Protein Design AI Platform

Building Future Materials with AI & Biotech.

A Protein Design AI Platform for High-Value Resource Recovery. Powered by a proprietary model trained on 460M data points, we design Unnatural Super Enzymes — fast, precise, and economical — to recover the high-value resources of tomorrow.

DFRN / 2026 SEOUL · DAEJEON
460M
Protein Data Trained
15%
PET Activity Expression Rate
3 ea
Pipelines · PET / Nylon / REEs

Securing strategic resources
now defines industrial competitiveness.

Oil shocks, semiconductor material disputes, Middle East conflicts — external resource dependency has repeatedly triggered economic and industrial crises worldwide. The technology to smartly recover and circulate resources is the industrial competitiveness of the next era.

100%
Crude Oil Imported

100% of domestic crude oil demand is met through imports, with 62% concentrated in the Middle East.

SRC. GS Caltex Press Release
8.6%
Mineral Self-Sufficiency

Import dependency on China for critical minerals — lithium, nickel, cobalt, manganese, graphite — ranges from 70 to 100%.

SRC. Korea Institute of Geoscience, 2024
7~16%
Resource Recycling Rate

Recycling rates for petrochemical products and mineral materials remain in single digits, with no circular economy structure in place.

SRC. National Statistics (Multi-agency)

We design Unnatural Super Enzymes with AI
to realize strategic resource recovery.

Using a proprietary deep learning model trained on massive protein data, we design enzymes that selectively react with target resources, then validate and express their function in our Wet-Lab. Building on this, we recover and supply the high-value resources that future industries require.

STAGE 01 AI Protein Design LLM · Sequence Gen. STAGE 02 Wet-Lab. Validate · Express STAGE 03 Resource Recovery High-Value Materials Supply AI-BIO LOOP circl:o PLATFORM

Design → Validate → Recover · Closed Loop

It doesn't stop at AI-generated candidates. Through synthesis and reactivity validation in our in-house Wet Lab., empirical data feeds back into model retraining — a closed loop that rapidly drives up functional expression rates and deploys proven enzymes into real resource recovery processes.

01
Sequence LLM-Based Protein Design
Llama + LoRA stacking architecture to predict and generate target enzyme sequences.
02
Data-Driven Screening & Learning
Perplexity filtering, sequence–structure clustering, and molecular dynamics simulation for precise candidate extraction.
03
High-Activity Enzyme IP via Wet-Lab
Only IPs validated through in-house synthesis and reactivity testing are assetized — enabling free commercialization.
04
High-Value Resource Recovery & Supply
Deploying developed enzymes into real processes to recover and supply PET, Nylon, and rare resources — building tomorrow's raw material supply chain.

Proven enzymes,
backed by a measurable AI advantage.

DFRN's technology is proven on two fronts: enzyme development outcomes directly designed and validated through our own AI-Bio platform, and the competitive advantage of the underlying AI model itself.

PART A Validated Enzyme Outcomes

From PETase design to functional candidates — in under 12 months.

While competing AI models remain below 1–5% functional expression, we have already achieved 15%.

Jul 2025
Sequence-Based Model Development
Jan 2026
1st Synthesis · 32 Variants
38%Protein Expression
Mar 2026
2nd Synthesis · 20 Variants
Advanced In-silico Screening
75%Protein Expression
Apr 2026
3rd Synthesis · 120 Variants
3+ High-Activity Enzymes Secured
15%PET Activity Expression
2–3months
Commercially viable enzymes generated at scale in a single cycle.
By innovating the discovery and securing mechanism, we build IP barriers faster and higher.
PART B AI Model's Competitive Edge

Faster. More versatile. A measurable advantage.

Compared to competing protein generation models, training cost is 1/9 and memory usage is 1/1.5. Protein length generation is unlimited, with rapid target switching via LoRA fine-tuning.

Cost-effective

No large-scale
hardware infrastructure needed.

Competitor
40,968
GB
DFRN
4,560
GB
8.9× Training Cost Reduction
  • Low-power, high-efficiency model with low carbon footprint
  • Operable on compact infrastructure
Parallel Processing

Multiple projects
running simultaneously.

Competitor
3,100
MB
DFRN
2,090
MB
1.5× Memory Efficiency
  • Concurrent multi-project execution
  • Optimal per-target batch operation (Design → Synthesis → Retrain)
Universal & Scalable

Unlimited protein
length generation.

Competitor
300~500
AA
DFRN
UNLIMITED
AA
Versatility & Expandability
  • Rapid target enzyme switching via LoRA
  • Easy model expansion from small to large molecules
We are Differian.

From the AI-Bio Platform
to a Resource Recovery Complex.

Beyond being a technology company — becoming the new standard for resource recovery. Three phases of a bold blueprint, building one pillar of the future resource industry.

Ph#1
Validated

AI-Bio Platform Validation

via PET-Degrading Enzyme

Functional validation of our proprietary AI-Bio platform. We proved the model's real-world activity expression capability through high-difficulty PET-degrading enzyme design.

  • High-difficulty PETase designed and validated
  • 15% functional expression rate achieved
  • Dry & Wet Lab. integrated infrastructure built
Ph#2
Expanding

Pipeline Expansion on the Platform

PA6 · PA6,6 · REEs

Leveraging the validated platform to expand into high-value resources. We sequentially develop and assetize enzymes for nylon and rare earth element recovery.

  • Nylon (PA6, PA6,6) degrading enzymes
  • REEs (Rare Earth Elements) binding enzymes
  • Scaling high-value IP licensing business
Ph#3
Planned

Resource Recovery Complex

Recovery at Industrial Scale

Bringing secured IPs into actual processes to build an integrated resource recovery complex. Completing the transition from technology supplier to raw material supplier.

  • Pilot facility → commercial plant scale-up
  • High-value raw material supply business
  • A new pillar in the global resource supply chain

Milestones & Achievements

Apr 2026
Successful target material degradation experiment using proprietary AI-designed enzyme
Jan 2026
Successful PET depolymerization experiment using in-house biocatalyst
Dec 2025
Selected as KIBO (Korea Technology Finance Corp.) Star Valley Company
Dec 2025
Approved as Daejeon Corporate R&D Center
Aug 2025
Signed industrial PET workwear resource recovery research contract (Lindström)
Aug 2025
Venture company certification
Jul 2025
Seed investment secured (Korea Investment Accelerator)
Jun 2025
KIBO Venture Camp 16th cohort — Top 10 Outstanding Companies
Nov 2024
1st Asan University Demo Day — Encouragement Award (Climate Tech Track)
Nov 2024
5th KAIST Lean Startup Camp — Excellence Award
Sep 2024
KAIST Convergence Capstone Design — Selected Company
Aug 2024
KAIST Deep-Tech Startup Rapid Prototype Support Program — Selected
— DIFFERENT FUTURE THROUGH REACTION IN NATURE —

Looking for partners
to build future materials together.

We welcome all inquiries — research collaboration, IP licensing, business partnerships, and investment.

value@circlo.kr