FXNL
Projects

Project / Manufacturing / Process Control

Turning Production Data Into Real-Time Control.

FXNL built an intelligent fermentation monitoring and control system that combines sensor data, visual analysis, historical production data and expert knowledge to continuously monitor batches and make bounded adjustments as conditions change.

Industry
Food & Beverage Manufacturing
Solution
AI + Process Automation
Data Sources
Sensors · Cameras · Production Systems
Scope
End-to-End

The Impact

From periodic monitoring to continuous control.

The previous process depended on multiple employees gathering information from separate systems, inspecting batches and manually interpreting conditions. The new system continuously assembles those signals and responds while the production environment is still changing.

Before

Hours / Days

to assemble, interpret and act on production information

After

~1–3 min

typical polling interval per connected data source

Individual sources are typically polled every 1–3 minutes, depending on the connected system — not all at once and not on a single fixed schedule.

Operational Impact

Less manual monitoring · Faster intervention · Greater batch-to-batch consistency · Reduced process variation · Better production records

The Challenge

The data existed. The response was delayed.

Production information was spread across sensors, dashboards, physical inspections, manual tests and employee experience. Different people understood different parts of the process, making it difficult to assemble a complete picture of a batch quickly.

By the time the information was collected and analyzed, conditions could already have changed—forcing teams to make adjustments using an incomplete or delayed view of production.

Before

  1. Separate Sensors
  2. Manual Inspection
  3. Gather Data
  4. Human Analysis
  5. Historical Comparison
  6. Adjustment
  7. Repeat

After

  1. Sensors + Cameras
  2. Continuous Monitoring
  3. AI-Assisted Interpretation
  4. Bounded Adjustment
  5. Measure Again

Continuous feedback loop — measure again, then adjust

The System

Sense. Interpret. Adjust. Measure Again.

Instead of treating each sensor or production system independently, FXNL created an orchestration layer that combines current conditions, historical production data and expert knowledge into a continuous feedback loop.

Monitored Signals

  • Temperature
  • Humidity
  • CO2
  • Sugar / Starch
  • Light
  • Color
  • Froth / Surface
  • Yeast Conditions

Monitor

Poll sensors, cameras and production systems

Interpret

Combine signals into a current read of the batch

Interprets Using

  • Real-Time Conditions
  • Historical Batches
  • Expert Knowledge

Decide

Determine an appropriate, bounded response

Adjust

Apply permitted changes within defined limits

Within Boundaries

  • Temperature
  • Humidity
  • Lighting
  • Formulation Variables

Active Batch

Conditions change — then the batch is measured again

Measure Again — loop returns to Monitor

Exceptions & Oversight

  • Dashboard
  • Email
  • SMS
  • Expert Override

When conditions fall outside acceptable ranges, the system raises alerts and production experts retain full visibility and override capability.

What the System Sees

A batch is more than one measurement.

Fermentation depends on multiple interacting conditions. The system combines product measurements, environmental data and visual signals rather than evaluating each source in isolation.

Product

  • Sugar / starch
  • Yeast conditions

Environment

  • Temperature
  • Humidity
  • CO2
  • Light exposure

Visual

  • Color
  • Froth
  • Surface conditions

Connected sources are monitored at intervals appropriate to each device and system.

Expert Knowledge

Capturing what experienced people just know.

Not all of the manufacturer’s production knowledge existed in a database or operating procedure.

Experienced employees had learned to recognize subtle changes in color, froth, fermentation behavior and other characteristics through years of production. Different employees also held different parts of that knowledge.

FXNL worked with those experts to identify meaningful signals, understand what they indicated, connect them to measurable conditions and outcomes, and incorporate that knowledge into the system’s monitoring and decision logic.

Some of the most valuable production data wasn’t stored in a system. It was stored in people’s experience.

The system operationalizes and scales portions of that expertise while keeping production experts in oversight and exception-management roles.

Responsibility

Automation with defined boundaries.

The system handles continuous monitoring and permitted adjustments while production experts retain visibility, override capability and responsibility for exceptions.

AI / System

  • Interprets current production conditions
  • Relates signals to historical patterns
  • Identifies deviations and relationships
  • Determines appropriate bounded responses

Automation / Control

  • Polls connected devices and systems
  • Executes permitted environmental adjustments
  • Logs conditions and interventions
  • Sends dashboard, email and SMS alerts

Human / Expert

  • Defines operating knowledge and boundaries
  • Oversees batches and system behavior
  • Investigates exceptions
  • Overrides adjustments when necessary

Implementation

The Hard Part Was Connecting the Physical and Digital Process.

The manufacturer already had sensors, production equipment, historical information and experienced employees. The challenge was turning those separate sources into one coherent operating system.

FXNL had to understand how numerous proprietary and third-party devices exposed data, how often each system could be queried, what its measurements actually represented, and how changes in one production variable could affect others.

The work also required translating years of production experience into signals, operating boundaries and decision logic the system could use.

The challenge wasn’t collecting more data. It was making the existing data actionable while it still mattered.

FXNL’s End-to-End Role

  1. Discover
  2. Map
  3. Architect
  4. Build
  5. Integrate
  6. Deploy

Built to Extend

One system. More production processes.

The current system supports kombucha fermentation and production monitoring. Its architecture can be extended to additional products, production lines, sensors and control systems as the manufacturer’s operations evolve.

  • Kombucha
  • Additional Products
  • Additional Production Systems

Kombucha production is deployed today. Additional products and production systems describe potential future extensions, not current deployments.

Have a workflow like this?

FXNL helps businesses connect fragmented data, systems and expertise into practical AI solutions that improve how work gets done.

Start a Conversation