Kuppola transforms unstructured research spreadsheets into structured, queryable science — stored in infrastructure you own. No ELN required. No vendor lock-in. Cancel anytime and keep everything.
Excel is flexible enough for scientists to use, but too unstructured to be queryable, reproducible, or auditable at scale.
Structuring Excel data requires someone who understands both the biology and data modeling. That person is rare. That time is expensive.
Scientific logic buried in column names, compound IDs, and cross-sheet patterns is invisible to generic tools — but critical to reproducible science.
When structured systems don't reflect how experiments actually run, scientists revert to Excel. The investment goes to waste.
Research labs generate data in Excel because it is flexible and requires no setup. But Excel is not queryable, not relational, not reproducible, and not auditable.
The problem isn't that scientists don't want structure. It's that the cost of creating that structure is too high. Migrating even a single study to any structured system requires months of manual work — and the moment the schema doesn't fit the scientist's mental model, adoption collapses.
Kuppola eliminates that cost entirely. The AI understands your data the way a domain expert would — and the structured result lives in infrastructure you already own.
Kuppola infers structure from your spreadsheets, builds a queryable registry in your own database, and lets you validate everything before committing a single record.
Trained on biological domain knowledge, Kuppola understands that S001_M001_Liver is a compound ID encoding a study, mouse, and tissue type. Generic ETL tools do not.
Kuppola is a complete standalone research data registry. Structured entities, queryable relationships, version history, and audit trails — all in your PostgreSQL database. No ELN required.
Test the generated schema against your actual data before committing anything. Run natural language queries, see how your rows map, and refine iteratively — without touching production.
Scientists upload an Excel file and get a structured, queryable registry in minutes. No back-and-forth with IT. No data engineers. No months-long implementation projects.
A multi-agent AI pipeline that handles the full translation — understanding your data, classifying your domain, mapping relationships, and generating a validated schema.
Parse Excel structure, decode column intent, and detect compound IDs and cross-sheet references.
Identify the research area and experimental design to apply the right biological knowledge.
Infer the entity hierarchy and map fields, relationships, and validation rules from your data.
Generate a validated schema, simulate it against your data, refine, then commit to your registry.
Kuppola is designed so that you could cancel your license tomorrow and keep every record, every schema, and every relationship — in open formats, in infrastructure you control.
All data is stored in PostgreSQL and S3 — industry-standard, open, and readable by any tool. Kuppola adds AI-powered structure on top, but the underlying data is never proprietary.
When a license expires, Kuppola enters read-only mode — not deletion mode. You can query, export, and access everything. Nightly Parquet snapshots to your S3 bucket continue regardless of plan state.
Deployment options
Fastest to start. Kuppola manages infrastructure. Full data export available at any time.
Kuppola deploys into your VPC. All data stays in your RDS and S3. You supply the AWS account and Anthropic API key.
Run everything including AI inference in your own environment. Air-gapped deployments supported.
Kuppola is a complete standalone registry. If you already use an ELN, Kuppola can sync structured data to it — but you never need one to get full value.
Sync Kuppola schemas and entities to your Benchling registry. Kuppola remains the source of truth — Benchling is a sync target. Ideal for teams already invested in Benchling workflows.
Export structured Kuppola data to Revvity Signals. Same adapter pattern — Kuppola schema is always the canonical source, Signals is the downstream target.
Kuppola's integration layer is a pluggable adapter interface. If your ELN supports file-based import (CSV, JSON), Kuppola can export to it today. Native adapters for more platforms are in the roadmap.
Every product decision at Kuppola comes back to these.
The AI understands biology, not just data types. Experimental context, domain conventions, and scientific relationships are first-class concerns — not afterthoughts.
We believe you should be able to walk away from any software vendor and keep everything you built with it. Kuppola is designed so that decision is always yours to make.
Data you can query is data you can use. Every hour Kuppola saves on structuring data is an hour a scientist can spend analyzing it, replicating it, or building on it.
Beta launching Q3 2025. We're running custom pilots — bring your messiest Excel file and we'll show you what Kuppola does with it. No ELN required to get started.
Request a Pilot →