BLK PHX LABS phoenix mark
BLK PHX LABS· Biomanufacturing intelligence

The operating system
for biological manufacturing.

Cell and gene therapy is manufactured one patient at a time, on paper batch records and the memory of a few experienced operators. LabOps.AI turns that process knowledge into software that QA teams can review, audit, and trust.

  • Cell & Gene Therapy
  • AI-Native
  • GMP Manufacturing
  • Human-in-the-Loop QA
Get in touch See what's built

The Problem

CAR-T and other autologous therapies are made one batch per patient. A single batch takes over 200 labor hours across manufacturing, QA, QC, and logistics, and labor is roughly 71% of batch production cost.1 Manufacturing and release alone frequently take two to four weeks, with QC testing and release typically two to three.2 Between 4% and 7% of patients never receive their CAR-T product because manufacturing failed.3

  • 200+ hrs

    Labor per CAR-T batch, across manufacturing, QA, QC, and logistics1

  • ~71%

    Share of batch production cost that is labor1

  • 4–7%

    Patients who never receive their CAR-T product due to manufacturing failure3

Autologous CAR-T vein-to-vein journey Seven stages: apheresis, shipping, activation and transduction, expansion, QC release, shipping, and infusion. Manufacturing and release often take two to four weeks, and QC release typically two to three. LabOps.AI works across expansion and QC release: deviations, batch record review, and release readiness. Where LabOps.AI works Manufacturing + release: often 2–4 wks Apheresis Ship Activation /Transduction Expansion QC releasetypically 2–3 wks Ship Infusion Autologous CAR-T vein-to-vein journey Seven stages from apheresis to infusion. Manufacturing and release often take two to four weeks, and QC release typically two to three. LabOps.AI works across expansion and QC release. Where LabOps.AI works Apheresis Ship Activation / TransductionMfg + release: often 2–4 wks Expansion QC releasetypically 2–3 wks Ship Infusion
Where LabOps.AI works: deviations, batch record review, release readiness. Illustrative; durations vary by product.2

The manufacturing intelligence needed to run these processes safely and at scale does not exist as software. It lives in the heads of scientists and operators who are retiring, burning out, or moving on.

No institutional memory

Critical process knowledge is undocumented or siloed in individuals. When operators leave, the knowledge leaves with them.

Deviation blindness

Deviations are investigated days or weeks after the fact, often surfacing only at batch record review or release testing, when the patient's dose is already committed.

Regulatory drag

GMP documentation is manual, error-prone, and consumes operator time that should go to process improvement. Review-by-exception is the goal; paper makes it impossible.

Scale barrier

Processes validated at small scale break at commercial scale because no system captures what actually drives product quality.

Why Now

Paper is finally leaving the floor.

CGT CDMOs are moving to purpose-built MES platforms.4 That creates structured process data for the first time, which an intelligence layer needs.

FDA is writing the AI rulebook.

FDA's January 2025 draft guidance set out a risk-based credibility framework for AI used to support regulatory decisions on drugs and biologics,5 and CDER's 2026 guidance agenda includes AI/ML in pharmaceutical manufacturing.6

Unvalidated AI is now an enforcement issue.

In April 2026 FDA issued a warning letter to a drug manufacturer that had used AI agents to write its specifications, procedures, and production records, citing failures of Quality Unit oversight.78 The market needs AI that is validated, auditable, and keeps humans in the loop. That is the design premise of LabOps.AI.

The workforce can't scale fast enough.

Experienced CGT manufacturing talent is scarce, so the knowledge needs to be captured in systems rather than individuals.

What We're Building

LabOps.AI is an AI-native intelligence platform that captures, structures, and activates the process knowledge inside a biomanufacturing facility, connecting batch records, equipment data, deviations, and regulatory history into a unified operational brain. Every output is decision support: QA reviews and approves.

Built today

Status · Built

Literature intelligence pipeline

RAG pipeline over CGT manufacturing and regulatory literature, with automated daily and weekly briefings.

Watch today's briefing →
Status · Built

Operations dashboard

FastAPI and React dashboard for batch parameter tracking, a deviation feed, and instrument-export ingestion.

