Carry the process from bench to partner tank

What we'd propose

• Digital CDMO

  A standard digital technology-transfer package for partner sites

  A structured, version-controlled transfer package that carries a fermentation process to a contract manufacturer as data: process description, equipment specifications, critical quality attributes and recorded process conditions in one access-controlled form, so every receiving site matches against the same reference.

  • Parameterised transfer templates

    One structure every site fills

    Define process descriptions, equipment specifications and critical quality attributes as a parameterised template, so each transfer produces a package in the same shape instead of a bespoke document set the receiving site has to re-interpret.

  • Version control and access control

    Every change tracked and scoped

    Hold process parameters and protocol documents under version control with role-based access, so a receiving site always works from the current recipe and 21st.BIO keeps a record of what was shared, with whom, and when.

  • Auditable hand-off workflow

    Traceability through the transfer

    Track each transfer step with an immutable log, so the reference every partner site is matched against — and the conditions it was matched under — is recorded rather than remembered.

  • Each receiving site verifies against a common reference instead of adapting a bespoke document set.
  • The interval between hand-off and a reproducible run — the part of the revenue timeline 21st.BIO can influence — shortens.
  • What was transferred, to whom and under which parameters is recorded rather than reconstructed later.

  We've done this before

  • Process Transfer Reporting System
  • Strategic Integration Guides for Pharmaceutical Control Systems

  See how we do Digital CDMO
• Digital CDMO

  Real-time monitoring for the pilot fermentation plant

  Instrument the 3,000-litre pilot bioreactors, bring their process data into one time-series model, and run deviation detection against the batch currently in the tank, so a signal about a sensitive fungal run arrives while the batch is still live.

  • Bioreactor data acquisition

    Process data off the tanks

    Connect the pilot bioreactor sensors and controllers through OPC UA (Open Platform Communications Unified Architecture) so process values leave the equipment in a documented, vendor-neutral form rather than staying inside a closed control system.

  • Deviation detection against a live batch

    Signals during the run

    Build the normal operating envelope for each critical parameter from prior runs, then flag drift in shear, oxygen transfer or temperature against the current batch, so a signal reaches operators in minutes rather than in a post-run report.

  • Non-invasive vision monitoring

    Watching what sensors miss

    Add external camera-based monitoring for foam and other visual indicators that in-tank probes do not capture, giving operators an early read on conditions before they threaten a batch of high-value strain.

  • Deviations surface against the batch that is running, not the one already lost.
  • One set of pilot-plant process data serves operations, process development and R&D instead of three separate extracts.
  • The conditions behind a good or bad run are recorded as data, ready to compare against the next campaign.

  We've done this before

  • Bioreactor Control: Computer Vision for Non-Invasive Foam Management
  • Transforming Raw Bioprocess Data into Actionable Insights

  See how we do Digital CDMO
• Enterprise AI

  One data model for Davis strain data and Søborg process data

  An ontology-based data platform that defines the entities both sites share — strain, run, vessel, parameter, result — once, then loads Davis R&D data and Søborg pilot-plant data against that single model, so a fermentation outcome can be queried against the strain that produced it.

  • Shared strain-to-process ontology

    One agreed set of terms

    Define strain, run, vessel, parameter and result as explicit entities with agreed relationships, so a query written once returns comparable answers across both sites instead of two local dialects of the same data.

  • Pipelines from lab and pilot systems

    Loading both sites

    Build ingestion for genomic, metabolic and phenotypic data from Davis and for bioreactor process data from Søborg, with schema validation at the boundary so bad records fail loudly rather than propagate quietly.

  • Cross-site query and retrieval

    Ask across both sites at once

    Expose the model so a process scientist can relate a Søborg run to a Davis strain directly, without a manual data pull across the time-zone gap for each question.

  • A fermentation result can be traced to its strain and conditions as a query, not a manual assembly.
  • Integration is done once against a shared model instead of once per point-to-point link between the sites.
  • New vessels, strains and future sites attach to the model rather than triggering another reconciliation.

  We've done this before

  • Ontology-Driven Ecosystem for Biomanufacturing Data
  • Scaling Biotech Manufacturing via Universal IT/OT Architecture

  See how we do Enterprise AI
• Digital Lab

  Automated data capture for strain screening

  Integrate automated imaging such as Reshape Biotech and the analytical instruments in Davis into a shared R&D data layer, so phenotypic results are captured as data at source and screening runs at higher throughput without adding manual transcription.

  • Imaging and instrument integration

    Results captured at source

    Connect automated imaging and analytical instruments so results arrive with strain identity, method and timestamp attached, instead of being read off a screen and re-keyed into another system.

  • Shared R&D data layer

    One queryable strain record

    Bring genomic, metabolic and phenotypic results into one queryable store, so strain candidates can be compared across experiments rather than within single runs.

  • Screening workflow orchestration

    Campaigns tracked end to end

    Schedule, track and capture results for high-throughput screening campaigns with full traceability, so throughput rises without loss of provenance.

  • Phenotypic results are captured as data rather than transcribed, so screening scales without proportional manual effort.
  • More strain variants can be evaluated per campaign, widening the design space explored before selection.
  • Every screening result carries its strain, method and timestamp, ready to feed the shared model.

  We've done this before

  • Fully Automated Autosampling Control System
  • Creating a Digital Lab for a Global Biotech Player

  See how we do Digital Lab
• Agents

  AI agents for strain and batch documentation

  Narrow, reviewable agents that take the repetitive part of the regulatory record: drafting strain-development and batch summaries from source data, checking a document against its template before review, and finding every record a strain or specification change affects. A named person approves every output.

  • Drafting from source records

    First drafts from system data

    Generate the first draft of a strain-development or batch document directly from the underlying system data, so the author edits and judges rather than assembles from scratch.

  • Template and completeness checking

    Gaps found before review

    Check a submitted document against its template and the relevant regulatory checklist, returning missing or inconsistent sections before it enters the human review queue.

  • Change-impact search across records

    Which records a change touches

    When a strain lineage, method or specification changes, retrieve every controlled record that references it so the scope of an update — for a GRAS maintenance file or an FDA notification — is known at the start.

  • Documents reach review complete, so the record for a notification is kept current rather than assembled against a deadline.
  • The scope of a strain or specification change is established by search rather than by recollection.
  • Every output traces back to the source records it came from and is signed off by a named reviewer.

  We've done this before

  • Process Transfer Reporting System

  See how we do Agents