☰

Circular Food Technology ApS (Agrain)

One shared data layer for scale-up

A Danish food company upcycling brewers' spent grain into ingredients, scaling from forty to twelve hundred tons

Industry
Food technology (upcycled ingredients from brewers' spent grain)
Headquarters
Copenhagen, Denmark
Public information as of
January 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of Circular Food Technology ApS (Agrain)'s published strategy and is not endorsed by, or produced in cooperation with, Circular Food Technology ApS (Agrain). Company website

Strategic priorities

Agrain is a Danish food technology company, legally Circular Food Technology ApS (CVR 39615231), founded in 2018 and operating commercially as Agrain. The company upcycles brewers' spent grain (BSG) — the wet protein-and-fibre residue left after brewing — into standardised flour and ingredient blends for bakery and food manufacturers. Production runs at the Bjæverskov facility about 45 minutes from Copenhagen, with a commercial headquarters on Pilestræde in central Copenhagen. Current output sits at roughly 30-40 tons per month against a publicly stated target of 1,200 tons per month — a 30-fold scale-up that is the central question for the business today.

The feedstock arrives in a fundamentally different condition from typical industrial inputs. BSG leaves the brewery at roughly 80 percent moisture and spoils within 12 to 24 hours, so the supply chain is a perishable reverse logistics network collecting from at least eight partner breweries across Greater Copenhagen and Zealand — Braunstein in Køge, Bryghuset Møn, Nørrebro Bryghus and Herslev Bryghus among them. Combined with more than 200 distinct input variants across breweries, beer styles and moisture levels, the standardisation work happens at the drying, milling and blending stages inside the Bjæverskov facility.

Agrain holds Organic and Kosher certifications and sells in both B2C (the Agrain-branded crackers and flour) and B2B (ingredient supply to industrial bakery customers) channels. The B2B side is the strategic priority, and a €3.5 million capital raise in May 2023 and additional funding signals point toward a Series B round dedicated to the 1,200-ton expansion. Investors include Collateral Good Ventures, North East Family Office, EIFO (Denmark's Export and Investment Fund) and Planetary Impact Ventures.

The current operating model depends heavily on manual coordination. Blending logic lives in Excel spreadsheets managed by named individuals, brewery data arrives by email and PDF, lab results sit in local LIMS or paper notebooks, and machine values are trapped in PLCs (programmable logic controllers, the industrial computers running each line). The 1,200-ton target cannot be served by the same process. The question for the next phase is whether the scale-up is designed into the data layer or built on top of it.

Challenges we see

  • Operations Manufacturing

    Standardising 200-plus input variants into one flour specification

    The Bjæverskov facility processes BSG from more than 200 unique input types — different breweries, beer styles and moisture levels — yet has to deliver a standardised output flour that meets B2B customer specifications.

    Where a blending recipe has to absorb that much variability at industrial throughput, the recipe itself becomes the load-bearing artefact: it has to be readable by the line, editable from the lab and auditable for every batch, not held in a single operator's spreadsheet.

  • Operations Operations

    Collecting perishable feedstock from 8+ breweries on a just-in-time schedule

    BSG leaves the brewery at around 80 percent moisture and spoils within 12 to 24 hours. Collection runs across Greater Copenhagen and Zealand to the Bjæverskov facility, with named partners including Braunstein, Bryghuset Møn, Nørrebro Bryghus and Herslev Bryghus.

    Where brewery schedules change and spoilage windows are measured in hours, dispatch decisions have to follow the bin, the calendar and the route in near real time rather than on a printed timetable, which is the difference between a logistics plan and a logistics system.

  • Operations Energy

    Keeping drying energy within unit economics as moisture varies

    Removing water from 80 percent to under 10 percent moisture is the most energy-intensive step on the line, and energy is the second-largest operating cost after labour and logistics at a Danish production site.

