{"core-extras.96d330537a":"Base IR","core-extras.bee831a169":"Substrate","core-extras.61a0822826":"Apps on Core","core-extras.15438cc098":"not supported","core-extras.ad565d9d01":"Feature","core-extras.0155bf03e5":"Segment legend","core-extras.8b7110078e":"Algorithm groups by data type","core-extras.25ab4276a6":"Scrollable capability matrix","page-breadcrumb.c766e66518":"Breadcrumb","section_st_a1_display-split.7c9a7c0610":"Detail","section_st_f1_cta-dark.90e40d5043":"Get started","trace-section.e257621ef4":"Constraint trace","uc-research-business-glyph.b76995f081":"Replication as part of delivery","uc-research-business-glyph.e8925256ba":"The record is built while the work happens","uc-research-business-glyph.f7893137fc":"At publication","uc-research-business-glyph.0bddb7562e":"What the result stands on","uc-research-business-glyph.e5e429bcc9":"Data","uc-research-business-glyph.adac69379a":"Code","uc-research-business-glyph.d443a11855":"Environment","uc-research-business-glyph.ba833d1e79":"Model version","uc-research-business-glyph.5485b8031d":"Published figure","uc-research-business-glyph.ee1d36fe5f":"Run again","uc-research-business-glyph.44203f9062":"Who starts from the same record","uc-research-business-glyph.a74f16c117":"Reviewer","uc-research-business-glyph.51346bb80f":"No reconstruction asked for","uc-research-business-glyph.794b34c144":"Collaborator","uc-research-business-glyph.b1f969c735":"The same environment","uc-research-business-glyph.6c3b378e82":"Auditor","uc-research-business-glyph.1f10b34f28":"A traceable path","uc-research-business-glyph.29af8e7a97":"Funder","uc-research-business-glyph.43d894ffc3":"Evidence already on file","uc-research-business-glyph.a64c8bdfae":"Attached at publication, not after","uc-research-business-glyph.601eb30b3a":"Runs in your jurisdiction","uc-research-consumer-glyph.49b4158990":"Read the study, check the path","uc-research-consumer-glyph.cbd0e794d5":"What it was made from, in plain view","uc-research-consumer-glyph.79f2447030":"The figure","uc-research-consumer-glyph.3eaad00bc0":"The data","uc-research-consumer-glyph.32e3ee7a8f":"The code","uc-research-consumer-glyph.c7dcc1acd7":"The research lead","uc-research-consumer-glyph.2a61676d14":"Run it yourself","uc-research-consumer-glyph.b319b06ba1":"Follow the link","uc-research-consumer-glyph.4c5cf05124":"It points at one fixed version.","uc-research-consumer-glyph.8f5f0afd80":"See what was used","uc-research-consumer-glyph.d76b6f09a6":"Data, code and settings are listed.","uc-research-consumer-glyph.0652efd181":"Run it again","uc-research-consumer-glyph.5e808ba841":"The packet replays on your own computer.","uc-research-consumer-glyph.2e352f53a1":"Open where the licence allows","uc-research-consumer-glyph.624192882f":"Kept in Europe","uc-research-technical-glyph.52dd2c1cbc":"Reproduction packet","uc-research-technical-glyph.155e3227b8":"fixed inputs, recorded environment","uc-research-technical-glyph.6f9d9c8004":"Same inputs, same figure","uc-research-technical-glyph.e5e429bcc9":"Data","uc-research-technical-glyph.adac69379a":"Code","uc-research-technical-glyph.d443a11855":"Environment","uc-research-technical-glyph.32fe9f5bda":"Seed","uc-research-technical-glyph.ba833d1e79":"Model version","uc-research-technical-glyph.5485b8031d":"Published figure","uc-research-technical-glyph.c0f85d6679":"Replay","uc-research-technical-glyph.b16f1f6654":"Difference","uc-research-technical-glyph.96341ecc13":"What the packet proves","uc-research-technical-glyph.4c13fe7983":"bit exact","uc-research-technical-glyph.aed3b8c6a7":"Verified","uc-research-technical-glyph.6d525b7156":"Execution","uc-research-technical-glyph.76d24dfc4f":"CPU native","uc-research-technical-glyph.64e63eea6a":"Residency","uc-research-technical-glyph.1e9cb061f2":"European Union","uc-research-technical-glyph.34492da1ce":"In the packet","uc-research-technical-glyph.acadf78430":"Hypothesis","uc-research-technical-glyph.1d197d39c5":"Where it runs","uc-research-technical-glyph.89178d7783":"Integrate via API","uc-research-technical-glyph.195920b8ed":"Hosted in the EU","uc-research-technical-glyph.46a62054bf":"Air gapped","uc-research-technical-glyph.a64c8bdfae":"Attached at publication, not after","uc-research-technical-glyph.601eb30b3a":"Runs in your jurisdiction","use-case-research-org.be601df25e":"Research","use-case-research-org.14a29ba383":"Use Cases","use-case-research-org.50853a6cd1":"Existing notebooks, scripts, and lab information systems keep working. Dweve wraps\n            them in a deterministic decision layer that records every input, every rule, and\n            every PI signature as the figure is produced.","use-case-research-org.190684eaf9":"Research, technical","use-case-research-org.48cd18dabc":"Your data","use-case-research-org.8bf34da1c7":"One platform","use-case-research-org.b3498e29f2":"Closed bundle for classified or sensitive data.","use-case-research-org.ec9dfc4fd3":"Fully offline","use-case-research-org.9640c6d419":"Run on your lab