☰

7Talos S.A.

Decentralized CAR-T manufacturing

An AI-driven CAR-T cell therapy company built on a data platform, modular automation, and a distributed production model

Industry
CAR-T Cell Therapy
Headquarters
Athens, Greece
Public information as of
January 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of 7Talos S.A.'s published strategy and is not endorsed by, or produced in cooperation with, 7Talos S.A.. Company website

Strategic priorities

7Talos is a Greek biotech company developing AI-guided CAR-T cell therapies for haematological cancers, using microfluidic technology to automate the genetic modification and expansion of patient-derived T cells. The core innovation is a decentralised manufacturing model: portable microfluidic bioreactors deployed at local healthcare facilities, with centralised quality oversight from a hub team. This contrasts with the conventional CAR-T model where a patient's cells travel to a centralised manufacturing centre, are modified, and return — a process that takes two to four weeks and costs USD 400,000-500,000 per dose.

The company's Multi-Omics programme integrates genomic, transcriptomic, proteomic, and metabolomic data with bioreactor telemetry to build a closed-loop AI system that guides cell expansion in real time. The goal is to reduce manufacturing time to under one week and lower the per-dose cost sufficiently to reach a broader patient population. The pipeline includes haematology indications with IND filings anticipated across multiple indications.

The digital challenge is distinct from conventional biopharmaceutical manufacturing: the production equipment sits in hospital environments outside 7Talos's direct control, the operators are hospital medical staff without bioprocess engineering training, and the regulatory requirements (21 CFR Part 11, GxP) apply to distributed nodes rather than a single controlled facility. Data integrity and system security are existential requirements — not just operational improvements.

Challenges we see

  • Supply Chain Digital

    Verifying code and data integrity across distributed bioreactors

    The decentralised manufacturing model requires loading patient-specific cell preparation parameters and manufacturing protocols into bioreactors located at hospital facilities. Any unauthorised modification to the loaded code or data before execution could produce an incorrect cell product that harms the patient.

    When the bioreactor is located in a hospital corridor rather than a 7Talos-controlled cleanroom, the integrity of the loaded parameters depends on the security of the hospital IT environment. A parameter modification that is invisible to the centralised system produces a cell product that appears valid but contains the wrong genetic programme.

  • R&D Digital

    Multi-omics data fragmentation preventing closed-loop AI feedback

    The Multi-Omics programme generates heterogeneous data — genomic sequences, transcriptomic profiles, proteomic assays, metabolomic measurements, and bioreactor telemetry — from different instruments, in different formats, at different time intervals. Without an ontology-driven integration layer, these data streams cannot be automatically correlated.

    Where multi-omics data sits in disconnected systems, the AI closed-loop system cannot access a unified view of cell state. The real-time feedback loop that adjusts bioreactor parameters closes too slowly — or not at all — because correlating genomic data with telemetry data requires manual data assembly rather than automatic inference.

  • Labor Shortages Labor

    Medical staff without bioprocess expertise operating AI-powered bioreactors

    Hospital-based operators at local facilities have clinical expertise but not bioprocess engineering training. An AI-powered bioreactor that recommends parameter adjustments may be perceived as opaque and untrustworthy if the operator cannot see the reasoning behind the recommendation.

    When operators default to manual mode to override AI recommendations they do not understand, the efficiency gains from the AI system are negated and the manual intervention introduces a human error risk that the AI system was designed to reduce.

  • Regulatory Pressure Digital

    GxP audit trails across distributed manufacturing nodes

    21 CFR Part 11 and GxP requirements mandate complete, tamper-evident audit trails for all manufacturing steps. In a decentralised model where bioreactors are in hospital facilities, maintaining a continuous, verifiable audit trail from a remote location is technically complex.

    Where audit trail data is collected intermittently or manually, the gap between recordings means a batch manufacturing record may not reflect what actually happened during the entire production window. Regulatory inspectors evaluating a distributed manufacturing site need the same evidence confidence as they would for a single-site facility.

  • Manufacturing Manufacturing

    Scaling microfluidic bioreactors to clinical production

    The microfluidic bioreactor platform has demonstrated proof-of-concept at R&D scale, but the transition to GMP-compliant clinical production requires consistent performance across multiple units, interoperability with different hospital infrastructure configurations, and predictive modelling to anticipate scale-up deviations.

