It's time to build bio: introducing the Genome API

For the last decade, the consumer genome has been locked behind apps, PDFs, and proprietary pipelines. Companies sequenced genomes, packaged the results into closed reports, and treated the underlying data as an asset to protect and monetise rather than a foundation to build on. Builders were left with whatever they were given.

This was a choice, not a constraint. We can choose differently.

Today, The Genome Computer Company is opening early access to the Genome API.

With a single HTTPS endpoint, partners can order sequencing, track samples through the lab, and receive a complete .genome/1.0 bundle and VCF as soon as results are ready.

What the API does

The Genome API exposes the full stack behind one call: collection, sequencing, variant calling, annotation, interpretation, and AI-native formatting. A kit ships to your customer. The sample returns to our accredited labs. The full pipeline runs. What lands in your application is the genome itself, not a report.

Until now, building on genomic data meant choosing between three bad options:

• Scraping consumer reports.
• Integrating vendor SDKs that hide the raw genome.
• Standing up your own lab.

All three trade away direct access to the underlying data.

Whole genome 30x, whole genome 100x, SNP genotyping, and imputed SNP panels are all orderable through the same endpoint. The same vertically integrated pipeline we run for direct customers is now available programmatically, with no lab negotiations and no bespoke integrations. You receive the data, you control the interpretation, and the format you receive it in is open and durable.

The same API also accepts existing VCFs. Partners whose users arrive with a file from a legacy provider, a clinical run, or a research dataset can submit it to a conversion endpoint and receive a fully-annotated .genomebundle in hours rather than weeks. Same output format, same downstream surface — fresh sequencing and existing files converge on a single primitive.

Clinical-grade infrastructure

Sequencing is performed in CLIA-, CAP-, and NATA-accredited, ISO 15189-certified laboratories. Data handling aligns with HIPAA controls.

Every bundle includes a manifest that pins:

• Pipeline version.
• Reference build (GRCh38).
• Guideline snapshots used at interpretation time.

Every result is auditable. Every claim traces back to a versioned source.

A Public Benefit Corporation

Genome Computer operates under Humankind, a Delaware Public Benefit Corporation. A PBC restructures corporate purpose: a traditional corporation is obligated to maximise shareholder value, while a PBC is required to balance shareholder interests with a defined public benefit and the interests of affected stakeholders.

Humankind’s charter is explicit:

To advance human understanding through the collective analysis of genomic and phenotypic data, while ensuring that identifiable individual data remains under individual control and is not sold or licensed for commercial use.

The genomics industry has repeatedly failed at this boundary. User data has been sold to pharmaceutical companies. Genetic databases have been handed to law enforcement without consent. Privacy terms have been rewritten retroactively when revenue ran short. When genomes are the most valuable asset on the balance sheet, monetising them becomes the default outcome.

A PBC changes that constraint. Selling identifiable genetic data would violate fiduciary duty rather than fulfil it, and that commitment is embedded in legal documents that survive leadership transitions, financial crises, and acquisitions — the exact moments at which privacy promises have historically failed.

For an API that other companies build on top of, this matters more than for a direct-to-consumer service. Two operating commitments follow from the charter:

Open, durable formats. .genome/1.0 is Apache-licensed and will remain so. The format is designed to outlive the company, and a bundle a partner ships today will still be parseable in 2030 regardless of what happens to us.

Transparent science. Interpretation logic is embedded in the bundle, not hidden behind an API. Results can be audited, recomputed, and challenged.

Why .genome matters

Every order returns two artifacts: a VCF for compatibility with existing bioinformatics pipelines, and a .genome bundle for everything else.

.genome/1.0 is designed for AI and software systems rather than human specialists. The gains over reading the equivalent annotated VCF are measurable: an LLM uses 3–7× fewer tokens of context, and produces 10–20× better interpretive accuracy on standard genomic-reasoning evaluations. Four properties of the format account for this.

It is AI-readable by construction. Variant data, interpretations, and rules are explicitly structured and versioned. There are no ambiguous encodings and no inferred meanings.

It is deterministic. PharmGKB metabolizer status, ClinVar pathogenicity, and ACMG actionability are computed at pipeline time against pinned guideline versions, so there is no runtime guessing.

It is queryable in milliseconds. Gene-level questions reduce to simple columnar reads, with no indexing complexity or specialised tooling required.

It is versioned end-to-end. Each bundle pins guideline versions and timestamps, so a result re-interpreted next year produces a clean diff against the version generated this year.

What you can build

Things that should already exist:

• An agent that blocks the wrong drug before it is prescribed.
• A nutrition system that adjusts a user’s diet based on how they actually metabolise food.
• A longevity coach grounded in the user’s own genome rather than population averages.
• A research platform where consented genomes flow into cohorts in minutes instead of months.
• An assistant that grounds its advice in the user’s biology, not in self-reported answers and survey data.

Each of these requires direct application-layer access to genomic data formatted for a software reader. Each one is now possible.

Request access

The Genome API is in private beta with a small group of partners. We are looking for teams shipping real products this year — clinical, consumer, research, or agent-based — that want to ship faster and get feedback into the roadmap.

Request access at genome.computer/api or write to api@genome.computer.

The genome is a building block. What are you going to build with it?