drafts the prompt, and spawns an autonomous research agent with a time budget to test your hypothesis.

How does paper syncing work?

Save papers to Zotero, read and highlight on any device (reMarkable, iPad, desktop). Nicolas extracts highlights using rmscene and PyMuPDF, generates AI summaries, and builds your knowledge base. Run /brew or distillate --sync to sync.

What is an engagement score?

Every paper gets an engagement score from 0-100 measuring interaction depth. It combines highlight density (30%), page coverage (40%), and highlight volume (30%). A skimmed paper scores 15-25; thorough highlighting across most pages scores 70+.

Do I need an API key for AI features?

Yes, AI features require ANTHROPIC_API_KEY in your .env file. Without it, AI features are skipped entirely and papers use their abstract as a fallback summary. You can configure custom models via CLAUDE_SMART_MODEL and CLAUDE_FAST_MODEL.

How do I export experiment charts?

Use the --chart flag to export clean metric charts as PNG, designed for sharing on X/Twitter or dropping into papers. The charts include metric sparklines and are auto-generated in lab notebooks.

What happens if state.json gets corrupted?

Distillate automatically backs up corrupt state.json as state.json.bak and starts with fresh state. No data is silently lost - the backup file preserves whatever was there. You can restore from the .bak file manually.

Can Distillate work with the Zotero Integration Obsidian plugin?

Yes, they complement each other. If a note with the same citekey exists, Distillate preserves existing content and appends its sections between <!-- distillate:start --> and <!-- distillate:end --> markers. On re-sync, only content between markers is replaced.

Categories

Use cases

Browse all AI tools on NeedAnAI