A free Mac utility designed for Claude Code and Codex

Stop paying the
Context Tax.

Messy files make you search. They make your agents search too. AI-Ready Folder turns the files, links, and screenshots you choose into one organized local folder—for you, Claude Code, and Codex—without moving the originals.

Get the free Mac beta on September 12

Email collection is being connected. No address is collected yet.

Once open, submitting asks us to send the beta link and essential release information. Privacy Policy

Estimate My Context Tax Your originals stay where they are.
Illustrative example
PRE-RELEASE BUILDFree beta · September 12Apple Silicon · macOS 14+

THE CONTEXT TAX

Your AI cannot use context
it cannot find.

Before useful work begins, someone has to find the right material, decide what matters, and deliver it in a form the agent can use. When context is scattered, you pay once in preparation—and again when the agent searches, rereads, or works from an incomplete picture.

6.1 workweeks/yearspent finding information across work applications1
More than 1 in 4Codex retrievals missed every required file2
57% more callswhen starting context was irrelevant rather than retrieval-derived3

YOU

The setup tax

  • SpeedSearch, reopen, attach, and explain.
  • CostFocused time becomes context prep.
  • QualityImportant material can be omitted.
CONTEXTtaxpaid twice

YOUR AGENT

The retrieval tax

  • SpeedMore reads, searches, and correction.
  • CostMore calls, rounds, and context.
  • QualityIncomplete inputs risk incomplete output.

Third-party figures describe the problem—not measured AI-Ready Folder savings. See evidence and assumptions.

HOW IT WORKS

Give Claude and Codex a map,
not a junk drawer.

AI-Ready Folder creates a visible local workspace over the context you choose. Browse it in Finder. Start your agents from the same map instead of an improvised pile of attachments and explanations.

BEFORE · SCATTERED
DOCStrategy_final_FINAL.docxDownloads
IMGScreenshot 2026-08-27.pngDesktop
URLAgent researchBrowser
XLSOperating model.xlsxDocuments
AI-READY
FOLDER
You choose
what enters
AFTER · NAVIGABLE
AI-Ready Folder
◇ INDEX.md Start here▾ Product↳ README.md · project map↳ Strategy_final_FINAL.docx · linked▾ Finance↳ README.md · project map↳ Operating model.xlsx · linked▸ Research
Illustrative example
01

Choose what belongs.

Add folders, files, public article links, or screenshots. Review scope and exclusions before anything is built.

02

Build the map.

Projects, subprojects, a root INDEX.md, useful README.md files, and recognizable links to originals.

03

Use it yourself—or with an agent.

Browse in Finder, follow a filename to its original, or begin a supported Claude Code or Codex workflow.

You see projects. Your agents see a navigable context structure. Both lead back to the same originals.

KEEP IT CURRENT

Add what matters.
Keep the map current.

The first sort is only the beginning. Context maintenance becomes a small, repeatable action instead of another cleanup project.

QUICK ADD
+Drop a file, screenshot, or linkChoose its project · keep the original visible

Throw in the next thing.

Paste a public article. Drop a file or screenshot. Preview a clipboard image and choose where its new original belongs.

AGENT CONTINUITY
CCbounded summaryCX

Carry useful work across agents—optionally.

Turn bounded Claude Code or Codex session events into structured summaries the other agent can find later.

Off by default · explicit consent · no hidden reasoning
REPORTS
Workspace pulse

See how the workspace is changing.

Review source mix, recent changes, unresolved items, and captured agent activity—without pretending to score your mind.

YOUR ESTIMATE

How much is context
costing you now?

Five quick choices estimate the time you spend finding existing material, supplying it to AI, and repairing work after context was missed.

30–45 secondsNo email requiredNo AI callVisible assumptions

Calculated in your browser. We do not need your files, prompts, paths, or project names.

Context Tax EstimateRuns locally

1 of 3 · Find it

On a normal workday, how much time do you spend finding things you already know exist?

Count files, links, notes, or screenshots. Do not count research for something new.

BUILT FOR CONTROL

Your files stay
where you put them.

AI-Ready Folder creates a derivative map over the material you choose. It does not reorganize your source folders or quietly scan the rest of your Mac.

01

You choose the scope.

Only folders, files, public links, and clipboard images you deliberately add enter the workflow.

02

Existing originals are unchanged.

The live workspace points back to your files. It does not move, rename, edit, overwrite, or delete them.

03

Local and remote work are separated.

Scanning, metadata inspection, text extraction, and screenshot OCR happen locally. Any later remote processing is bounded and consented.

04

Sign in when processing begins.

Download and inspect first. Passwordless sign-in begins only before the first AI-assisted processing run.

05

The Mac release must pass its trust gate.

The beta ships only after Developer ID signing and Apple notarization from the official site.

A

Built by Tessera. A team developing tools that help people keep their context useful across AI workflows.

FREE MAC BETA · SEPTEMBER 12

Turn your scattered files into a workspace you—and your agents—can actually use.

Bring the files, links, screenshots, and project history you choose into one visible local map. Keep every existing original where it is.

Get the free Mac beta on September 12

Email collection is being connected. No address is collected yet.

Apple Silicon · macOS 14+ · No credit card

QUESTIONS, ANSWERED

Trust starts with
specifics.

Does AI-Ready Folder move or duplicate my files?+

No. Existing originals remain where they are. The generated workspace uses recognizable symbolic links that open those originals. Public URLs appear as .webloc files. A clipboard image becomes one new original only after you preview and confirm it.

What happens locally, and what may use cloud AI?+

Folder scanning, metadata inspection, text extraction, fingerprinting, and screenshot OCR happen on your Mac. After scope review and consent, bounded extracted text or metadata may be sent for sorting and generated summaries. Original file binaries and screenshot images are not uploaded.

Which Macs and file types are supported?+

The first beta targets Apple Silicon Macs running macOS 14 or later. The final file-type list will match the extraction fixtures that pass the release gate; unsupported items remain honest metadata-only pointers.

Which Claude Code and Codex workflows are supported?+

V2 targets supported CLI versions of Claude Code and OpenAI Codex. Claude Desktop, claude.ai, ChatGPT web or desktop, IDE extensions, and other agents are not promised for this beta.

What happens when an original moves or is renamed?+

The reference is marked missing and AI-Ready Folder offers a scoped repair. It will not silently replace a file with a similarly named one; ambiguous matches remain your decision.

What can optional session capture access?+

It is off by default and connected separately per supported agent. It collects bounded lifecycle events for structured summaries—not hidden reasoning or a full transcript archive.

Why is an account needed when processing begins?+

An account is not needed to download, install, select sources, or inspect the local scope. Passwordless sign-in begins only before the first remote AI-processing run for consent, abuse protection, and processing limits.

Who is Tessera?+

Tessera is the team building AI-Ready Folder. We are developing tools that help people keep useful context available across AI workflows.

EVIDENCE & BOUNDARIES

1

Based on the 2021 Qatalog/Cornell Workgeist study: 59 minutes per workday finding information. 6.1 workweeks uses 250 workdays and 40 hours/week. Self-reported, vendor-sponsored, and not limited to local files or our users. Source ↗

2

Agent Retrieval Bench, July 2026: Codex trajectories missed every benchmark-required file in 27.2%–29.3% of 287 logged repository cases. This measured context acquisition—not patch success. Source ↗

3

In a 45-sample seed-intervention pilot, irrelevant starting context averaged 8.49 calls versus 5.42 with retrieval-derived context—57% more using retrieval as the denominator. Source ↗

Third-party evidence establishes the problem, not AI-Ready Folder's effect. Product-specific claims require beta evidence.