hardware, on-prem.","use-case-research-org.3345c1a90d":"Your lab cluster","use-case-research-org.55423877a1":"Dweve-operated public Mesh; processing boundary documented in the agreement.","use-case-research-org.61810d3c5a":"Business Workspace","use-case-research-org.2b1884c5a7":"One platform, three postures","use-case-research-org.215475056e":"WHERE IT RUNS","use-case-research-org.92a16f5060":"SDK: TS/Python/Go/Rust","use-case-research-org.4b3258f7ec":"RFC 7807","use-case-research-org.92af23aac4":"PROTOBUF","use-case-research-org.dabbb6f4e1":"OPENAPI 3.1","use-case-research-org.3c8c96455a":"Integrate via REST or gRPC against a typed OpenAPI 3.1 surface. Managed through Fabric on the public Dweve Mesh for pilots. Licensed for your lab cluster. Air-gapped for classified data. Same API, same trace format, same replay packets across all three.","use-case-research-org.0ff8b5b425":"Three deployment postures.","use-case-research-org.94188e2c6c":"One API.","use-case-research-org.34bfbaefff":"How you integrate","use-case-research-org.80b30e032e":"Bit-identical output","use-case-research-org.f2aada205c":"Same notebook","use-case-research-org.e2194037eb":"PI signs the final figure.","use-case-research-org.8eeb40cd07":"Sign packet","use-case-research-org.d74ce837aa":"Pre-registered rule applied to outputs.","use-case-research-org.7baf054b92":"Apply rule","use-case-research-org.a35e6eece0":"Cells produce raw outputs.","use-case-research-org.05aa7221aa":"Run cells","use-case-research-org.6e4db855b4":"Inputs logged before the notebook runs.","use-case-research-org.9678dd9fa7":"Pin inputs","use-case-research-org.478360978f":"Four guarantees","use-case-research-org.c00f9811f5":"WHAT THE LAYER DOES","use-case-research-org.3b720bec28":"SIGNED REPLAY","use-case-research-org.6757848af7":"DETERMINISTIC","use-case-research-org.44193f9e72":"Existing Jupyter notebooks, R scripts, and Stata do-files keep working. Dweve wraps them in a deterministic decision layer that records every cell, every rule, and every PI signature as the figure is produced. The replay packet is the audit artefact. Inputs are logged before the notebook runs, the cells produce their raw outputs, the pre-registered rule is applied to them, and the PI signs the final figure. The notebook keeps working, and the packet it produces is what an audit reads.","use-case-research-org.fb834ca63c":"bit-identical replay.","use-case-research-org.ee5ae02b3a":"Notebooks wrapped,","use-case-research-org.ae9f1986d7":"Analysis determinism","use-case-research-org.83d5dcbc4e":"One signed hypothesis","use-case-research-org.01c501a76c":"Four steps","use-case-research-org.7aa986a02b":"PI signs; packet is the audit artefact.","use-case-research-org.8b3d8d665a":"Sign","use-case-research-org.df8e013ec0":"Deterministic compute runs the analysis.","use-case-research-org.b1b392607d":"Run","use-case-research-org.1d4dc675c0":"Rule and model version frozen at registration.","use-case-research-org.5a6b2d30fe":"Freeze","use-case-research-org.23a0e0dad2":"Hypothesis framed, registered before tests.","use-case-research-org.d92a04e3ca":"Pose","use-case-research-org.1779864412":"PINNED","use-case-research-org.d3737a6efb":"FROZEN","use-case-research-org.fddbe6d281":"Pre-registration is the workflow most damaged by post-hoc analysis. Dweve freezes the hypothesis, the rule, and the model version at registration time. The replay reconstructs the exact analysis the PI committed to before looking at the data. Four steps hold it in place: the hypothesis is framed and registered before any test is run, the rule and the model version are frozen at registration, deterministic compute runs the analysis, and the PI signs the packet that results. That packet is what a replication attempt reads, so the analysis it reconstructs is the one committed to in advance.","use-case-research-org.9f84443f8b":"Hypotheses frozen,","use-case-research-org.eb88c9b949":"Pre-registration","use-case-research-org.078a0e1e72":"The decision layer never crosses a trust boundary that fails the GDPR or ethics\n            rules. The same topology ships on all three postures.","use-case-research-org.d5a9f6f81a":"Three postures","use-case-research-org.8ab2c84ed7":"One architecture","use-case-research-org.9d6d963345":"No outbound calls. Optional EU-only Zenodo endpoint.","use-case-research-org.8e05a902ed":"Egress","use-case-research-org.8c0e6f760f":"Signed packets written once, read many, EU-only replica.","use-case-research-org.dcf17fd359":"Packet store","use-case-research-org.6de6f8a697":"CPU-native engine, no GPU farm, no cloud ML APIs.","use-case-research-org.7f59a1f1d5":"Decision","use-case-research-org.7d853df929":"TLS terminate in EU, mTLS to the decision layer.","use-case-research-org.92096d78df":"Ingress","use-case-research-org.7f166ac611":"Four boundaries","use-case-research-org.338ac926aa":"TOPOLOGY","use-case-research-org.6b4217aac7":"EU ONLY","use-case-research-org.5e905a995e":"PACKET