☰

LabGenius Therapeutics

Closing the last mile in an automated discovery lab

Industry
TechBio / Antibody Discovery
Headquarters
London, United Kingdom
Public information as of
January 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of LabGenius Therapeutics's published strategy and is not endorsed by, or produced in cooperation with, LabGenius Therapeutics. Company website

Strategic priorities

LabGenius Therapeutics raised £35 million in Series B funding in May 2024, led by M Ventures (Merck KGaA), to advance multispecific antibodies for solid tumors through its EVA™ platform — a ML-driven robotic system that runs closed-loop Design-Build-Test-Learn (DBTL) cycles to optimize antibody candidates. The company employs over 60 scientists and engineers at its London facility and runs a second major phase of its Sanofi partnership (NANOBODY® optimization) that began in late 2025.

The EVA™ platform uses Multi-Objective Bayesian Optimization (MOBO) to co-optimize multiple antibody properties simultaneously. The computational models depend entirely on the quality and velocity of experimental data returned from the lab, which means the DBTL cycle is only as fast as the slowest data step feeding it. LabGenius has publicly disclosed that roughly 30 percent of laboratory actions still involve manual sample preparation, assay steps, or manual data ingestion into Benchling.

The company is pre-clinical but Series B milestone achievement requires a credible path to clinical operations. This means the flexible R&D data environment — acceptable for discovery — has to become a validated, audit-ready GxP data chain before any internal asset enters a Phase I trial. The same infrastructure must also support the Sanofi collaboration, which adds partner data governance and the requirement to reconfigure laboratory workflows rapidly for different project demands.

Challenges we see

  • Operations Integration

    Automating the last 30 percent of laboratory actions

    Despite extensive automation through EVA™, approximately 30 percent of laboratory actions remain manual, involving sample preparation, specialized assay steps, and manual data ingestion into systems like Benchling, according to Q&A with Head of Automation Mohammad Akhlaq.

    Where manual steps sit inside an otherwise automated DBTL cycle, the cycle time is set by the slowest step. Automating the remaining sample preparation and data capture steps closes that gap without changing what the automated equipment already does.

  • Digital Manufacturing

    Adding new lab hardware without re-engineering the integration

    LabGenius operates in a high-growth environment requiring constant addition of new robotic arms, liquid handlers, and specialized sensors from diverse vendors using different communication protocols.

    When each new piece of equipment requires a custom integration project, the lab layout becomes rigid and the cost of reconfiguration scales with the number of vendor relationships rather than with the scientific value of the hardware being added.

  • Compliance Regulatory

    Building validated data chains as assets move toward the clinic

    LabGenius is pre-clinical and must transition the flexible, data-rich R&D environment to validated GxP-compliant systems as it advances wholly-owned TCE and ADC assets toward Phase I.

    A clinical data submission requires each piece of evidence to carry an unbroken chain from the instrument that produced it to the submission it appears in. Building that chain after an asset is ready for filing is a retracing; building it now is an enabler.

  • Digital Integration

    Keeping ML models trained on complete rather than fragmentary data

    The efficiency of MOBO-driven optimization cycles depends entirely on the quality, velocity, and contextualization of experimental data returned from the lab to cloud-based ML environments on Google Cloud Platform.

    When data leaves an instrument and arrives at the ML environment without the experimental context that explains it, the model is trained on values rather than on understanding — and the optimization surfaces patterns in measurement rather than in biology.

  • Digital Operations

    Governing access to proprietary ML models and partner data across locations

    LabGenius operates a hybrid model with remote data scientists accessing sensitive IP, proprietary ML models, and DNA sequences on GCP while servicing global partners including Sanofi. NIS2 compliance requirements for life sciences entities are approaching.

    The attack surface for a company whose core IP lives in cloud-hosted models and whose collaborators are global is larger than the physical perimeter, which means perimeter-based access controls leave the models and sequences exposed to whichever location a credential is used from.

Opportunities, by urgency and business impact

Each bubble is one opportunity, numbered to match the list below. Further right means it bites sooner; higher means a bigger effect on the business. A bigger bubble means a bigger implementation effort.

Source: A4BEE analysis of public sources

  1. Automating the last 30 percent of laboratory data capture

    Roughly 30 percent of laboratory actions remain manual, including sample preparation and data ingestion into Benchling, creating bottlenecks in the high-throughput DBTL cycle and introducing transcription errors into ML training data.

    Deploy custom IoT connectors and retrofit legacy lab equipment to enable automatic data acquisition, eliminating manual transcription and accelerating the ML training feedback loop.