    When production-scale bioreactor performance is not modelled in advance, the first clinical runs become empirical learning exercises rather than predictable manufacturing steps — increasing the risk of failed batches that delay IND timelines and consume the limited patient samples available for early clinical work.

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. Decentralized manufacturing data platform for distributed bioreactor network

    The current centralised CAR-T manufacturing model involves significant logistical overhead and long production cycles, often taking weeks to deliver modified cells to patients while incurring high costs that limit patient access.

    Implement a data platform that unifies bioreactor telemetry, multi-omics data streams, and manufacturing protocol management across distributed hospital-based nodes — enabling centralised quality oversight to monitor and verify every distributed production run in real time.

    • 7Talos decentralised manufacturing strategy
    • CAR-T centralised vs distributed production cost analysis

  2. Digital twin for CAR-T biopharma process scale-up

    Many advanced therapies fail during transition from lab-scale R&D to GMP-compliant production because process parameters do not scale linearly and the failure only becomes apparent when the physical run is executed — at significant cost and delay.

    Develop a digital twin of the CAR-T microfluidic production workflow to simulate cell expansion behaviour, optimise bioprocess parameters, and run virtual fail-fast iterations before committing to a physical run — reducing the empirical trial count during scale-up.

    • 7Talos IND filing timeline for haematology indications
    • Microfluidic bioreactor scale-up requirements documentation

  3. GxP regulatory automation for distributed manufacturing

    Ensuring continuous GxP compliance across a decentralised network of hospital-based bioreactors is manually intensive, requires full audit trails for every production run, and creates audit risk when data collection depends on hospital IT infrastructure.

    Deploy automated regulatory intelligence tools that continuously verify data integrity across distributed nodes, generate audit-ready batch records automatically from instrument data, and flag compliance deviations before they accumulate into inspection findings.

    • 21 CFR Part 11 and GxP requirements for distributed manufacturing
    • FDA EMA guidance on CAR-T decentralised production

  4. Ontology-driven multi-omics and bioreactor data integration

    Multi-layered biological information — genomics, transcriptomics, proteomics — is often disconnected from physical bioreactor telemetry, preventing a unified view of the cell expansion process that the AI closed-loop system requires to function.

    Establish an ontology-driven data ecosystem that creates a single source of truth, automatically correlating LIMS samples to SCADA batches with full data traceability and enabling the AI closed-loop system to access a unified view of cell state in real time.

    • 7Talos Multi-Omics programme objectives
    • Closed-loop AI feedback system requirements

  5. AR/VR workforce enablement for hospital-based operators

    Hospital-based medical staff operating CAR-T bioreactors lack bioprocess engineering expertise and may default to manual override when AI recommendations are not understood — negating efficiency gains and increasing the human error risk the system is designed to reduce.

    Combine UX redesign of the bioreactor human-machine interface with AR/VR-guided training and real-time troubleshooting support, enabling non-specialist hospital staff to operate AI-guided equipment confidently and correctly.

    • Hospital partner operator training needs assessment
    • Digital operator competency framework for decentralised manufacturing

What we'd propose

  • Enterprise AI

    Ontology-Driven Biomanufacturing Data Platform

    We deploy a centralised data lakehouse using semantic modelling to unify disparate IT/OT data streams — bioreactor telemetry, multi-omics instrument data, manufacturing protocols, and quality records — into a single source of truth for GxP environments, enabling real-time analytics and automated compliance reporting across all distributed manufacturing nodes.

    • Bioreactor telemetry ingestion at distributed nodes

      Every bioreactor connected to the platform in real time

      Deploy lightweight, validated edge clients at each hospital-based bioreactor that stream telemetry data to the centralised platform in real time — covering cell density, nutrient concentrations, temperature profiles, gas flow rates, and AI recommendation logs — with tamper-evident transmission.

    • Multi-omics data ontology

      Genomics, transcriptomics, and bioreactor data in one model

      Build an ontology-driven data model that maps genomic sequences, transcriptomic profiles, proteomic assays, and metabolomic measurements to the corresponding bioreactor production run, enabling the AI closed-loop system to query cell state across all data dimensions simultaneously.

    • Automated batch record generation

      GxP batch records assembled from instrument data

      Configure automated extraction of production data from the platform to populate a GxP-compliant electronic batch record for each run — eliminating manual transcription, ensuring completeness, and generating an audit-ready record the moment the run concludes.