STORE","use-case-research-org.7e826f7af3":"TRUST","use-case-research-org.138a224dc4":"DMZ","use-case-research-org.c83c4eed51":"Three operating models, one evidence architecture. Managed Fabric runs on the public Mesh. Licensed operation brings direct products into your lab cluster. Air-gapped operation protects classified or sensitive-subject data. Each model keeps decisions and evidence inside its declared trust boundary. The same four boundaries are drawn in each one: an ingress that terminates TLS in the EU, the decision layer behind the trust boundary, the packet store, and an egress that makes no outbound calls.","use-case-research-org.3ae8029f90":"per posture.","use-case-research-org.52c48d0dc9":"Where the layers sit,","use-case-research-org.f9790fb715":"Deployment topology","use-case-research-org.2c2bb99178":"Determinism is asserted in the suite, not promised in prose. The same oracle run on\n            your hardware produces the same answer as on ours.","use-case-research-org.064f903275":"Not aspirational","use-case-research-org.aed3b8c6a7":"Verified","use-case-research-org.ed0ff2382e":"No rounding error crosses the unit-of-last-place threshold.","use-case-research-org.aadefaf8ed":"Correctly rounded","use-case-research-org.cc376f246e":"Corners and ranges exercised automatically in CI.","use-case-research-org.e02eea5453":"Property tests","use-case-research-org.a24d9583df":"Each result compared to a high-precision reference.","use-case-research-org.47909e2975":"256-bit oracle","use-case-research-org.4ac72fd74a":"Bounded iteration count, no floating-point unit.","use-case-research-org.97dab07ad3":"CORDIC compute","use-case-research-org.b9a7f2f458":"How the bound is proven","use-case-research-org.dc0507035e":"VERIFICATION","use-case-research-org.481d3e216b":"PROPERTY TESTED","use-case-research-org.fdd71ee1c4":"MPFR 256-BIT","use-case-research-org.51ab88da6e":"0 ULP","use-case-research-org.30400b7314":"Every numerical computation in the analysis path is verified against MPFR at 256-bit precision. The published result is 0 ULP, correctly rounded across the documented input domain, with property-based tests driving the corners on every release. The bound is asserted in the test suite on every release rather than promised in prose, so the same oracle run on your hardware returns what it returns on ours.","use-case-research-org.bed4aa0e07":"checked against an oracle.","use-case-research-org.0479003b2b":"0 ULP transcendentals,","use-case-research-org.f3e7f51455":"Verified figures","use-case-research-org.e63a99077f":"One signed chain","use-case-research-org.65409fb2e6":"Final figure reproduces bit-identical.","use-case-research-org.5faa59d4bc":"Result","use-case-research-org.5cb4913d6e":"Cleaning rule applied, deterministic pipeline.","use-case-research-org.12f4c13a6a":"Process","use-case-research-org.940b7a1235":"Raw samples hashed and sealed at intake.","use-case-research-org.da433cd41e":"Raw","use-case-research-org.e3ae3bc8ea":"Source dataset, ethics, DUA signed.","use-case-research-org.6da13addb0":"Source","use-case-research-org.414cbcc9ec":"PROVENANCE","use-case-research-org.da4b6b1d2a":"RESULT","use-case-research-org.7515346cdb":"PROCESSED","use-case-research-org.c5db7969dc":"RAW","use-case-research-org.e9ae0ed784":"SOURCE","use-case-research-org.15bcc1d322":"Every dataset links to its lineage back to source. Source, raw, processed, features, result. Each step signed, each rule versioned, each PI named. The replay reconstructs the exact figure years later. Four steps make the chain: the source dataset arrives with its ethics approval and data use agreement signed, the raw samples are hashed and sealed at intake, a versioned cleaning rule runs through a deterministic pipeline, and the final figure reproduces bit-identical. Because it is a chain rather than a folder, no link can be re-cut without the next one showing it.","use-case-research-org.033b330e09":"chained and signed.","use-case-research-org.0d62d74600":"Source to figure,","use-case-research-org.278845c433":"Data provenance","use-case-research-org.f5d4349e08":"Full trace","use-case-research-org.3477f4d91d":"Same answer","use-case-research-org.17a22df4e0":"Managed Fabric, licensed on your hardware, or air-gapped.","use-case-research-org.65cfa78989":"Sovereign","use-case-research-org.e56a362553":"OpenAPI 3.1 spec, typed errors (RFC 7807).","use-case-research-org.cb885a925a":"REST + gRPC","use-case-research-org.8fca8dd445":"Data, code, environment, PI, signature.","use-case-research-org.0788e3e212":"Signed packet","use-case-research-org.6cbcecba9f":"Same input, same output, every machine, every release.","use-case-research-org.02ae1c41fe":"Exact replay","use-case-research-org.0753940ae7":"Four properties","use-case-research-org.06ebfec4be":"WHAT IT GIVES YOU","use-case-research-org.30d7cf38f4":"Every figure, table, and number in a paper runs on a deterministic engine. Bit-exact replay across machines and releases. Same figure on the PI laptop, the cluster, and the reviewer workstation. No