    • Q&A with Mohammad Akhlaq, Head of Automation, LabGenius (Medium)
    • LabGenius Strategic Account Intelligence Report

  2. Using modular standards to make hardware add-or-replace rather than re-engineer

    Proprietary vendor ecosystems create custom integration projects and rigid lab layouts that prevent the rapid reconfiguration needed for diverse partnership projects and scaling demands.

    Introduce MTP (Module Type Package) standard to create a Plug and Produce lab environment where hardware can be added or replaced without disrupting the central orchestration layer.

    • LabGenius Strategic Account Intelligence Report

  3. Building audit-ready GxP data chains before the first clinical filing

    The flexible R&D data environment is incompatible with the validated, locked-down requirements for clinical data, creating a compliance cliff as LabGenius advances assets toward the clinic.

    Implement GAMP5-compliant cloud governance and data integrity frameworks (ALCOA Plus principles) so EVA platform data chains can withstand FDA and EMA regulatory audits.

    • LabGenius Strategic Account Intelligence Report
    • PMC12688275 — Research publication on multispecific antibody developability

  4. Predicting manufacturability before committing to scale-up experiments

    Multispecific antibodies that perform well in discovery may prove impossible or prohibitively expensive to manufacture at scale; classical in silico developability screens have limited predictive power.

    Build Digital Twin simulations of bioprocesses to predict antibody behavior in large-scale fermenters and purification units before committing to expensive wet-lab scale-up experiments.

    • PMC12688275 — Research publication on multispecific antibody developability

  5. Extending identity verification to every access point for cloud-hosted IP

    Remote data scientists accessing sensitive IP on GCP combined with partner data sharing requirements create multiple potential entry points for credential compromise and lateral movement.

    Implement Zero Trust security architecture with continuous identity verification and IEC 62443-aligned controls protecting both cloud ML assets and laboratory automation systems.

    • LabGenius job postings
    • LabGenius Strategic Account Intelligence Report

What we'd propose

  • Digital Lab

    Lab Digitalization and MTP Implementation

    Deploy Module Type Package (MTP) standards to transform LabGenius' laboratory into a reconfigurable Plug and Produce environment enabling rapid scaling and vendor-agnostic hardware integration.

    • MTP-based equipment integration

      Universal driver for lab hardware

      Implement VDI/VDE/NAMUR 2658-compliant interfaces enabling any MTP-ready analyser, robotic arm, or liquid handler to connect to the central orchestration layer without custom coding for each new piece of equipment.

    • Flexible lab architecture design

      Modular reconfiguration capability

      Design lab layouts as modular units that can be physically and digitally reorganized to accommodate different Sanofi projects or internal pipeline requirements within days rather than weeks.

    • OPC UA connectivity backbone

      Standardized machine communication

      Establish a secure OPC UA (Open Platform Communications Unified Architecture) backbone connecting all lab equipment to ensure real-time data flow from sensors to the EVA ML platform without protocol translation bottlenecks.

    • New hardware connects without a custom integration project, so the lab scales with scientific demand rather than with engineering capacity.
    • The Sanofi partnership benefits from rapid workflow reconfiguration between projects without restructuring the automation layer.
    • The same architecture is reusable as the London facility expands.

    We've done this before

    See how we do Digital Lab
  • Enterprise AI

    Industrial Data Platform for ML Training

    Build an ontology-based data platform with automatic pipelines to ensure 100 percent reliable data flow from high-throughput assays to Benchling and the EVA ML environment on GCP.

    • Automated data ingestion from instruments

      Zero-touch sensor to cloud

      Deploy IoT connectors and custom drivers to automatically capture data from all analytical instruments, eliminating manual transcription and ensuring real-time availability for ML model training.

    • Benchling integration

      LIMS connectivity with bidirectional pipelines

      Build bi-directional data pipelines between laboratory equipment and Benchling to maintain a single source of truth for all experimental metadata and results.

    • Data contextualization engine

      Ontology-based semantic layer

      Implement semantic data models that automatically tag and contextualize raw sensor readings with experimental parameters, ensuring ML algorithms receive properly annotated training data rather than values without provenance.

    • The MOBO model trains on complete experimental records, not on data selected by manual transcription.
    • A single source of truth in Benchling means the EVA platform and the scientists working from it see the same data.
    • Instrument-to-cloud pipelines run continuously, so the DBTL cycle runs at the speed of the biology rather than at the speed of documentation.

    We've done this before

    See how we do Enterprise AI
  • Digital Lab

    GxP-Ready Cloud Governance Framework

    Implement GAMP5-compliant cloud migration and data integrity frameworks ensuring the EVA platform data can withstand FDA and EMA regulatory audits as LabGenius advances assets toward clinical development.