    • The AI closed-loop system operates on a unified data view rather than assembling data manually from disconnected sources — reducing the time from data availability to parameter recommendation from hours to seconds.
    • GxP batch records are complete and audit-ready at the end of each run, rather than requiring days of manual assembly by quality personnel.
    • The platform scales to any number of distributed nodes without requiring a proportional increase in quality assurance headcount.

    We've done this before

    See how we do Enterprise AI
  • Digital CDMO

    MTP-Compliant Modular Automation for Decentralised Bioreactors

    We implement the Module Type Package standard to enable vendor-agnostic Plug and Produce integration of the microfluidic bioreactors with local hospital facility control systems, ensuring that each distributed node can be commissioned, operated, and decommissioned without requiring custom engineering at each site.

    • MTP equipment descriptions for microfluidic bioreactors

      Bioreactors described in a standard machine-readable format

      Create MTP-compliant equipment descriptions for the 7Talos microfluidic bioreactor platform, covering process parameters, alarm handling, and material connections — enabling the hospital's facility management system to recognise and interface with the bioreactor automatically on connection.

    • Plug and Produce commissioning workflow

      New hospital site operational in days, not weeks

      Build a standardised commissioning workflow that uses the MTP equipment package to configure the distributed node's network connections, SCADA interface, and audit trail integration in a matter of days rather than requiring on-site engineering for each new hospital installation.

    • Interoperability with hospital infrastructure

      Connect to hospital LIMS and HVAC without custom code

      Develop integration adapters that use MTP and HL7/FHIR standards to connect the bioreactor system to hospital LIMS and facility management systems — covering patient sample registration, environmental monitoring data, and equipment status — without requiring modifications to the hospital's existing infrastructure.

    • New hospital sites are operational within days of equipment delivery rather than weeks of on-site engineering, accelerating the network roll-out timeline.
    • The MTP equipment package means the bioreactor is self-describing to the hospital's control system — reducing the specialised engineering support required at each distributed node.
    • Interoperability with hospital infrastructure enables patient sample data to flow automatically into the manufacturing record without manual re-entry.

    We've done this before

    See how we do Digital CDMO
  • Digital Lab

    Digital Twin and Predictive Bioprocess Modelling

    We build a cloud-agnostic digital twin of the CAR-T microfluidic production workflow — including cell culture kinetics, genetic modification dynamics, and scale-up deviations — to simulate cell expansion behaviour, optimise bioprocess parameters, and run virtual fail-fast iterations before committing to a physical production run.

    • CAR-T process simulation platform

      Cell expansion virtualised before physical execution

      Develop a calibrated simulation model of the CAR-T production workflow that replicates the specific cell culture kinetics, transfection efficiency dynamics, and expansion curve of the 7Talos platform — enabling operators to test parameter changes in simulation before running at physical scale.

    • Scale-up deviation prediction

      Scale-up risks identified before clinical runs

      Build a scale-up correlation model that predicts how parameters that work at R&D scale will behave at clinical production scale, identifying the specific parameters most likely to deviate and the correction actions that should be pre-prepared before the first clinical run.

    • AI model retraining from physical run data

      Digital twin improves with every completed run

      Configure the digital twin to receive physical run data automatically after each completed production run, retraining the underlying cell culture model to reduce prediction error over time — making each successive run more predictable than the last.

    • Physical optimisation runs are reduced because simulation identifies viable parameter windows before the bench work begins — each avoided failed run preserves a limited patient sample and keeps the IND timeline on schedule.
    • The first clinical production run is better prepared because the scale-up model has identified and pre-addressed the parameters most likely to deviate.
    • The digital twin improves with experience, making each successive cell type or indication faster to establish than the one before it.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    AR/VR Workforce Enablement for Hospital-Based Operators

    We combine a UX redesign of the bioreactor human-machine interface with immersive AR/VR training modules and real-time AR-guided troubleshooting, enabling hospital-based medical staff with no bioprocess engineering background to operate the AI-guided system confidently and contribute to quality outcomes.

    • Immersive bioreactor operation training

      Practice CAR-T production without patient samples at risk

      Build a library of VR training scenarios covering the complete CAR-T production workflow — sample receipt, bioreactor inoculation, parameter monitoring, harvest — allowing hospital operators to build procedural competency in simulation before operating with actual patient cells.