floating-point drift, no chip-dependent rounding. The packet carries the data, the code, the environment, the PI and the signature, and the same four properties hold on managed Fabric, licensed on your hardware, or air-gapped.","use-case-research-org.14cfcf3fb2":"Data, code, environment,","use-case-research-org.a6d4aac61a":"Reproduction packets","use-case-research-org.43decaea2c":"You do not need a lab of your own to challenge a result. The packet is in plain\n            language and the data, the code, and the PI are named on the paper.","use-case-research-org.a0a0e902a0":"Research, consumer","use-case-research-org.c605f28db5":"On the record","use-case-research-org.a0f441d201":"Your study","use-case-research-org.8b1df414b4":"Same packet the journal sees.","use-case-research-org.1c54132b8e":"Record","use-case-research-org.2874739dbe":"Independent replication, published.","use-case-research-org.16462c5c97":"Replicate","use-case-research-org.75f527181b":"Free","use-case-research-org.fc3df8177f":"Journal comment, named.","use-case-research-org.153d7a58b3":"Comment","use-case-research-org.06310f4e07":"Replay the packet on your laptop.","use-case-research-org.75e6a088a0":"Four routes","use-case-research-org.bc3e09058f":"YOUR PATHS","use-case-research-org.394020b42e":"RETRACT","use-case-research-org.b1a13fadf5":"COMMENT","use-case-research-org.87d43cf9f7":"REPLICATE","use-case-research-org.cf1951d08f":"Every published result is replicable. The same packet the PI signed is the packet you can replicate on your own hardware. If you disagree with the figure, the replication path is on the paper.","use-case-research-org.573d91c3b3":"can challenge.","use-case-research-org.0adf6c84a9":"A result you","use-case-research-org.b21e4a35ed":"Your replication path","use-case-research-org.11234a16b5":"One record","use-case-research-org.06686324e6":"One figure","use-case-research-org.5d9d30ba83":"Paper + packet accepted by journal.","use-case-research-org.5656400564":"Publish","use-case-research-org.f8ebf6a23b":"Pre-registration frozen, rule pinned.","use-case-research-org.59b0353672":"Design","use-case-research-org.aa4b0a1987":"Hypothesis framed, pre-registered.","use-case-research-org.92f5fe6952":"Four outcomes","use-case-research-org.a76f8ee287":"HOW IT WORKS","use-case-research-org.ff0fc30c74":"PUBLISH","use-case-research-org.ad173e04d8":"RUN","use-case-research-org.8d2fccfa70":"DESIGN","use-case-research-org.5f4a3c7c18":"POSE","use-case-research-org.393141a3b4":"A figure in a paper is no longer a black box. The hypothesis, the data, the code, and the PI sit together from the moment the analysis runs. If the reviewer asks, the author can show why the figure came out the way it did. Four moments are on the record: the hypothesis is framed and pre-registered, the pre-registration is frozen with the rule pinned, deterministic compute runs the analysis, and the paper reaches the journal with its packet.","use-case-research-org.dfb39fb4ea":"with its record attached.","use-case-research-org.782638932a":"A figure moves","use-case-research-org.1b542b2751":"How peer review works","use-case-research-org.37c0201d91":"Replay on your laptop.","use-case-research-org.993a665d35":"In charge","use-case-research-org.905cb326c7":"You","use-case-research-org.4d73d375e8":"Built and run under EU rules.","use-case-research-org.ef5fdf226f":"Footing","use-case-research-org.c82d1b09cf":"EU","use-case-research-org.057ef5b849":"The permanent identifier.","use-case-research-org.5d31a45fe6":"DOI","use-case-research-org.8e16e5b2c4":"Real","use-case-research-org.1641a6eed5":"A reason, not a guess.","use-case-research-org.89b86ab0e6":"Language","use-case-research-org.34118050b2":"Plain","use-case-research-org.3ff1e4c972":"When your lab runs on Dweve","use-case-research-org.f31727ff20":"WHAT YOU GET","use-case-research-org.126cbdcb1c":"EU DATA","use-case-research-org.87536362b1":"NAMED PI","use-case-research-org.b838c172bb":"PLAIN LANGUAGE","use-case-research-org.70351f83c8":"When your research lab uses Dweve, the study you read is not a black box. It is tied to a DOI, a named PI, and a packet you can replay on your own laptop. If you ask why, the answer is on the paper. Four things come with it: plain language, a permanent identifier, a named investigator, and a packet you can open on your own computer without needing the statistics to read it.","use-case-research-org.a050979b99":"can replay.","use-case-research-org.9213b7f1b0":"A study you","use-case-research-org.f1c33192de":"Your side of the study","use-case-research-org.088de83576":"The data stays inside your jurisdiction. The same platform scales from a single\n            PhD project to a full multi-lab consortium without re-platforming.","use-case-research-org.e713cc03be":"Research, business","use-case-research-org.938036d5be":"Closed bundle for classified or sensitive-subject data.","use-case-research-org.2e54241686":"Run on your lab cluster. Same API, same trace format.","use-case-research-org.08a557101e":"Your