Auditable decision infrastructure for genome editing programs
Helix Studio — System of Record for Genome Editing Decisions
Deterministic design, policy gated risk assessment, and replayable evidence bundles for CRISPR and Prime Editing decisions.
Need governed decisions for a program? Go Studio. Need to verify an export? Install the CLI.
No phenotype promises. No AI mysticism. Just decisions you can defend in review, QA, and procurement.
Helix is the first Genome IDE built to function as a system of record for genome editing decisions.
Live system snapshot
Intent
Targets · Edit intents
Policy
Evidence gates · Constraints
Evidence
Bundles · Hashes · Exports
Products
Two products. One deterministic engine.
Helix exists to turn genome editing into a decision workflow: generate evidence, enforce a policy gate, and export a reviewable bundle. Helix Studio is the Genome IDE (GUI + CLI) for designing and defending decisions. Helix CLI is the headless toolchain for enforcing the same standards at scale.
Helix exists to block experiments that should never be run.
For scientists and platform leads designing and defending experiments.
Helix Studio (GUI + CLI)
- Interactive workspace for designing edits and reviewing evidence.
- Policy gates, deterministic exports, and audit-ready artifacts.
- Bundles the Helix CLI for headless replay and verification.
For infra teams enforcing standards at scale.
Helix CLI (headless)
- Run evidence generation in CI, HPC, or batch automation lanes.
- Deterministic verification, diffing, and bundle tooling.
- SDK-friendly outputs for integration and automation.
Governance guarantees
These are not features. These are things that cannot go wrong.
When millions of dollars and months of wet lab time are at stake.
Determinism
Same inputs. Same outputs. Or a signed explanation for why not.
Policy enforcement
If required evidence is missing, Helix refuses to produce a decision and tells you what is required.
Assumption visibility
Every outcome is tied to an explicit regime, calibration domain, and constraint set — no invisible defaults.
Replayability
Any reviewer can rerun the computation byte for byte and verify exported artifacts.
One buyer facing workflow.
Calm. Inevitable. Reviewable — built for QA, platform leads, and regulated teams.
Design intent defined
Target class and edit intent declared.
Start from what you are trying to change, not which tool you happened to open.
Constraints declared
Hard bounds and allowed regimes set.
Constraint sets make assumptions explicit and keep teams from drifting into undefined territory.
Policy selected
Evidence requirements chosen up front.
Policies define what must be present before a decision is allowed to exist.
Evidence computed
Computational evidence generated under declared assumptions.
Outcome simulation becomes decision evidence; predicted outcomes become implied distributions (assumption bound).
Decision gated
Policy enforced before export.
If required evidence is missing, Helix refuses the decision and tells you what to provide.
Artifact exported
Evidence bundle packaged for review.
Export replayable PDFs and machine-readable bundles with hashes, versions, and provenance.
Decision reviewed later
Rerun and verify byte for byte.
Auditors can reproduce results and trace every assumption back to its source regime.
The unit of value
The Unit of Value: An Edit Program
Helix prices, governs, and audits at this level — without needing to say “enterprise”.
An Edit Program is
- A target class or therapeutic objective
- A bounded set of edit intents
- A declared policy and calibration regime
- A persistent decision history
What Helix makes durable
- Decision evidence bundles with hashes and provenance
- Risk enumeration gated by policy (including off-targets)
- Implied distributions (assumption bound) instead of “predictions”
- Exports ready for review by QA, platform, and procurement
Boundary conditions
What Helix Refuses to Do
Helix does not predict biological truth. It produces reviewable, reproducible computational evidence and enforces explicit boundaries on what is and is not known.
Ship uncertainty as certainty
Helix does not turn weak evidence into definitive claims. Outputs stay explicitly bounded by assumptions.
Hide policy or calibration regimes
No silent defaults. Decisions are always associated with an explicit policy and calibration domain.
Emit decisions without required evidence
If required inputs are missing, Helix blocks the decision and lists the missing evidence.
Make results non-reproducible
Every export carries IDs, versions, and hashes so reviewers can replay and verify results later.
Proof on page
Methods, benchmarks, and security receipts.
Big claims demand proof. Here are the artifacts teams review before every pilot, surfaced directly on the homepage.
Downloadable artifact with footer metadata.
Sample design report (PDF)
Pull the actual single page evidence bundle we hand pilots. Footer carries build + model IDs, deterministic hash, timestamp, and generator signature so reviewers can cite it.
Download design sample →Downloadable artifact for sequencing proof.
Sample NGS outcome report (PDF)
Inspect a real alignment/NGS QC export for the chr20 multiplex pilot (1.5M reads). Runtime stats and outcome tables are stamped with the same footer metadata.
Download NGS sample →Determinism checks ship in CLI + Tools outputs.
Deterministic Instruments
Design, evidence generation, and QC exports are deterministic. Each carries a schema version + SHA-256 hash so `helix-cli verify --deterministic` can replay and diff results verbatim.
Reproducibility notes →Tracking throughput per hardware class.
Benchmarked on Real Runs
Internal pilot: chr20 multiplex amplicon set (1.5M reads). 41 s on a single A100, 6.2 min CPU fallback. Results logged against NIST GIAB to keep reality checks public.
Pilot benchmark snapshot →Key deliverable in every pilot.
Validation Workflow
Each pilot compares Helix evidence bundles against wet lab CRISPR + Prime Editing outcomes. Deliverables include agreement tables, QC deltas, and automation recipes.
Pilot deliverables →Docs + commands published publicly.
Security & Supply Chain
Telemetry ships disabled, offline mode is first-class, installers are signed, and every drop has SBOM + checksum with copy/paste verification commands.
Security posture →Early Access / Beta Cohort
Built for defensible decisions.
Private beta is open now. General availability coming soon. If you need reproducible evidence, explicit policy gates, and an audit trail for genome editing decisions, let’s talk.