OneSchema alternative — explainable, self-hosted CSV import
Ce contenu n’est pas encore disponible dans votre langue.
Mildport is a self-hosted OneSchema alternative. It gives your users the same embeddable spreadsheet-import experience — upload, map, validate, fix, submit — but the engine runs in your infrastructure, and every mapping decision is explainable rather than opaque.
Teams comparing OneSchema are usually optimizing for one of two things: clean, validated rows reaching their backend, and control over where that data is processed. Mildport is built to give you both without a cloud round-trip.
OneSchema vs Mildport at a glance
Section titled “OneSchema vs Mildport at a glance”| OneSchema (cloud) | Mildport (self-hosted) | |
|---|---|---|
| Hosting | Vendor cloud | Your infrastructure (Docker Compose or Helm) |
| Data path | Through vendor cloud | Stays in your VPC |
| Validation | Rules-based, cloud-run | Typed contract + deterministic matcher, run in your env |
| Mapping transparency | Vendor-managed | Visible scores + replayable, event-sourced pipeline |
| Schema setup | Templates built by hand in the vendor dashboard | No-code editor embeddable in your own admin, or drafted from your API spec |
| AI assistance | Vendor AI | Optional evidence-gated judge, bring your own model, off by default |
| Data cleanup | Vendor AI transforms | Plain-language cleanups, previewed before apply, replayed free on similar files |
| Inputs | CSV / spreadsheets | CSV, TSV, XLSX, ODS, JSON, XML, PDF tables, OCR’d scans |
| Delivery | Vendor APIs | License-keyed REST API + webhooks with an audit trail |
| Import evidence | Vendor logs | Decision log per import + signed receipts a third party can verify offline |
| Tool or protocol? | A tool: the import runs, the trail stays with the vendor | A protocol: receipt, ambiguity ledger, changeset and a portable import contract |
Why teams choose Mildport
Section titled “Why teams choose Mildport”- Explainable validation, not a black box. Every header → field decision carries a score you can see, and the normalization pipeline is event-sourced so you can replay exactly what happened to a row. See Capture-first imports.
- Self-hosted by design. Files never leave your VPC; the engine never phones home. This is the path for regulated data and security-reviewed deployments.
- Beyond clean CSVs. Real customer files include messy spreadsheets, PDF tables and scanned documents. Mildport decodes them through dedicated sidecars, not just well-formed CSV.
- Your AI, your call. The deterministic matcher does the predictable work; the optional judge only touches the uncertain tail, with your model and a visible rationale.
- Schema from your API, not from a dashboard. OneSchema’s template editor makes you rebuild your data model by hand. Mildport drafts the import schema from the API spec you already have — review, publish, and every embed updates without a release. The editor itself is embeddable, so your customers can manage their own fields inside your product, versioned with one-click rollback.
When OneSchema is the better fit
Section titled “When OneSchema is the better fit”If your inputs are reliably clean CSVs, you want a fully managed cloud product, and routing data through a vendor cloud is fine for your compliance posture, a hosted importer is less to operate. Mildport is for when self-hosting and explainability are requirements, not nice-to-haves.
Getting started
Section titled “Getting started”npm install @capitality-io/mildport-widgetDefine the element, point it at your self-hosted engine, and listen for clean rows:
el.addEventListener('import-applied', e => sync(e.detail));Full walkthrough in the Quickstart; run the engine yourself with the self-hosting overview.
Is Mildport a self-hosted alternative to OneSchema? Yes — an embeddable importer like OneSchema, but proprietary self-hosted software with the engine running in your own infrastructure.
How does Mildport handle validation compared with OneSchema? It validates against a typed contract and maps columns with a deterministic, explainable matcher — visible scores and a replayable pipeline — with an optional AI judge for the uncertain tail only.
Can I run Mildport without sending data to a vendor cloud? Yes. It runs in your environment (Docker Compose or Kubernetes via Helm), licenses verify offline, and files stay in your VPC end to end.
See it on your own files at mildport.com, or read What is Mildport?