hardware","use-case-research-org.0058df0a6e":"Dweve-operated public Mesh; processing boundary documented in the agreement.","use-case-research-org.7853827da5":"LICENSED","use-case-research-org.67b752639a":"MANAGED MESH","use-case-research-org.5ddd6023e9":"Two questions decide procurement: where does it run, and which rules does it satisfy. It runs managed through Fabric on the public Dweve Mesh, on your hardware, or fully air-gapped for classified data. GDPR, FAIR, and the EU AI Act are design constraints the platform satisfies by construction.","use-case-research-org.3fb9f3f8b5":"Run on your hardware.","use-case-research-org.560486613b":"Start with the managed service.","use-case-research-org.1d197d39c5":"Where it runs","use-case-research-org.1f3f87ef63":"The funder report, the ethics audit, and the peer review all pull from one packet.\n            One source of truth, three audiences.","use-case-research-org.9aab9fcbfd":"One source of truth","use-case-research-org.4bd812de1a":"Four frameworks","use-case-research-org.0298a442a4":"Packets archived per funder retention rule.","use-case-research-org.2621c6fd51":"Archive","use-case-research-org.6c14c85aa9":"Funder report pulls from the same packets.","use-case-research-org.ee45c30326":"Report","use-case-research-org.ec84692ba4":"Work proceeds with packets produced as it happens.","use-case-research-org.ca46da0ee0":"Data management plan signed before work starts.","use-case-research-org.ae2f98a099":"Plan","use-case-research-org.c4d9bc6083":"COMPLIANCE","use-case-research-org.77d0f39625":"DFG","use-case-research-org.ffc17ee091":"NWO","use-case-research-org.574fc55fa6":"ERC","use-case-research-org.c0d273fb48":"HORIZON","use-case-research-org.7475c02863":"Funding agencies and ethics boards increasingly require reproducibility evidence. Dweve produces the artefacts each framework expects, as the work happens. EU Horizon Europe, ERC, NWO, and DFG reports all draw from the same packets. The sequence is the same under each: a data management plan is signed before the work starts, packets are produced as the work happens, the funder report pulls from those packets, and they are archived under the retention rule the funder sets. The funder report, the ethics audit and the peer review end up as three readings of one record rather than three reconstructions.","use-case-research-org.63ecd37e98":"answered in hours.","use-case-research-org.acb46f2545":"A grant report","use-case-research-org.5de1a82431":"Funding and ethics","use-case-research-org.93b48c8079":"One packet","use-case-research-org.185d5e5e4d":"One paper","use-case-research-org.b3c679f5b3":"Signed packet reconstructs every figure.","use-case-research-org.83c6d723cb":"Packet","use-case-research-org.6052459d7b":"CPU + OS + toolchain captured for replay.","use-case-research-org.d443a11855":"Environment","use-case-research-org.d5721485f0":"Git snapshot, container image, hash pinned.","use-case-research-org.adac69379a":"Code","use-case-research-org.ffa028950f":"DOI cited, FAIR, EU-only mirror.","use-case-research-org.e5e429bcc9":"Data","use-case-research-org.c02908a2e1":"WHAT CHANGES THIS PAPER","use-case-research-org.b375f14e2e":"PACKET","use-case-research-org.086bfb42cb":"ENV","use-case-research-org.16de25af88":"CODE","use-case-research-org.580393f5a9":"DATA","use-case-research-org.01556fa507":"Every paper produces one signed packet. Data DOI-cited, code hashed, environment captured, figure pinned. The reviewer, the replicator, and the audit firm all pull the same artefact. One source of truth, three audiences. The data is DOI-cited and FAIR with an EU-only mirror, the code is a git snapshot and container image with the hash pinned, the environment records CPU, operating system and toolchain, and the signed packet reconstructs every figure from those three.","use-case-research-org.256a5d9627":"all attached.","use-case-research-org.73e7b8ac40":"The reproduction packet","use-case-research-org.5d1cc3c5c6":"Jurisdiction","use-case-research-org.b59107995f":"Opaque figures","use-case-research-org.f68cece41b":"Trace attached","use-case-research-org.bb28a44c66":"Signed packet the reviewer replays in under a minute","use-case-research-org.4e250b4fde":"REPLAY","use-case-research-org.40e8937632":"PI Jansen, signature, timestamp","use-case-research-org.501e489055":"OWNER","use-case-research-org.30ee932fc8":"Pre-registration 2026-0184, hypothesis H3","use-case-research-org.aaa398a28b":"RULE","use-case-research-org.c43713f495":"Raw samples, ethics record, DUA","use-case-research-org.beb64c0d3c":"ethics-EUR-2026-014.pdf","use-case-research-org.250c86f186":"Pre-registration 2026-0184 hypothesis H3, tested against 14,820 samples collected under ethics approval EUR-2026-014. Feature extractor v2.4 applied, deterministic compute produced the survival curve. PI Jansen signed the figure on 2026-06-04. The same figure replays bit-identical, with the data, the rule, and the PI on one screen.","use-case-research-org.7f4a184ece":"THE