Read the full case study here

‍

The Challenge - Exploring systematic yield improvement for a scaled-up production strain

With over two decades of experience, CanBiocin Inc. (Canada) develops and manufactures species‑specific probiotics for companion animals and livestock, offering freeze‑dried and microencapsulated formulations that support gut health. The company has established industry-leading practices for probiotic strain optimization, scaling candidates from microtiter plates to large fermenters.

‍

While effective, traditional methods are fundamentally limited by manual experimentation with each optimization cycle spanning several weeks. Thus, to accelerate R&D timelines and efficiently improve its core metrics in the scale - up process, CanBiocin gained interest in modern technologies such as high-throughput screening and AI and decided to partner with Differential Bio to e x plore their potential, starting with an already scaled-up Limosilactobacillus reuteri strain.

‍

The Results - A data‑driven lift for fermentation performance

Deploying Differential Bio's lab-in-the-loop AI strain optimization approach, CanBiocin could drive several production‑relevant conditions, following a balanced optimization approach:

‍

• 34.1% increase in biomass yield
• 27.6% reduction in media cost
• 3,000+ formulations screened in 4 weeks (a 34× increase in experimental throughput versus manual workflows)
• Simpler recipe: top formulation uses 8 ingredients (down from 11)
• 283% ROI after 10 batches, with the investment breaking even within the first three production batches

‍

Together, these outcomes improve unit economics while de‑risking subsequent scale‑up and tech transfer.

‍

How We Did It

Our five‑step optimization framework compresses months of iteration into weeks:

‍

• Step 1 - Knowledge transfer: align on goals, share strain and current media.
• Step 2 - High‑throughput data capture: thousands of 96‑well microtiter experiments run in parallel with real‑time growth monitoring.‍
• Step 3 - AI model build: train models mapping media inputs to growth outcomes to predict high‑value regions of the design space.‍
• Step 4 - Optimization & selection: Identify top candidates balancing yield, cost, and formulation simplicity.‍
• Step 5 - Scale‑up validation: confirm leading formulations in 4L bioreactors under production‑like conditions and deliver media specs, protocols, and model insights for implementation.

‍

A Partnership Built for Scale

Differential Bio deployed its integrated Virtual Scale-up Platform, coupling robotic data generation with AI-guided optimization, to optimize the fermentation media for CanBiocin’s commercial L. reuteri production strain - delivering higher biomass yield while reducing media cost. This successful collaboration not only marks a shift from manual iteration to data-driven optimization, but also establishes a partnership for the future, equipping CanBiocin with a scalable platform to accelerate innovation and maintain a competitive edge in probiotic development.

‍

“We chose Differential Bio for their integrated approach combining robotic lab data generation with AI-driven optimization. We are happy with this project and look forward to strengthening our position as a leading probiotic ingredient developer together.” - Jake Burlet, CEO at CanBiocin

‍

What’s Next

This successful collaboration demonstrates how AI‑assisted media design and robotic experimentation can unlock performance gains even for already scaled‑up strains. We’re continuing to expand this approach across other lactic acid bacteria and probiotics programs, unlocking more value through media optimisation and beyond, powered by robotic data capture and AI‑based optimisation.

‍

Read the full case study here

‍