Roadmap
Now means available to Founding Customers in the current Private Beta — not that it is nearly ready. Everything else is honestly labelled, including the things we have tried and cannot yet do well.
Updated 2 September 2026
Now
Available to Founding Customers in the current Private Beta. If it is in this list, we can show it to you in a live demo.
Evidence Programmes
Run a named subject — a supplier, a site, a contract — against a set of requirements, and get one finding per requirement with the evidence attached.
Supplier Compliance Review
A starting template of reviewer-written requirements, in the language a compliance reviewer already uses.
Your own requirements
Write requirements in your own words. They persist between runs, so the same question is asked the same way next quarter.
Semantic evidence discovery
Hybrid retrieval — keyword, semantic and section signals merged — so evidence is found when the document does not use your vocabulary.
Requirement → evidence → finding
Requirements persist between runs and every finding is kept on record, under the run that produced it.
Source-linked findings
Three classes of document are kept permanently separate: evidence DataStruct could support, sources it relied on but could not confirm, and material that is merely related.
Reviewer decisions
Confirm, reject, or mark for follow-up. Attributed to a named person and audit logged. The system records no verdict of its own.
Programme runs and history
Every run has its own identity, date and status. Open an earlier run to see what was found and what the reviewer decided then; decisions never carry over to a new run.
Requirement editing and archiving
Correct a requirement’s wording for future runs — earlier findings keep the wording they were evaluated against — or archive it so future runs skip it while its history stays.
OCR for scanned documents
Scanned PDFs are read rather than silently ingested as empty, and the deployment reports whether OCR is available.
Spreadsheet evidence with row locators
XLSX registers and trackers are ingested as first-class documents. Spreadsheet evidence names the sheet and the row, or the row range, the cited passage came from; where only the sheet is known, only the sheet is named.
Evidence Programme report
Export one run as a DOCX report or as XLSX, CSV or JSON: requirements, what DataStruct found, evidence states, reviewer decisions and sources, with the human-review boundary stated. PDF is not offered.
Document reports and dataset exports
Document reports in HTML, DOCX and JSON; extraction datasets in CSV, XLSX and JSON.
Access control and audit
Workspace roles with granular permissions, an audit log on sensitive actions, and GDPR export and erasure.
In progress
Being built and measured now. Some of it is in the product already but not yet something we would ask you to depend on.
Evidence support accuracy
Improving how confidently DataStruct separates supporting evidence from related material. We measure this against a fixed internal benchmark rather than by impression.
Broader evidence templates in the app
More programme templates and review modes exist behind the interface than the screen currently exposes.
First-run and programme experience
Shortening the distance between signing in for the first time and seeing a real finding.
Next
Committed direction. Shaped by what Founding Customers tell us matters.
ISO and quality audit preparation
Programme templates aimed at audit evidence packs rather than supplier review.
Cyber supplier assessment
Requirement sets for security questionnaires and supplier assurance.
ESG supplier due diligence
Disclosure and due-diligence requirement sets over supplier documentation.
Team review workflows
Assignment, queues and sign-off across a reviewing team rather than one reviewer.
Improved reporting
Richer evidence packs, including a print-ready format.
Customer-managed AI
Bring your own model credentials, and choose where inference runs.
Enterprise identity
Single sign-on, once it is proven in production rather than merely implemented.
Retention controls
Customer-configurable retention windows and self-serve hard deletion.
Exploring
Problems we think are real and have not solved. Listed because pretending otherwise would be the easy thing to do.
Contradiction detection
Surfacing when two documents disagree. Our own twelve-case conflict benchmark surfaced none of them, so this is research, not a feature.
Document supersession and version lineage
Preferring the current version of a superseded record. Measured at 0% on three of four internal corpora.
Investigation mode
Following a question across a corpus as a line of enquiry rather than a single answer.
Evidence change monitoring
Telling you when evidence you relied on has changed or expired.
Private deployment options
Isolated or customer-hosted deployment for regulated environments.
Our roadmap reflects DataStruct's current product direction and may evolve as we learn from customers and validate new workflows. We do not publish dates we cannot guarantee.