RECORD SHOWS","use-case-research-org.7a78b3966d":"How was figure 3 produced?","use-case-research-org.3fd17fa07f":"THE REVIEWER ASKS","use-case-research-org.a4fa213203":"Every figure, table, and number leaves a signed packet: the data, the rule, the environment, the PI, and a bit-exact replay. The reviewer opens one file and watches the same figure come back.","use-case-research-org.407e99bffe":"The packet the reviewer replays in under a minute.","use-case-research-org.404756588e":"The analysis is the easy part. Reconstructing the why of every figure six months\n              later, in front of reviewer 2, is the hard part.","use-case-research-org.b3ea657ff0":"Each blocks review","use-case-research-org.ec95d99629":"Four gaps","use-case-research-org.e10fad6d31":"A chat log does not replay the same way twice.","use-case-research-org.fade4edeec":"No evidence","use-case-research-org.e7c750f4bc":"Foreign-hosted data fails GDPR + ethics rules.","use-case-research-org.e03947cc0c":"Sovereignty","use-case-research-org.144b5ad1ae":"A figure is not a result a reviewer can replay.","use-case-research-org.8479f2b690":"Black box","use-case-research-org.b0402d8654":"It invents a citation instead of pulling the real source.","use-case-research-org.a978d8e92c":"Hallucination","use-case-research-org.782cef5501":"Four structural gaps","use-case-research-org.e6ff64bc5a":"WHY GENERIC AI FAILS PEER REVIEW","use-case-research-org.001fec2804":"A research director does not buy a platform. She survives peer review, replication studies, and funding audits. FAIR data principles, Open Access mandates, GDPR, and the EU AI Act each want the same thing: the hypothesis, the data, the code, the environment, and the PI of every figure. Today that record lives in OSF, Zenodo, GitHub, and email threads that were never meant to agree.","use-case-research-org.a223c6b8d6":"a series of signed packets.","use-case-research-org.786bccf585":"A study is","use-case-research-org.8f4f71b63c":"The research pipeline","use-case-research-org.a4fb970cd2":"See where it runs","use-case-research-org.485649d24f":"Talk to a research engineer","use-case-research-org.3b74e3fb86":"Pick one workflow: pre-registration, data ingestion, analysis run, or peer review response. Run it with your own data and watch the reproduction packet assemble as the work moves. Managed through Fabric on the public Dweve Mesh, on your hardware, or air-gapped.","use-case-research-org.65d12da14e":"Start with the figure you most dread reproducing.","use-case-research-org.87e1309154":"Reproducible","use-case-research-org.d75a4de2ef":"EU AI Act","use-case-research-org.a5b7d57a85":"GDPR","use-case-research-org.c61a7c7f39":"FAIR","use-case-research-org.55e2ce0ca5":"Open Access","use-case-research-org.dd35a81646":"Provenance","use-case-research-org.52dd2c1cbc":"Reproduction packet","use-case-research-org.2cf1e080ec":"Bit-exact replay for every figure, table, and number in a paper. CPU-native, 0 ULP verified, EU-sovereign. Integrate via API, host in the EU, or run air-gapped on classified data.","use-case-research-org.c7df82990c":"from recorded inputs.","use-case-research-org.4579755910":"Reproduce the figure","use-case-research-org.9a402e5819":"Your study card","use-case-research-org.83c1151840":"A result is easier to trust when you can inspect the path behind it. Dweve links the study to its DOI, research lead, data, and a packet you can replay on your own computer.","use-case-research-org.1b5e7db2e4":"Check how it was made.","use-case-research-org.77168c1785":"Read the study.","use-case-research-org.bc408cf6c5":"See a reproduction packet","use-case-research-org.08e41f7da0":"Publish each result with the hypothesis, data, code, environment, and model version already attached. Reviewers, collaborators, and auditors start from the same executable record instead of asking the team to reconstruct it months later when reviewer 2 asks.","use-case-research-org.1b14e4a2c8":"part of delivery.","use-case-research-org.a1621309c9":"Make replication","use-case-research-org.0814ece775":"Packet reconstructs every figure on your laptop","use-case-research-org.b4b6270d9c":"10.5281/zenodo.2026.0184, permanent","use-case-research-org.989842c17d":"Jansen, Vermeulen, El Amrani (EUR)","use-case-research-org.dad57a4c24":"AUTHORS","use-case-research-org.071490b130":", on every CPU tested, every compiler, every release","use-case-research-org.5397e0583f":"Yes","use-case-research-org.64b080ec54":"ANSWER","use-case-research-org.7d3f908304":"Do fixed-point kernels replay bit-identically across CPUs?","use-case-research-org.dcf89b04f7":"QUESTION","use-case-research-org.2d1dfaefde":"ACCEPTED","use-case-research-org.84a6746267":"Your study, in plain language","use-case-research-org.085cbca419":"Replay reconstructs the same figure","use-case-research-org.05bd762e38":"Every dataset links to its lineage back to source","use-case-research-org.eba1448014":"Final figure reproduces bit-identical","use-case-research-org.0be77c50de":"Feature