Conference template shelf
Multi-user collaboration
Unlimited local compile
AI-native writing UX

LLM-Native Research Writing

Write top‑tier papers in plain language.

Otterflow is a research paper writing assistant that turns natural-language instructions into LaTeX-ready manuscript changes. Start from real conference templates, not a blank file.

Ask for an abstract, rewrite, rebuttal pass, or cleaner bibliography.

Keep literature, local data, citations, and revisions inside one writing flow.


PDF preview

Section 3.2 · Evidence-grounded inference
1
Reasoning quality improves when retrieval traces stay grounded in source evidence.
We anchor the long-context claim in prior work and keep the citation visible inside the manuscript flow
NeurIPS 2026 · Long-context reasoning

Scaling Laws for Reasoners

Quoted evidence

Grounded retrieval improves citation precision and reduces unsupported synthesis in manuscript drafting.
instead of sending the writer to a separate references workflow.
BibTeX readyREADY TO IMPORT
@inproceedings{reasoners2026,
  title={Scaling Laws for Reasoners},
  booktitle={NeurIPS}
}

Citation Flow

Find papers, keep BibTeX, place citations where the argument needs them.

Otterflow can turn literature search and paper selection into citation-ready manuscript operations. The goal is not another tab full of references. The goal is a stronger paragraph with the citation already in place.

Paper discovery to BibTeX-ready records

Inline citations and references placed back into the draft


Natural-language tasks

Draft the abstract from these contributions.

Manuscript sections
GENERATING...
01Abstract
02Related Work
03Rebuttal

Intelligent By Default

Tell Otterflow the paper task. It handles the mechanics.

Describe the writing move in ordinary language. The LLM interprets the scholarly intent, plans the section-level change, and pushes the manuscript forward without forcing you to think like an editor or a formatter.

Draft abstract, related work, method updates, and rebuttal edits

Paper-aware writing behavior instead of generic chat responses


Template shelf#1

NeurIPS 2026

ready
OFFICIAL TEMPLATE
Submission-ready structure

Ready For Real Venues

Start from the template the venue actually expects.

Open a ready-made paper skeleton for NeurIPS, ICLR, ICML, ACL, CVPR, and other conference styles, then move directly into drafting instead of rebuilding the structure from scratch.

Submission-shaped layouts from the first minute

A faster path from idea to conference-ready draft


DeepResearchSRL-001
search papersSTATUS: VERIFIED // ID_672
inspect abstractsSTATUS: PROCESSING // BUFF_OK
extract evidenceSTATUS: PENDING // WAIT_04
System State: StableERR: 0.00%
Local CSV
SPREADSHEET_MODE
#
METRIC
VALUE
DELTA
01
F1_SCORE
0.834
+2.4%
02
PRECISION
0.812
-0.5%
03
RECALL
0.856
+1.8%
04
LATENCY
1.82s
STABLE
05
THROUGHPUT
124/s
+4.1%
06
ERR_RATE
0.00%
0.0%
Chart + table draft
SCALE: 1:0.5
VERIFIED OBSERVATION
Result paragraphPerformance gains remain strongest on long-context benchmarks.

Evidence To Manuscript

Blend DeepResearch and local CSV files into the manuscript.

Run deep research for source discovery, then bring local data into the same writing loop. Otterflow helps turn findings into tables, figures, result paragraphs, and sharper claims.

DeepResearch for grounded literature and source review

CSV to table, chart, caption, and results-section workflow

Desktop Distribution

Bring Otterflow into the writing loop.

Otterflow is now available on macOS and Windows. Use this page to go straight to platform-specific downloads, installs, and update paths.

macOS

Now available

Built for manuscript-first writing workflows on Apple laptops and desktops.

Checking latest build...

Checking latest build...

Windows

Now available

Prepared for research teams that write, revise, and compile on Windows.

Checking latest build...

Checking latest build...