Status · Built

Audit trail

User authentication and audit logging designed against 21 CFR Part 11 requirements. Compliance itself belongs to a validated system at the customer site.

Status · Built

Automated test suite

Automated tests cover the API, authentication, audit coverage, data ingestion, and instrument adapters.

Roadmap

Planned capabilities. Not shipped.

Roadmap · CQA / CPP Extraction

Process Intelligence Engine

Identifies Critical Quality Attributes and Critical Process Parameters from unstructured batch records, SOPs, and deviation reports.

Roadmap · Early Deviation Signals

Deviation Detection

Flags process anomalies early enough for operators and QA to intervene, with every flag reviewed and dispositioned by a human.

Roadmap · Knowledge Architecture

Institutional Memory

Structures tacit operator knowledge into retrievable, auditable records that survive personnel turnover and scale across sites.

Roadmap · GMP Documentation

Regulatory Readiness

Structured, audit-ready documentation workflows that shorten batch record review. QA reviews and approves every record.

Landscape

LabOps.AI sits beside the systems that run a GMP floor, not in place of them. Execution systems and automated hardware generate the data; LabOps.AI reasons over it so QA and operations can act sooner.

Category
Examples
LabOps.AI difference
Pharma MES / eBR
Apprentice.io Tempo, in use at Minaris for CGT4
MES executes the batch. LabOps.AI reasons over the data the batch produces. Complementary, not a replacement.
AI process modeling
BioCurie: mechanistic + ML process modeling for CGT and RNA manufacturing9
Upstream process development, vs. LabOps.AI's GMP operations and QA layer.
Automation hardware
Closed, automated cell-processing systems that run manufacturing steps inside a single instrument
Hardware vendors capture their own data. LabOps.AI is vendor-neutral across sites and equipment.

Founder

Built by a scientist-operator: a former clinical lab director with cGMP biobanking experience, who now runs high-volume operations inside one of the most automated logistics networks in the world.

Fabian Alvarez-Primo, PhD

Founder · LabOps.AI
  • PhD, Biomedical Materials Science & Engineering, UTEP
  • Former CLIA/COLA Clinical Laboratory Director (LC-MS, NGS)
  • Postdoctoral research, TTUHSC / Southwest Brain Bank: iPSC reprogramming, neuropsychiatric transcriptomics, cGMP biobanking
  • Operations Manager II, Amazon
  • 8 peer-reviewed publications
LinkedIn ↗ GitHub ↗ Google Scholar ↗
BLK PHX LABS

Get in Touch

Building in CGT manufacturing, QA, or process development? I'd like to hear what breaks on your floor.

fabian@blkphxlabs.com
Email Fabian Watch today's briefing

Investors and design partners: same address.

Sources

  1. “CAR-T cell manufacturing landscape: Lessons from the past decade and considerations for early clinical development.” Molecular Therapy: Methods & Clinical Development 32(2), 2024. PMC11088187 · doi:10.1016/j.omtm.2024.101250
  2. “Promises and challenges of a decentralized CAR T-cell manufacturing model.” Frontiers in Transplantation, 2023. PMC11235344 · doi:10.3389/frtra.2023.1238535
  3. “CAR-T cell manufacturing: Major process parameters and next-generation strategies.” Journal of Experimental Medicine, 2024. PMC10791545 · doi:10.1084/jem.20230903
  4. Outsourced Pharma, “Minaris Regenerative Medicine Selects Apprentice.io's Tempo Manufacturing Execution System To Scale Commercial Production,” Sept 2023. outsourcedpharma.com
  5. U.S. FDA, “Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products,” draft guidance, January 2025. fda.gov
  6. IntuitionLabs, “FDA AI Guidance 2026: Drug Manufacturing & Digital Health.” intuitionlabs.ai (secondary summary of the CDER 2026 guidance agenda)
  7. U.S. FDA, Warning Letter to Purolea Cosmetics Lab, MARCS-CMS 722591, April 2, 2026. fda.gov
  8. DLA Piper, “FDA warning letter highlights risks of using AI in drug manufacturing,” April 21, 2026. dlapiper.com
  9. Cure, “BioCurie leads AI-powered biomanufacturing revolution.” wewillcure.com