Best overall AI solution for pharma and best for quality control
Quality issues in pharma don’t just slow down a batch – they trigger investigations, scrap, CAPAs, and regulatory headaches. Averroes.ai is designed specifically to prevent that.
Our platform plugs into your existing inspection cameras or AOI systems and turns every captured image into a high-fidelity QC checkpoint powered by production-grade AI.
Instead of relying on brittle rules or inconsistent manual review, Averroes learns directly from your images, identifies defects, flags anomalies that haven’t been seen before, and helps teams catch quality leaks before they spread across a batch.
Typical deployments in regulated environments see large reductions in false rejects, major drops in review time, and improved batch-release confidence because teams can trust what the system sees.
Averroes also supports continuous learning – essential for lines where ingredients change, lighting varies, or equipment drifts. And because it’s no-code, QC and process engineers can update models in minutes without waiting for data scientists or outside vendors.
Features
No-code AI inspection builder for detection, classification, segmentation, and review
99%+ classification accuracy and extremely low false reject rates
Works with existing vision hardware (no new cameras required)
Few-shot training: 20–40 images per defect class
WatchDog anomaly detection for unknown or untrained defect types
Real-time monitoring dashboards with trend alerts and batch-level visibility
Continuous learning with feedback loops and model versioning
Supports on-prem, hybrid, and cloud-agnostic deployments
Virtual metrology capabilities for high-precision inspection steps
Pros:
Reduces batch failures and compliance risk with extremely accurate, consistent inspections
Minimal change management thanks to compatibility with your current hardware
Rapid model updates keep pace with formulation changes or new SKUs
Low labeling burden accelerates rollout across multiple lines or sites
Strong fit for sterile, packaging, fill-finish, and container-closure integrity workflows
Cons:
Requires stable image capture processes to reach peak performance
If you care about accelerating drug discovery without turning your R&D into a black box, Insilico Medicine is one of the serious players to look at.
Their Pharma.AI platform ties together biology, chemistry, and clinical prediction in a single workflow, so you are not juggling five tools and ten exports just to move from target to lead. Under the hood, it combines PandaOmics for target discovery, Chemistry42 for generative molecule design, and inClinico for clinical trial success prediction.
The result is a software stack that can move you from “interesting hypothesis” to AI-designed candidates with supporting evidence in a fraction of the usual timeline.
It is still geared mainly toward small molecules and does come with a learning curve, but for teams with the right scientific and data muscle, the leverage is significant.
Features
Pharma.AI suite with PandaOmics, Chemistry42, and inClinico
Generative chemistry engine with multi-parameter optimization and novelty scoring
AI-driven target identification and prioritization using multi-omics and literature data
Virtual screening, property prediction, and retrosynthesis support
Clinical trial outcome prediction and indication prioritization
APIs and integration options for plugging into existing discovery workflows
Pros:
True end-to-end platform from target discovery through candidate and trial prediction
Generative models help design and triage novel small molecules quickly
Integrated filters for novelty, synthetic accessibility, and medicinal chemistry “sanity checks”
Strong validation record with multiple AI-designed assets advancing into the clinic
Can meaningfully compress early discovery timelines and reduce wet-lab iteration
Cons:
Best suited to teams with solid computational chemistry/data science capabilities
Primarily focused on small molecules, with more limited support for other modalities
Model complexity can feel opaque for users needing high interpretability and detailed rationale
Best for optimizing clinical trial strategy and decision-making from unstructured data
IBM Watson takes a data-heavy part of pharma and makes it manageable. Instead of sifting through genomic datasets, trial reports, scientific papers, regulatory filings, pathology notes, and internal documents, Watson uses its NLP engine to extract relationships, signals, and patterns that people would miss or take months to uncover.
This makes it especially useful for teams working on clinical trial design, patient stratification, biomarker identification, and safety assessment where unstructured text is the bottleneck.
Watson isn’t the fastest molecule generator or a fully turnkey discovery suite, but its strength lies in depth – the quality of its data integration, the maturity of its NLP models, and its explainability features that help researchers trust what they’re looking at.
With the shift to watsonx–powered foundation models, IBM has doubled down on transparency, regulatory alignment, and hybrid deployments for large enterprises that need scale and auditability.
Key Features
Advanced NLP for extracting entities, relationships, biomarkers, variants, and outcomes from massive biomedical literature
Smart Document Understanding to structure unstructured trial documents, regulatory files, and clinical reports
Explainable model outputs that support validation and regulatory review
Predictive analytics for trial outcomes, patient matching, and study feasibility
Integration with IBM watsonx for foundation model–based reasoning
Hybrid cloud deployment across secure on-prem, private cloud, or regulated environments
Pros:
Excellent for teams dealing with large volumes of unstructured scientific and clinical data
Strong interpretability features, improving trust and easing regulatory documentation
Proven track record in oncology, biologics, and real-world evidence projects
Scales well for large pharma organizations that need secure, hybrid infrastructures
Broad flexible toolkit rather than a narrow, locked-down solution
Cons:
Requires technical maturity and data engineering resources to fully operationalize
Not a turnkey discovery engine; best used as part of a larger R&D workflow
Best for high-throughput, AI-driven biological discovery and multimodal target/molecule generation
Recursion takes a very different approach to drug discovery – instead of starting with known targets or narrow hypotheses, they generate enormous biological and chemical datasets and let AI map relationships no one has seen before.