extractor pinned to model version","use-case-research-org.54e01da89b":"FEATURES","use-case-research-org.3afc885c61":"rule v3","use-case-research-org.372d071310":"Cleaning rule v3, deterministic pipeline","use-case-research-org.82fc93a50b":", hashed and sealed at intake","use-case-research-org.227e204bae":"14,820 samples","use-case-research-org.8a1532013d":"DUA signed","use-case-research-org.0e99f1272a":"EUR lab consortium, ethics approval 2026-014","use-case-research-org.a9ff4c278e":"PER PAPER / CHAINED","use-case-research-org.23429d2f6f":"Data provenance, on the record","use-case-research-org.11dc9e1952":"Status:","use-case-research-org.f8e9e1863f":"Replay packet","use-case-research-org.5e1dc0db8a":"Code archive","use-case-research-org.ef8513063a":"Raw + processed data, DOI-cited, EU-only mirror.","use-case-research-org.545b68a284":"Data packet","use-case-research-org.9c5083fb98":"Replay reconstructs the full lifecycle","use-case-research-org.1037c5db6c":"Each step has a named PI and a timestamp","use-case-research-org.dbb0c335a3":"PRE-REG 2026-0184","use-case-research-org.74f1d9d503":"Hypothesis lifecycle, on the record","use-case-research-org.0102c8644b":"FILED / TRACEABLE / REPRODUCIBLE","use-case-research-org.3c9e174d9c":"REPLAYABLE","use-case-research-org.869395dd34":"Data, code, environment pinned","use-case-research-org.ce0d0bb09d":"PERMANENT","use-case-research-org.7fced9588f":"ORCID","use-case-research-org.73af38860a":"Bit-exact replay of fixed-point kernels","use-case-research-org.bf654c3810":"TITLE","use-case-research-org.de72c478b4":"arXiv / DOI / ORCID","use-case-research-org.366bb8056e":"PAPER / 2026-0184","use-case-research-org.d19e97c9a9":"REPRODUCIBLE","use-case-research-org.505324a7dd":"PEER REVIEWED","use-case-research-org.7a1c4e536c":"OPEN ACCESS","use-case-research-org.fd815047d7":"EU-ONLY","use-case-research-org.1dabba21cd":"✓","use-case-research-org.fbcec86144":"Business Workspace","use-case-research-org.87feb88216":"Licensed","use-case-research-org.c73cc64719":"Air-gap","use-case-research-org.869360a6f4":"Bit","use-case-research-org.c8e5998f6a":"API","use-case-research-org.9170a8b2fb":"EU","use-case-research-org.2876814bd7":"Integer","use-case-research-org.adbc560c32":"MPFR","use-case-research-org.f23ba2a6ee":"proptest","use-case-research-org.be258ce66f":"0 ULP","use-case-research-org.cc4dbff4ea":"PRE-REGISTRATION","use-case-research-org.5d0b2f6bd4":"OBSERVED","use-case-research-org.80db3832a5":"OUTCOMES","use-case-research-org.d072aa117d":"WORKFLOW","use-case-research-org.feb892b2eb":"DEPLOYMENT","use-case-research-org.f830ea0076":"READER","use-case-research-org.00e59c6e64":"PEER REVIEW","use-case-research-org.6f38f66d9a":"PLATFORM","use-case-research-org.39bda14904":"EVERY RELEASE","use-case-research-org.5ce41bc339":"ARCHITECTURE","use-case-research-org.46b9927e1d":"DETERMINISM","use-case-research-org.pipeline.signedPi":"Signed by the principal investigator","use-case-research-org.pipeline.versionPinned":"Protocol version pinned","use-case-research-org.pipeline.running":"Analysis running in the recorded environment","use-case-research-org.pipeline.awaitingReview":"Publication packet awaiting review","use-case-research-org.pipeline.reviewPacket":"Review packet","use-case-research-org.pipeline.fourQuestions":"Four questions, four records","use-case-research-org.pipeline.questionPlanned":"What was planned?","use-case-research-org.pipeline.answerPrereg":"Pre-registration","use-case-research-org.pipeline.questionData":"Which data was used?","use-case-research-org.pipeline.answerDoi":"Dataset DOI","use-case-research-org.pipeline.questionRan":"What actually ran?","use-case-research-org.pipeline.answerEnv":"Code and environment","use-case-research-org.pipeline.questionOwns":"Who owns the result?","use-case-research-org.pipeline.answerPi":"Named investigator","use-case-research-org.pipeline.oneStudy":"One study","use-case-research-org.pipeline.onePacket":"One replayable packet","use-case-research-org.e780933db6":"A study is a series of packets, each one signed as the work happens rather than reconstructed months later.","use-case-research-org.0acb838238":"One question, one file","use-case-research-org.f277f1e344":"One file answers the reviewer's question, and the same file replays the figure that answered it.","use-case-research-org.2a6a716288":"Data, code, environment and packet are produced together per paper, so three audiences read one artefact.","use-case-research-org.b5dee6b30c":"The funder report, the ethics audit and the peer review are three readings of one record, not three reconstructions.","use-case-research-org.3ff13b67dd":"Procurement is choosing a boundary. Nothing inside the boundary changes when you choose a different one.","use-case-research-org.516b06e363":"The study you read is tied to a permanent identifier, a named investigator and a packet you can open on your own computer.","use-case-research-org.52fb1eb424":"The hypothesis, the data, the code