Now in early access

Your research archive,
finally intelligent

Incipit turns fieldwork scans into a persistent, searchable, relationship-aware research archive. Shaped by your intuition. Verified by your expertise.

incipit.dev
El problema indígena y la tierra
Amauta · Lima, Peru · septiembre 1928
PublicationAmauta ✓ Verified
Dateseptiembre 1928 ✓ Verified
LanguageSpanish
AuthorJ.C. Mariátegui ✓ Verified
José Carlos MariáteguiPeruindigenismotierra
The Problem

The historian's workflow is fundamentally broken

You scan 60 documents in a day and they all come out as IMG_0047.pdf. Metadata lives in your head. Connections live on sticky notes. The hunch you had in Lima about a document you saw in New York? Gone.

📄

Documents without identity

Hundreds of files named IMG_0047 or scan_final_v2. The original archive reference codes get lost in the renaming shuffle.

🔍

OCR that can't read your sources

Degraded microfiche, century-old typography, handwritten marginalia, multilingual text. Traditional OCR mangles exactly the documents that matter most.

🧠

Connections only in your head

A Bolivian newspaper mentions the same diplomat as a Puerto Rican letter from a different decade. You'd only notice if you remembered both.

💬

AI tools weren't built for this

You can chat with a handful of documents. But that's a workaround, not a workflow. Your research needs structured metadata, verified facts, and an archive that scales.

How It Works

A research brain that compounds with every document

Incipit doesn't just read documents. It builds an archive that gets smarter over time, shaped by your research goals and your intuition.

👁️

Vision-First Ingestion

AI reads the actual image of your scan, not a broken OCR layer. Degraded microfiche, old Spanish typography, multi-column layouts. It reads what OCR cannot.

Trust Tiers

Every extracted field gets a confidence score. You verify before anything is committed. T1 for confirmed, T2 for high-confidence, T3 for uncertain. No silent guessing.

🔗

Connection Surfacing

Every new document is compared against your entire archive: entities, dates, themes. Informed by your research context. Meaningful connections, not just keyword matches.

📝

Standing Queries

Record a hunch at upload time. That note becomes a live query that activates when a matching document arrives, weeks or months later.

📜

Provenance & Changelog

Where it came from, how you got it, every rename and correction: logged permanently. That gibberish filename might be a catalog reference you need later.

📚

Instant Citations

Chicago/Turabian citations generated from verified metadata. Copy-paste ready. If a field is uncertain, the citation reflects that. Never fabricates.

Why Incipit

A workaround is not a workflow

General-purpose AI tools

  • Upload a handful of documents to a chat project
  • Get prose answers you have to verify yourself
  • No structured metadata or confidence scores
  • No provenance tracking or changelogs
  • No standing queries or automatic connection surfacing
  • Cap at 10 to 40 files per project
I

Incipit

  • A persistent archive that scales with your research
  • Structured, searchable metadata with trust tiers
  • You verify every field before it's committed
  • Provenance, changelogs, and original filenames preserved
  • Research notes become standing queries that activate over time
  • Built for researchers, not developers
Origin

“Built by a historian who spent years in archives across eight countries and built the tool that should have existed.”

Every feature in Incipit comes from direct experience in national archives, university special collections, and private document repositories across Latin America. This isn't a tool designed by engineers guessing at a workflow. It's the tool a researcher needed and nobody had built.

8
Countries researched
500+
Primary sources
Early Access

Your documents deserve
a real archive

Stop losing connections between sources. Start building a research brain that compounds with every document you feed it.