Their Recursion OS ties together automated wet labs, supercomputing (BioHive-2, built with NVIDIA), multimodal foundation models, and generative design tools across phenomics, transcriptomics, proteomics, and chemical space. This results in an industrialized workflow that’s built to uncover new targets, design molecules quickly, and validate them in a closed feedback loop.
The platform is one of the most comprehensive in TechBio. Recursion runs millions of experiments per week, converting every cellular image and biological output into structured digital data, and feeding it back into models that improve over time.
They’re also one of the few AI-native biotechs with clinical-stage programs across oncology and rare diseases, showing real-world traction beyond theoretical discovery.
Key Features
Recursion OS spanning target identification through early clinical development
Best for integrated AI plus automated biology in early discovery
Evotec sits at the intersection of advanced AI, automated biology, and large-scale drug discovery infrastructure. Unlike point-solution AI tools, Evotec operates more like an extension of a pharma R&D engine: target ID, screening, hit-to-lead, optimization, ADME-tox, and IND-enabling all sit under one roof.
Their strength comes from pairing in-silico modeling with high-throughput lab automation and deep therapeutic expertise across neuroscience, oncology, metabolic disease, kidney disorders, and anti-infectives.
Their portfolio of next-gen platforms – from machine learning–guided chemistry to automated wet-lab systems and PanOmics data layers – makes them a powerful option for pharma teams looking to compress early discovery timelines without giving up scientific rigor.
Add in partnerships with Novo Nordisk, Bristol Myers Squibb, Pfizer, Sandoz, and Just–Evotec Biologics, and they’ve built one of the most connected ecosystems in the industry.
Key Features
AI-integrated discovery platforms for hit identification, target validation, and lead optimization
Automated high-throughput screening and active learning loops
PanOmics data integration (phenomics, proteomics, transcriptomics)
Broad modality coverage from small molecules to biologics and protein degraders
Just–Evotec Biologics for end-to-end biologics design and manufacturing
Flexible partnering models ranging from CRO essentials to fully integrated discovery alliances
Proven performance through multiple pipeline milestones with big pharma
Pros:
Strong early-stage acceleration using AI plus automation rather than automation alone
Wide therapeutic expertise supported by deep omics-driven biology
Ability to run both small-molecule and biologics programs under a unified platform
High-quality partnerships validating technology and scientific approach
Well-suited for pharma companies looking to outsource significant chunks of discovery
Cons:
Large, complex infrastructure suited more to mid-size and large pharma than small biotech
Long timelines inherent to preclinical/clinical development still apply
Not a self-serve AI platform – requires collaborative engagement
Best for AI-driven manufacturing optimization and GxP-compliant pharma production
Aizon is laser-focused on the regulated, day-to-day reality of pharmaceutical manufacturing. Their platform replaces paper-based decision-making with real-time, GxP-compliant AI models that help teams improve yield, spot deviations early, reduce batch failures, and tighten process control across biologics and small-molecule production.
What makes Aizon stand out is the combination of pragmatic AI, industry-specific workflows, and regulatory alignment. The platform is built for FDA-regulated environments, not retrofitted for them.
Their digital twin technology for bioreactors, multi-site data unification, and batch-level predictive analytics give manufacturing teams a practical path toward smarter, more consistent execution without months of custom engineering.
Key Features
End-to-end SaaS solution purpose-built for GxP-regulated pharma manufacturing
Digital twin for bioreactors supporting real-time process prediction and optimization
Unifies structured and unstructured manufacturing data across sites and systems
Predictive models for deviation risk, yield optimization, and batch consistency
Built-in GMP, FDA, and ISO-aligned data integrity controls
Cloud-native architecture on AWS with rapid implementation (6 weeks typical)
Supports digital maturity progression through Execute, Unify, and Predict modules
Strong partner ecosystem for global onboarding and integration
Pros:
Designed specifically for pharma manufacturing
Clear regulatory advantage with GMP-ready workflows and FDA collaboration
Best for digital quality management and compliance automation in life sciences
Scilife is pragmatic and foundational: helping pharma, biotech, medical device, and ATMP companies build a quality system that keeps up with the scale and speed of modern operations.