    Where dryer setpoints have to follow a moisture reading that changes by the truckload, a closed control loop that adjusts burner intensity on the current batch protects the energy budget that the unit economics depend on, rather than letting it drift with operator judgement.

  • Digital Integration

    Joining brewery, lab and machine data into one view of the batch

    Brewery input data arrives by email or PDF, lab results sit in local LIMS or paper notebooks, and process values live inside machine PLCs — three locations that do not currently exchange data.

    When a perfect batch cannot be reproduced because the inputs, the lab results and the machine readings cannot be read together, the Golden Batch analysis that should sit inside the operation has to be reconstructed by hand after the fact, which is what the scale-up is up against.

  • Compliance Regulatory

    Carrying Organic and Kosher status through a blended product

    Organic and Kosher certifications require evidence that every grain in a finished bag came from certified batches, while the blending process itself mixes batches to deliver a consistent specification.

    Where blending mixes origin along with everything else, batch genealogy has to be reconstructed from the inputs and the recipe rather than read off the bag, which is what makes an audit cycle a manual investigation rather than a routine check.

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. Moving the blending recipe from a spreadsheet to a controlled line recipe

    The proprietary blending recipe currently lives as a complex Excel workbook managed by named individuals, with no direct connection to the dosing PLCs that execute the blend on the line.

    Porting the recipe logic into a controlled service that receives near-infrared input readings and writes setpoints directly to the dosing PLCs removes the human transcription step and gives the operation a recipe that survives the people who wrote it.

    • Agrain, Research analysis, 2026 (proprietary blending algorithm)
    • World Bio Market Insights, Agrain feature, 2024

  2. Predicting bin fill and spoilage across the brewery network

    Collection runs on fixed schedules with phone coordination, so a brewery schedule change can mean a truck arriving to an empty silo while another bin overflows; spoilage windows are 12 to 24 hours.

    IoT (Internet of Things) level and temperature sensors in each brewery bin feeding a central dashboard that predicts fill and spoilage lets dispatch sequence the most urgent collections first and turn a fixed timetable into a live routing decision.

    • Agrain, Research analysis, 2026 (dead freight and spoilage exposure)
    • Agrain, Research analysis, 2026 (partner brewery list)

  3. Closing the dryer control loop on inline moisture

    Incoming moisture varies truckload to truckload, and dryer setpoints are adjusted by hand on operator judgement, which produces either over-drying (energy waste and nutritional damage) or under-drying (mould risk).

    Inline near-infrared moisture measurement on the intake belt driving a closed-loop dryer controller adjusts burner intensity and residence time on the running batch, with energy use as the named outcome and a 10 to 15 percent reduction in gas and electricity cited as the realistic range.

    • Agrain, Research analysis, 2026 (energy as second-largest OpEx)
    • Agrain, Research analysis, 2026 (dryer control opportunity sizing)

  4. Building one data model across inputs, lab results and machine values

    Brewery inputs (PDF or email), lab quality results (local LIMS or notebook) and process values (PLC) live in three disconnected stores, so the same batch cannot be traced from input to finished bag in one query.

    An ontology-based data platform — defining entities for brewery batch, lab sample, machine reading and finished lot — lets a Golden Batch query join them across systems, shorten Root Cause Analysis from days to minutes and make a successful batch a reproducible recipe rather than a happy accident.

    • Agrain, Research analysis, 2026 (three data silos)
    • Agrain, Research analysis, 2026 (Golden Batch analysis opportunity)

  5. Generating batch-level Certificates of Analysis and sustainability reports

    B2B customers require a Certificate of Analysis per shipment, and the EU Green Claims Directive will require primary, batch-specific sustainability data rather than static averages; both currently require manual assembly.

    Automated generation of the Certificate of Analysis from integrated lab data, paired with batch-specific sustainability reports calculated from live energy and logistics data, turns the compliance load into a product the company ships with the flour.

