AI research workbench · macOS / Windows

From research questionto verifiable result.

SciFound is a desktop workbench for computational science and engineering. Let an agent read papers and project files, create or revise project files, and use available tools within the permissions you grant—then inspect the files and clickable citations.

01Local project storage02Your choice of model03Inspectable files and citations
protein-folding-notesProduct preview
Research taskClaude · API
Compare the assumptions in these two papers and save the verifiable differences to a project file.
01Read papers in projectDone
02Update findings.mdDone

I found three inspectable differences and saved them to the project file.

papers/method-a.pdf · P04-L18findings.md:12-34
Ask a follow-up or assign the next step…

Beyond chat

Research needs a workbench, not a longer chat transcript.

SciFound brings projects, files, agents, terminals, and inspectable citations into one desktop environment. Assign real work and keep the outputs inside the project for inspection and reuse.

How it works

Four steps make a workflow you can continue.

Explore the product
01

Open a project

Create a research project or open an existing local folder.

02

Assign the task

Describe the goal, source material, and constraints—without learning prompt syntax.

03

Let the agent work

Within your permissions, it reads materials, edits project files, and uses available tools.

04

Inspect the result

Open generated files and document citations, then continue or revise.

Real capabilities

Every step returns to your project.

01

Documents & citations

From papers to traceable understanding

Read project PDFs and DOCX files, then follow supported citations back to the relevant document location.
02

Files & terminal

From source material to project files

Create or revise project files, use available terminal tools within scope, and keep results and failures visible.
03

Inspect & continue

Continue from the actual file

Open file paths or document citations, inspect the source, and keep working without losing context.

Local first

Your project starts on your machine.

Project files, conversations, and project memory are stored in the local workspace you choose. Content is transmitted only when required by a model or a networked or sharing action you explicitly enable.

Understand the data flow
LOCAL

Project-scoped access

File operations stay within the current project boundary.

VISIBLE

Visible model source

Know which source is active for the current task.

INSPECT

Inspectable outcomes

Generated files and citations return to real objects.

Research notes

Practical writing on literature, projects, and AI workflows.

Explore all articles
Literature & citations01

Literature & citations · 6 min read

Make every important claim traceable to the paper

Read article
Research workflows02

Research workflows · 5 min read

Why an AI research workflow should start with a project

Read article
Models & data03

Models & data · 7 min read

Four questions to ask before choosing a model source

Read article

SCIFOUND PREVIEW

Put your next research task inside a real project.