Where most QMS tools are static and document-heavy, Scilife adds automation, traceability, risk management, training compliance, controlled printing, and design controls in one validated, 21 CFR 11–compliant platform. It’s built specifically for regulated life sciences teams that need audit readiness, cross-team accountability, and smart workflows without drowning in manual tasks.
For organizations struggling with siloed quality processes, recurring compliance gaps, or multi-site coordination issues, Scilife becomes a digital backbone – structured, traceable, and easy to scale.
Choosing the right AI platform is all about finding the tool that fits how your teams work, what your data looks like, and where your biggest bottlenecks sit.
Here are the criteria that matter most and how each of the 7 stacks up:
1. Match the Solution to Your Real Use Case
Getting this wrong is the fastest way to waste time and budget.
Pharma has wildly different problems depending on the team: discovery needs generative chemistry; QC needs consistency; clinical needs data interpretation; manufacturing needs deviation prediction.
Best Fit by Use Case:
Averroes.ai: Strongest for visual QC, real-time defect detection, and batch-release confidence.
Insilico Medicine: Ideal for small-molecule target discovery and molecule design.
IBM Watson: Great for teams drowning in unstructured trial, regulatory, or RWE data.
Recursion: Best for multimodal, high-throughput biological discovery.
Evotec: Suited to pharma organizations that want an outsourced discovery engine.
Aizon: Purpose-built for regulated manufacturing optimization.
Scilife: Best for QMS, compliance, design controls, and audit readiness.
Less Good:
Anything outside each platform’s specialization. For example, Aizon won’t help with R&D, and Scilife won’t design molecules.
2. Check Data Readiness and Compatibility
AI only performs as well as the data it receives.
Discovery tools need multi-omics depth. QC tools need stable imaging. Clinical tools need clean unstructured data ingestion. Manufacturing AI needs connected plant data.
Platforms Good at Handling Complex Data:
Recursion: Unmatched multimodal datasets and automated wet labs.
Evotec: Strong PanOmics integration across high-throughput workflows.
IBM Watson: Excellent at wrangling unstructured biomedical text.
More Sensitive to Data Conditions:
Averroes.ai: Needs consistent imaging to reach peak accuracy.
Aizon: Depends on digital maturity across sites.
Scilife: Analytics strength depends on clean quality records.
3. Evaluate Interpretability and Scientific Trust
Teams adopt AI faster when they can understand why the model made a call.
Strongest in Interpretability:
IBM Watson: Built specifically for explainable outputs.
Averroes.ai: Transparent defect evidence, model versioning, and review tooling.
Scilife: Audit trails and traceability baked in.
More Complex:
Insilico and Recursion – very powerful, but their generative/discovery models can feel abstract without strong internal science teams.
4. Consider Regulatory Alignment
Regulatory expectations shape every decision in pharma, so validation, traceability, and auditability matter more than raw speed.
Regulatory-Ready Leaders:
Averroes.ai: Built for regulated QC lines.
Aizon: GMP-focused with FDA collaboration.
Scilife: 21 CFR 11, ISO 13485, full QMS controls.
IBM Watson: Strong documentation and governance.
Less Suited Here:
Recursion and Evotec are strong scientifically but not QMS tools.
5. Integration and Workflow Fit
Pharma workflows are fragile. AI should slide in, not force a rebuild.
Best for Quick, Low-Disruption Integration:
Averroes.ai: Works with existing cameras and inspection setups.
Aizon: 6-week deployment model and site-friendly tools.
Scilife: Clear structure and template-driven rollout.
More Involved:
Recursion, Evotec, and Insilico require deeper technical onboarding.
6. Scalability and Long-Term Fit
The best solutions grow with your pipeline, SKUs, or sites.
Highly Scalable:
Averroes.ai, Aizon, IBM Watson, Recursion, Evotec
More Use-Case Bounded:
Scilife: Deep but QMS-specific.
Insilico: Powerful but small-molecule focused.
Looking To Stabilize Manufacturing Outcomes?
Get 99%+ accuracy with fewer disruptions.
Frequently Asked Questions
How do you measure ROI when implementing an AI solution in pharma?
ROI usually comes from reduced cycle times, fewer failed batches, faster discovery loops, or lower compliance burden. Teams track before/after metrics like throughput, deviations, or lead-selection timelines to quantify real impact.
Do AI platforms replace scientists, operators, or QA teams?
No. These tools augment expertise rather than replace it. Scientists, engineers, and QA specialists still make decisions; AI simply accelerates analysis, flags issues earlier, or handles repeatable tasks with higher consistency.
How long does it typically take to deploy a pharma-ready AI system?
Deployment ranges from weeks to several months depending on the platform and use case. Manufacturing-focused solutions like Aizon or Averroes deploy fastest, while deep discovery engines such as Recursion or Insilico require more setup and integration.
Can these AI systems handle multi-site or global operations?