    • GAMP5 compliance architecture

      Validated cloud infrastructure

      Design and deploy cloud environments on GCP that meet pharmaceutical regulatory requirements with full audit trails, access controls, and validated change management procedures.

    • Data integrity framework

      ALCOA Plus compliance engine

      Implement automated enforcement of Attributable, Legible, Contemporaneous, Original, and Accurate data principles across all EVA platform data streams, from instrument output to long-term retention.

    • Regulatory documentation package

      Audit-ready validation artifacts

      Deliver IQ/OQ/PQ documentation, traceability matrices, and computer system validation (CSV) records required for regulatory submissions to FDA and EMA.

    • A credible GxP data chain is a prerequisite for Series B milestone achievement and for any clinical trial application.
    • Building the validated infrastructure now avoids a retrofit after an asset is ready for filing.
    • The framework is reusable across every future regulatory submission from the London site.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    Digital Twin for Bioprocess Simulation

    Create digital simulations of antibody manufacturing processes to predict developability and manufacturability before committing to expensive wet-lab scale-up experiments.

    • Bioprocess digital twin

      Virtual fermentation modeling

      Build physics-informed models of large-scale bioreactor behavior that simulate how multispecific antibody candidates will perform during manufacturing scale-up, catching developability issues before they reach the pilot.

    • Developability prediction engine

      In silico screening enhancement

      Augment existing LabGenius developability screens with ML-enhanced models that predict aggregation, stability, and expression levels based on sequence features and process parameters.

    • Process optimization dashboard

      Real-time simulation interface

      Deploy visualization showing predicted outcomes across different manufacturing scenarios, enabling rapid comparison of process alternatives without physical experimentation.

    • Only candidates that are both biologically active and manufacturable at scale advance, reducing late-stage failure costs.
    • The Digital Twin is a reusable asset for every future ADC and TCE in the pipeline.
    • Simulation runs are faster and cheaper than physical scale-up experiments, accelerating the candidate selection decision.

    We've done this before

    See how we do Digital Lab
  • Enterprise AI

    Zero Trust Security Architecture

    Deploy identity-based Zero Trust security architecture protecting LabGenius' cloud-hosted ML models, DNA sequences, and laboratory automation systems from cyber threats while enabling secure collaboration with global pharma partners.

    • Identity-based access control

      Continuous verification layer

      Implement continuous identity verification for all users and systems accessing GCP resources, eliminating implicit trust and preventing lateral movement even when credentials are compromised.

    • OT security architecture

      IEC 62443-aligned controls

      Deploy industrial cybersecurity controls protecting laboratory PLCs, robotic controllers, and SCADA systems from unauthorized access, aligned with IEC 62443 for operational technology in life sciences environments.

    • NIS2 compliance framework

      Regulatory security alignment

      Establish security governance, incident reporting, and risk management procedures aligned with NIS2 directive requirements for life sciences critical infrastructure before the compliance deadline applies.

    • Proprietary EVA ML models and DNA sequences are protected regardless of which location a collaborator accesses them from.
    • Zero Trust architecture satisfies due diligence requirements from Sanofi and future pharma partners.
    • NIS2 alignment positions LabGenius ahead of the regulatory curve before the directive becomes applicable.

    We've done this before

    See how we do Enterprise AI

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what LabGenius Therapeutics's own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Lab Automation 70 → 95
The EVA platform is highly automated, but 30 percent of laboratory actions remain manual, which is where human error enters the DBTL cycle and where the ML training feedback loop slows down.
Data Integration 60 → 90
Benchling is in use, but data silos exist between the ML environment, LIMS, and analytical instruments — the MOBO cycle depends on closing those gaps.
Cloud Infrastructure 75 → 90
GCP is deployed with Terraform for infrastructure-as-code, but the environment lacks GxP validation and audit-ready documentation required for clinical transition.
Cybersecurity Posture 55 → 85
Basic cloud security controls are in place, but the hybrid model with remote data scientists and global pharma partner access requires Zero Trust architecture and IEC 62443-aligned OT controls.
Process Simulation 40 → 75
In silico developability screens exist, but a full bioprocess Digital Twin for predicting large-scale manufacturability has not yet been deployed.
Regulatory Readiness 45 → 85
The R&D-focused environment is not yet designed for clinical submissions; GAMP5 compliance, ALCOA Plus data integrity, and CSV documentation need to be built before Phase I filings.

Check this yourself

Our Service Portal has free self-assessments and market comparisons. These are the ones that line up with what we've read above — no sales call required.

Think we've read this right?

Talk to us

Related reading

This is an independent analysis prepared by A4BEE from publicly available information as of January 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with LabGenius Therapeutics, and may be incomplete or inaccurate. All company names and trademarks are the property of their respective owners. To request a correction or removal, contact [email protected].