    • HMI UX redesign for non-specialist operators

      AI recommendations visible and explainable

      Redesign the bioreactor human-machine interface to make AI recommendations visible, contextual, and actionable for medical staff without bioprocess engineering training — replacing opaque confidence scores with plain-language explanations of what the AI is observing and what action it recommends.

    • AR-guided real-time troubleshooting

      Expert guidance overlaid on the equipment

      Integrate an assisted-reality layer (RealWear or equivalent) that connects the on-site operator to a centralised 7Talos bioprocess specialist during active production events — enabling the specialist to see what the operator sees and provide step-by-step guidance without travelling to the hospital site.

    • Operators gain procedural competency before treating patients, reducing the deviation risk that comes from unfamiliarity with the equipment or the AI recommendation logic.
    • When an unexpected production event occurs, the on-site operator is not isolated — a specialist can guide them through the resolution in real time without waiting for a site visit.
    • The UX redesign builds trust in AI recommendations over time, reducing the manual override rate that currently negates the efficiency gains of the AI system.

    We've done this before

    See how we do Digital Lab
  • Digital CDMO

    GxP-Ready Security and Compliance Infrastructure

    We deploy high-availability, GxP-compliant cloud infrastructure with high-security connectivity to distributed hospital-based bioreactors — ensuring data integrity, execution flow verification, and tamper-evident audit trails across all manufacturing nodes to satisfy FDA and EMA regulatory requirements.

    • Secure bioreactor connectivity for distributed nodes

      Encrypted, authenticated connection from every hospital site

      Implement mutual TLS authentication and encrypted data channels from each hospital-based bioreactor to the centralised platform, ensuring that parameter loads and telemetry streams are verified as originating from the correct, approved bioreactor — preventing spoofing attacks on the parameter loading system.

    • Automated data integrity verification

      Every data point verified as unaltered since collection

      Deploy a continuous data integrity verification system that generates cryptographic attestations for each data point as it is collected at the bioreactor — enabling FDA and EMA inspectors to verify that the electronic records presented in a regulatory submission have not been altered since the production run.

    • Regulatory intelligence and compliance flagging

      Compliance gaps identified before they become audit findings

      Build a regulatory intelligence dashboard that continuously monitors distributed node compliance status — data completeness, audit trail gaps, electronic signature coverage — and alerts the quality team when a deviation is accumulating that could become an inspection finding.

    • Regulatory submissions for distributed manufacturing sites carry the same evidence confidence as single-site submissions — because the data integrity infrastructure is the same.
    • A data integrity incident at a distributed node is detected and contained within hours rather than being discovered during a regulatory inspection months later.
    • The compliance monitoring dashboard shifts the quality team from reactive firefighting to proactive risk management.

    We've done this before

    See how we do Digital CDMO

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what 7Talos S.A.'s own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Data Orchestration 35 → 95
Multi-omics instrument data, bioreactor telemetry, and LIMS records currently sit in disconnected systems without an ontology-driven integration layer. The AI closed-loop system requires a unified data view to function, and without it, the correlation analysis that would detect cell state deviation is assembled manually rather than performed automatically.
Interoperability 30 → 90
The microfluidic bioreactor has not yet been deployed with MTP-compliant integration at a hospital site. Each new hospital installation currently requires custom engineering to connect to local infrastructure, meaning the decentralised network expansion scales only as fast as bespoke integration projects can be completed.
AI Reliability 45 → 95
The AI closed-loop system has demonstrated proof-of-concept at R&D scale, but the digital twin required for scale-up prediction and the data infrastructure for real-time multi-omics integration have not been built. Without these, the AI system's recommendations are based on incomplete data and cannot yet fulfil the real-time steering function that the model depends on.
Regulatory Automation 25 → 95
No automated batch record generation system exists; GxP batch records for distributed manufacturing nodes are assembled manually. The data integrity verification required by 21 CFR Part 11 is not continuous, leaving a gap between what happened during a production run and what the regulatory record reflects.
Workforce Fluency 20 → 85
Hospital-based operators at the first deployment sites have had no immersive CAR-T bioreactor training and no exposure to the HMI interface before the equipment arrived. The training programme and UX redesign that would address digital hesitancy are not yet operational.
Security Resilience 30 → 95
Bioreactor connectivity to hospital IT environments has not yet been security-hardened. The parameter loading mechanism that would verify authenticity and prevent tampering from a compromised hospital network account has not been designed or implemented.

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 7Talos S.A., 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].