and the investigator travel with the figure from the moment the analysis runs.","use-case-research-org.a5063bd43a":"Three of the four ways to challenge a published figure need nothing more than the packet and your own computer.","use-case-research-org.152945194a":"Reproducibility is a property of the engine here, not a practice the team has to keep up on its own.","use-case-research-org.b998611ad9":"Lineage is a chain rather than a folder. No link can be re-cut without the next one showing it.","use-case-research-org.abaff5bab7":"Determinism is asserted in the suite on every release, not promised in prose.","use-case-research-org.e7cf01e44c":"Three operating models, one evidence architecture, and the same trust boundary drawn in three different places.","use-case-research-org.eeb732fdd5":"The replay reconstructs the analysis the investigator committed to before the data was looked at.","use-case-research-org.e3925ac97f":"The wrapped notebook keeps working, and the packet it produces is the audit artefact.","use-case-research-org.2007b7051e":"One typed surface across all three postures, so integrating once is integrating everywhere.","use-case-research-org.ef95aa4b6e":"One packet per paper holds the data with its identifier, the code at the exact commit, the environment that ran it and the signature that closed it.","use-case-research-org.5882949f28":"The reviewer, the independent replicator and the audit all open that same artefact, so there is no second version to reconcile.","use-case-research-org.b5e86f926d":"The identifier on the paper is permanent, so the study you read today is the study anyone else opens from the same link years from now.","use-case-research-org.0621db1610":"The packet is written in plain language first. You do not need the statistics to see which data was used, who signed the figure and how to run it again.","use-case-research-org.78d84a8e09":"Intake is where the chain starts. Raw samples are hashed and sealed on arrival, so every later step refers to a fixed object rather than a moving one.","use-case-research-org.2bfb88b0d8":"Feature extraction pins to a model version, which is what lets a figure be reproduced on a release that did not exist when it was published.","use-case-research-org.66c1f5342b":"One file, opened once","use-case-research-org.d7ab934854":"The same figure comes back","use-case-research-org.a844fcf834":"Registered","use-case-research-org.9b064af95c":"four stages","use-case-research-org.bca102c9e7":"One paper, four attached parts","use-case-research-org.3172ff9da3":"What a published paper ships alongside its figures","use-case-research-org.8e632ceca6":"Attached","use-case-research-org.25e8aa3979":"four parts","use-case-research-org.80cd124891":"Data, code, environment and packet are produced together, per paper.","use-case-research-org.54861100a9":"Five links from the source dataset to the published figure","use-case-research-org.66c95e7fc7":"Sealed","use-case-research-org.5d4960a033":"five links","use-case-research-org.825586b275":"The study card a reader receives, in five rows","use-case-research-org.7f8d7aeedb":"five rows","use-case-research-org.23b016944b":"One packet, three readers","use-case-research-org.8377801e5c":"How a funder report, an ethics audit and a peer review are served","use-case-research-org.73c20ded5e":"Reported","use-case-research-org.38aa72a188":"three audiences","use-case-research-org.5c24c519b5":"The packet is the source. The three reports are views of it.","use-case-research-org.7cf98075db":"The same room, three boundaries","use-case-research-org.500680c378":"Where the decision layer, the packet store and the replay sit","use-case-research-org.1bb360f468":"Bounded","use-case-research-org.89ecc13361":"three postures","use-case-research-org.8fd8407d60":"Choosing a posture is choosing a boundary, not a different product.","use-case-research-org.02e3f43a87":"How you challenge a published figure","use-case-research-org.f19ada6792":"Four steps, three of which need nothing but the packet","use-case-research-org.da6818fb20":"Open to all","use-case-research-org.de60e7db09":"four steps","use-case-research-org.b2c15fdbb0":"Disagreement is a route on the paper, not a request to the authors.","use-case-research-org.c8c4690501":"Two runs, one account","use-case-research-org.e417512918":"The signed run and the replayed run, compared step by step","use-case-research-org.1bf3ec5bf8":"Matched","use-case-research-org.d3aa136486":"two differ, two do not","use-case-research-org.e207169b99":"The machine and the release are free to change. The recorded inputs and the figure are not.","use-case-research-org.69b9755bde":"The record behind figure 3","use-case-research-org.c6b2d79983":"The reviewer question, the record that answers it, and the trace behind both","use-case-research-org.9992b5f893":"Replayed","use-case-research-org.b8fcb6ce37":"four trace lines","use-case-research-org.e3a49f3ac7":"Nothing in the answer was assembled after the question was asked."}