    • Agrain, Research analysis, 2026 (EU Green Claims Directive)
    • Agrain, Research analysis, 2026 (B2B COA requirement)

What we'd propose

  • Digital CDMO

    Smart factory architecture for the 1,200-ton expansion

    We design the IT/OT (Information Technology / Operational Technology — the enterprise and shop-floor systems respectively) architecture for the 1,200-ton facility before procurement: tag naming conventions, network topology and equipment data schemas agreed up front so the greenfield site opens with a single data path rather than assembling one afterwards.

    • Reference architecture for the new site

      One documented data path from day one

      Specify how equipment, line supervision, manufacturing execution and enterprise systems connect, including the segmentation model and the protocol standards, so every vendor on the project builds toward the same target from commissioning.

    • Dynamic recipe management for variable inputs

      A recipe that absorbs 200 input variants

      Implement a manufacturing execution system that holds the blending recipe as data and accepts near-infrared input readings from the intake belt as live setpoint inputs, so the recipe adjusts itself batch by batch rather than being re-typed each run.

    • OPC UA backbone across PLCs and sensors

      Process values leaving the controller

      Deploy OPC UA (Open Platform Communications Unified Architecture) on every PLC, sensor and analytical instrument so process values leave the equipment in a documented, vendor-neutral form rather than staying inside a closed controller.

    • The recipe is held in the line, not in a named individual's laptop.
    • Quality by Design principles arrive with the equipment instead of being added after.
    • The 30-fold scale-up uses the same headcount shape, because the architecture absorbs the variability.

    We've done this before

    See how we do Digital CDMO
  • Enterprise AI

    Bio-logistics IoT platform across the brewery network

    Connected bin sensors at the eight-plus partner breweries feed a dispatch and routing dashboard that predicts fill levels and spoilage risk, so collection becomes a live decision rather than a printed schedule.

    • Smart bin sensors in the brewery network

      Fill and temperature at the source

      Install ruggedised ultrasonic level and temperature sensors in each partner brewery silo so fill and spoilage risk are measured at the source rather than reported by phone.

    • Predictive dispatch engine

      Which bin gets the next truck

      An algorithm that combines bin fill rate, current temperature and the brewing schedule to predict when each bin will reach capacity and trigger collection before spoilage risk crosses the threshold.

    • Route optimisation dashboard

      The day's route is a calculation, not a printout

      A central visualisation showing real-time bin status across the brewery network with route calculations that prioritise the most urgent collections and minimise dead freight on light days.

    • Spoilage losses shrink because collections run on condition rather than calendar.
    • Dead freight disappears because the dispatch system knows the bin state, not the brewery's last call.
    • Logistics cost per ton becomes a measurable, optimisable line item.

    We've done this before

    See how we do Enterprise AI
  • Digital CDMO

    Closed-loop dryer control on inline moisture

    Inline near-infrared moisture measurement on the intake belt driving a model-assisted PID (Proportional-Integral-Derivative) controller on the dryer, so burner intensity and residence time follow the current batch rather than the last operator's instinct.

    • Inline near-infrared on the intake belt

      Fingerprinting every truckload

      Install near-infrared spectroscopy on the intake belt so moisture, protein and fat are measured per truckload before processing begins, removing the wait for a lab result before the dryer can be set.

    • Model-assisted PID on the dryer

      Burner intensity follows the moisture

      Implement a PID control loop with a model-based extension that adjusts burner intensity and residence time to the measured moisture, protecting nutrition and energy use on the same running batch.

    • Energy analytics dashboard

      kWh per ton, per shift

      A dashboard showing energy consumption per ton of output with trend and anomaly detection, so a 10 to 15 percent reduction in gas and electricity use shows up as a number on the operations wall.

    • Energy cost per ton becomes a controlled variable rather than a weather-dependent one.
    • Nutritional damage from over-drying drops out of the variability band.
    • The energy budget stops being the second-largest surprise in the monthly review.

