Growing shared archive
Retain reusable structure across compatible files rather than treating every file as an isolated archive.
COMING SOON · LOCAL ARCHIVING
Retain reusable data in one archive and restore selected files exactly. Designed for growing libraries of compatible content.
UK patent application GB2624952.4 · Filed 22 October 2026 · Pending
Windows progress: 92 scoped archive and command-line checks reviewed. Full desktop workflow qualification remains in progress.
Reuse can improve as compatible content joins the same library. Actual savings depend on the data and the structure that can be retained and reused.
Retain reusable structure across compatible files rather than treating every file as an isolated archive.
Process files in sections, with a bounded working set. Capacity depends on available storage and the qualified platform.
Reconstruct and verify the files you request. Other archived content can remain compact.
Compare original and retained sizes, including the archive and restoration record. Keep backups and app storage visible separately.
THE WORKFLOW
Add supported files or folders to a shared archive. Keep the original files while checking the result.
See what was retained and what was saved. Unique data and restoration dependencies remain part of the accounting.
Select files, choose a destination and verify the exact restored content.
RECORDED PRIVATE CANDIDATE
1.651 GB of source content became a 793.6 MB archive and restoration record, compared with a 1.490 GB same-content ZIP.
The recorded candidate was 46.8% smaller than ZIP on this cohort. Runtime and installation storage are additional; this is a private candidate result.
See the comparison →Personal-device, team, business-deployment and OEM licence models are being prepared. Desktop packages and customer onboarding have separate release checks; the current desktop candidate is macOS-first.
UK patent application GB2624952.4 was filed on 22 October 2026 and is pending. Gaming interfaces and AI integrations are separate integration work, with their own qualification steps.
Compatible additions can reuse retained structure. These recorded examples use different datasets and denominators.
Forty new files required 51,712 extra retained bytes for 125,985,692 source bytes: about 2,436:1 for that addition. The existing archive was about 496 MB and still counts. This is not a whole-archive or ZIP ratio.
A separate generated, related CSV study reached a 541,752-byte archive and restoration record after four imports, versus 15,466,144 bytes for ZIP. Runtime and installation storage are additional. Results depend on the data.
THE GLYX ARCHIVE FILM · NARRATED BY AETERNUM · 2:18
Cumulative reuse, a measured 2,436:1 addition, the separate 96.5% and 46.8% ZIP comparisons, and exact selected restoration.
English captions are included in the film. Recorded comparisons retain their stated workloads and measurement boundaries. UK patent application pending.
YOUR DATA. MORE POSSIBILITIES.
The opportunity extends beyond individual files: exact retained records can support other software when their meaning, access and current applicability are handled by that software.
Archive compatible files together and restore just the selected content. Keep retention, backups and recovery within your own policy.
Evaluate storage, transfer and repeated-data costs on the same content and access pattern, with restoration work included.
Explore compact persistence for tool data, structured histories and verified artifacts. Archive restoration supplies data; the application decides what it means and when to use it.
Pair exact records with Seed State reconstruction for recovery and repeated state operations in a qualified integration.
Proposed offers include simple personal/team subscriptions, a 50% share of verified net business savings, and annual deployment or OEM agreements where usage-based licensing fits better. AI services licence a deployment rather than paying automatically for each transient agent.
For a coordinated BBF stack, agree one commercial cost boundary so the same storage or processing saving is charged once.
Compare individual and combined offers →GLYX LICENSING · PRIVATE PROPOSAL
Choose a simple personal licence or evaluate a business savings share. The commercial model follows how you use the archive.
PEOPLE & SMALL TEAMS
A proposed device or team licence for creating and growing archives. Pricing will be set for the qualified release. The proposed licence preserves the ability to restore existing archives after expiry.
BUSINESS & AI SERVICES
Our recommended business model shares the measured financial benefit equally. Compare the same data and service requirements at actual storage rates, with restoration and access costs included. This GLYX tariff is a proposal for agreement.
APPLICATIONS & DEVICE MAKERS
For embedded or distributed use, agree annual capacity bands or OEM rights. Licence the application or service; each temporary AI agent is not automatically another paid seat.
Count physical stored bytes over time, including shared bases, indexes, restoration dependencies and required replicas. Compare equivalent retention, exact restoration and access performance with your existing compression or deduplication. Include changed compute, retrieval and transfer costs; use zero where there is no actual cost.
Illustration only: if an eligible monthly bill falls from £1,000 to £400 after all agreed costs, the £600 saving gives BBF £300 and leaves you £300. Your total is £700. These amounts are assumptions, not a benchmark or quote.
Free space on an already-paid disk is useful capacity; a cash saving arises when a charge or expansion purchase is actually avoided. A combined product agreement counts the same saving once.
Explore the savings calculator →Newer October studies connect exact archival persistence to Seed State's qualified reconstruction. A 17 October versioned-data comparison retained 57.3% fewer logical bytes than trained Zstandard, with exact full and selected-range reconstruction.
This is evidence for the tested archive/state component, separate from the 742-file GLYX desktop cohort. Physical allocation, processing time and comparator details are available together.
Read the latest combined study →NEW PRIVATE ENGINEERING STUDY · 19 OCTOBER 2026
The latest GLYX candidate completed the repeated-content workload in 894.07 ms versus 1,155.85 ms for the existing engine. Four of six tested workloads qualified for a complete-operation improvement, with median elapsed-time reductions of 14.75–22.65%.
All 1,656 exact-output checks passed, and matched archive sizes were unchanged. The candidate uses more working memory: the largest measured child-process peak was about 129 MB, versus about 34 MB on several baseline workloads. This is a private engine candidate, separate from the installed desktop release.
See all six workloads, timing boundaries and memory →