AI for computer science research has to start where the field publishes: conferences and preprints. AI for Research searches DBLP and arXiv next to OpenAlex and Semantic Scholar, answers questions with citations to those papers, drafts a conference paper to the venue's page limit, and formats it for IEEE, ACM or another style. It works for engineering research too, with Word, PDF and LaTeX in and out.
Sources for engineering and computing
| Source | What it holds | Where in the product | Credits |
|---|---|---|---|
| DBLP | Computer science publications by venue and author, including conference proceedings | Find Papers and Deep Research | Search 0 |
| arXiv | Preprints in computer science, physics, mathematics and engineering, flagged as not peer reviewed | Preprints source; Find Papers | Search 0 |
| OpenAlex and Semantic Scholar | 250M+ scholarly works across fields, with citation counts | All papers source; Find Papers | Search 0 |
| Web | Standards bodies' pages, technical reports, documentation | Web source in Deep Research | As the run's depth |
| My Library | Your own PDFs, including papers from IEEE Xplore or the ACM Digital Library through your institution | My Library source | As the run's depth |
Deep Research costs by depth: Summary 1 credit, Deep 15, Detailed 45 (paid plans). IEEE Xplore, the ACM Digital Library and Inspec are not searched.
Tools for engineering and computing
| Tool | What it does | Credits |
|---|---|---|
| Find Papers | Search DBLP, arXiv and the general indexes | 0 |
| Deep Research | Cited answers across conferences, journals and preprints | Summary 1, Deep 15, Detailed 45 (paid) |
| Conference paper writer | A 4-8 page paper with contribution statements, to the venue's page limit | 15 |
| Journal formatter | IEEE, ACM, Nature, APA, Vancouver or custom: section order, limits, reference style, and what still breaks the guidelines | 3 per 10,000 words, at most 12 |
| Venue templates | Blank IEEE and ACM templates in the editor | 0 |
| Word and LaTeX export | LaTeX, Word and PDF in; Word and PDF out on every plan, LaTeX out on paid plans | 0 |
| Scientific diagrams | Mermaid flowcharts, architectures and study designs in the editor, SVG or PNG export | 0; Draw with AI 1 |
| Research poster maker | A PPTX poster or a tikzposter .tex file for the conference | 3 |
| Data Lab | Compare benchmark runs with t-tests, Wilcoxon or Bayesian t-tests, calculated in the browser | 0; AI write-up 2 |
Industry R&D teams can also keep benchmarks, standards clauses and patent references in the Technical research workbook in Workbooks.
A worked example
Example (invented for illustration): a PhD student in software engineering is writing a paper on flaky test detection for a conference with an 8-page limit.
- In Find Papers, DBLP source, he searches "flaky tests" for the last five years and saves 30 papers from the main software engineering venues to his Library. 0 credits.
- He adds arXiv preprints from the last six months. 0 credits.
- A Deep Research Summary answer on "which features predict flaky tests, and how are detectors evaluated?" cites twelve of them. 1 credit.
- He draws the pipeline of his method with Draw with AI and edits the Mermaid code: 1 credit.
- He compares his detector with two baselines across 20 projects in Data Lab with a Wilcoxon signed-rank test, free.
- The Conference paper module drafts the paper from his notes, results and Library, 8 pages, with three contribution statements: 15 credits.
- He runs the journal formatter on the 7,000-word draft with the IEEE style: 3 credits.
Total: 20 credits, inside the Free plan's 45 a month. He moves the text into the venue's official LaTeX template for submission; LaTeX export is on paid plans, or he copies from the Word export.
Citation styles and reporting standards
IEEE and ACM reference styles are presets in the journal formatter, with blank venue templates in the editor, and both are among the 10,000+ styles for references. Many conferences accept only their own LaTeX class, so keep the official author kit at hand for the camera-ready. Computing has few formal reporting guidelines like those in medicine, but reviewers expect the dataset version, code revision, hardware and evaluation protocol for every result; the Technical research workbook has fields for exactly these, and the SAMPL guidelines help when you report statistical comparisons between methods.
What it does not do
- It does not search IEEE Xplore, the ACM Digital Library or Inspec. Bring those papers in as PDFs or export files.
- It does not run your code, experiments or simulations, and it does not train models.
- No CAD, finite element analysis or CFD.
- It does not submit to a conference or guarantee acceptance. Use the venue's official template and read its current call for papers.
- Drafts are written from your notes and sources; the contribution and the results must be yours. Check the venue's policy on AI-assisted writing.
Who uses it
- PhD students in computing and engineering: see AI for PhD students.
- Master's students writing a first paper: see AI for master's students.
- Faculty and lab leads: see AI for academics.
- R&D teams in industry: see AI for industry.
For compound and bioactivity sources, see chemistry and materials.
Learn more
- DBLP: a computer science literature search
- How to write a conference paper
- How to format a paper for a journal
- How to make a research poster
- Latest AI research papers
Last updated 2026-11-13