    We've done this before

    See how we do Digital CDMO
  • Enterprise AI

    Ontology-based data platform across inputs, lab and line

    An ontology-driven data platform that ingests brewery input data, lab results and machine values and exposes them through one model, so a Golden Batch analysis joins inputs, process and output in a single query rather than a manual reconciliation.

    • Multi-source data ingestion

      Three streams into one model

      Build ingestion for brewery batch data (via API or document capture), lab data from the LIMS (Laboratory Information Management System) and machine data from the PLCs, with schema validation at the boundary so bad records fail loudly rather than silently.

    • Semantic process model

      Brewery batch, lab sample, lot, reading

      Define an ontology covering brewery batch, lab sample, finished lot and process reading as explicit entities with agreed relationships, so the same batch can be traced from input to bag in one query.

    • Golden Batch analytics

      Best run versus current run

      Overlay the current production run against the optimal historical profile so deviations are visible while the batch is still being produced, and so a perfect batch becomes a recipe rather than a memory.

    • Root Cause Analysis moves from days of reconciliation to minutes of query.
    • A successful batch becomes a recipe the line can repeat, not a story the team tells.
    • Tribal knowledge turns into institutional data before the people who hold it move on.

    We've done this before

    See how we do Enterprise AI
  • Agents

    Agents for Certificates of Analysis and sustainability reports

    Narrow, reviewable agents that assemble the recurring document load: drafting Certificates of Analysis and batch-specific sustainability reports from the integrated lab, process and energy data, with a named reviewer approving every output before it leaves the building.

    • Drafting Certificates of Analysis from lab and process data

      Per-shipment CoAs without per-shipment work

      Generate the first draft of each Certificate of Analysis from the lab result, the batch genealogy and the process record, so the author reviews and signs rather than assembles.

    • Batch-level sustainability reports from live data

      Scope 3 by the bag, not by the year

      Calculate batch-specific carbon, water and waste footprints from live energy, material and logistics data so each shipment carries a sustainability report derived from this batch rather than a static average.

    • Compliance template checking before review

      Gaps caught before the document enters the queue

      Run a completeness check against the B2B customer template and the EU Green Claims Directive requirements before a document reaches a human reviewer, so missing or inconsistent sections are flagged at draft time rather than at audit.

    • The compliance load that scales with shipment volume stops scaling with headcount.
    • B2B customers receive a CoA and a sustainability report from the same dataset as the audit.
    • Every output is traceable to the source records and signed by a named reviewer.

    We've done this before

    See how we do Agents

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what Circular Food Technology ApS (Agrain)'s own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Process automation 25 → 85
The Bjæverskov facility currently relies on the blending recipe in an Excel workbook and operator judgement on the dryer. The 1,200-ton target depends on closed-loop control on the dryer and a manufacturing execution system that holds the blending recipe as data.
Data integration 20 → 88
Brewery inputs, lab results and machine values live in three separate stores that do not currently exchange data. The Golden Batch analysis the scale-up depends on cannot happen until they are joined under one model.
Supply chain visibility 18 → 78
Collection runs on fixed schedules with phone coordination and spoilage windows of 12 to 24 hours. Continuous visibility into bin fill and temperature at the brewery silos is what turns the schedule into a system.
Quality management 35 → 85
The Certificate of Analysis and the Organic and Kosher audit trail are produced manually today. A scale-up to 1,200 tons per month with the same headcount depends on the certificate being generated from the data rather than typed.
Energy optimisation 30 → 75
Dryer setpoints are adjusted by hand on operator judgement, leaving energy — the second-largest operating cost — exposed to variability in incoming moisture. Closing the loop on inline moisture measurement is the named target.
ESG reporting 40 → 88
Sustainability numbers are currently calculated from static averages, but the EU Green Claims Directive will require primary, batch-specific data and the B2B customers are starting to ask for it.

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 Circular Food Technology ApS (Agrain), 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].