Most platforms scale across multiple sites, but the readiness varies. Averroes, Aizon, IBM Watson, and Scilife support multi-site orchestration well, while heavy scientific platforms may require additional infrastructure and governance to scale globally.
Conclusion
Pharma teams are under pressure to make smarter, faster decisions without sacrificing compliance, yield, or patient safety.
Each of these 7 AI solutions plays a different role:
Averroes is the strongest fit for real-time manufacturing accuracy, reducing false rejects and tightening QC across lines.
Aizon focuses on predictive manufacturing performance and GxP-ready digital twins.
Insilico and Recursion push drug discovery forward but offer limited value on the shop floor.
IBM Watson helps teams extract insight from unstructured clinical and scientific data.
Evotec supports early R&D through integrated automation.
While Scilife strengthens quality systems and audit readiness.
If you’re exploring AI to improve inspection accuracy, stabilize processes, or reduce batch risk, book a free demo to see how Averroes supports pharma manufacturing teams.
Pharma teams are turning to AI because the pressure is real.
Discovery timelines stretch past 10 years, clinical failure rates hover around 90%, and quality issues still trigger costly investigations and delays.
The challenge isn’t finding AI tools – there are countless on the market – it’s knowing which ones solve the problem you have.
We’ll break down seven leading platforms and what each one is best suited for across R&D, trials, manufacturing, and quality.
Our Top 3 Picks
Best for Manufacturing & Quality Teams
Averroes.ai
VIEW NOWBest for R&D & Discovery Teams
Insilico Medicine
VIEW NOWBest for Clinical Strategy & Trial Design Teams
IBM Watson for Drug Discovery
VIEW NOW1. Averroes.ai
Best overall AI solution for pharma and best for quality control
Quality issues in pharma don’t just slow down a batch – they trigger investigations, scrap, CAPAs, and regulatory headaches. Averroes.ai is designed specifically to prevent that.
Our platform plugs into your existing inspection cameras or AOI systems and turns every captured image into a high-fidelity QC checkpoint powered by production-grade AI.
Instead of relying on brittle rules or inconsistent manual review, Averroes learns directly from your images, identifies defects, flags anomalies that haven’t been seen before, and helps teams catch quality leaks before they spread across a batch.
Typical deployments in regulated environments see large reductions in false rejects, major drops in review time, and improved batch-release confidence because teams can trust what the system sees.
Averroes also supports continuous learning – essential for lines where ingredients change, lighting varies, or equipment drifts. And because it’s no-code, QC and process engineers can update models in minutes without waiting for data scientists or outside vendors.
Features
Pros:
Cons:
Score: 4.8/5
View Now2. Insilico Medicine
Best for accelerating drug discovery
If you care about accelerating drug discovery without turning your R&D into a black box, Insilico Medicine is one of the serious players to look at.
Their Pharma.AI platform ties together biology, chemistry, and clinical prediction in a single workflow, so you are not juggling five tools and ten exports just to move from target to lead. Under the hood, it combines PandaOmics for target discovery, Chemistry42 for generative molecule design, and inClinico for clinical trial success prediction.
The result is a software stack that can move you from “interesting hypothesis” to AI-designed candidates with supporting evidence in a fraction of the usual timeline.
It is still geared mainly toward small molecules and does come with a learning curve, but for teams with the right scientific and data muscle, the leverage is significant.
Features
Pros:
Cons:
Score: 4.7/5
View Now3. IBM Watson for Drug Discovery
Best for optimizing clinical trial strategy and decision-making from unstructured data
IBM Watson takes a data-heavy part of pharma and makes it manageable. Instead of sifting through genomic datasets, trial reports, scientific papers, regulatory filings, pathology notes, and internal documents, Watson uses its NLP engine to extract relationships, signals, and patterns that people would miss or take months to uncover.
This makes it especially useful for teams working on clinical trial design, patient stratification, biomarker identification, and safety assessment where unstructured text is the bottleneck.
Watson isn’t the fastest molecule generator or a fully turnkey discovery suite, but its strength lies in depth – the quality of its data integration, the maturity of its NLP models, and its explainability features that help researchers trust what they’re looking at.
With the shift to watsonx–powered foundation models, IBM has doubled down on transparency, regulatory alignment, and hybrid deployments for large enterprises that need scale and auditability.
Key Features
Pros:
Cons:
Score: 4.6/5
View Now4. Recursion Pharmaceuticals
Best for high-throughput, AI-driven biological discovery and multimodal target/molecule generation
Recursion takes a very different approach to drug discovery – instead of starting with known targets or narrow hypotheses, they generate enormous biological and chemical datasets and let AI map relationships no one has seen before.
Their Recursion OS ties together automated wet labs, supercomputing (BioHive-2, built with NVIDIA), multimodal foundation models, and generative design tools across phenomics, transcriptomics, proteomics, and chemical space. This results in an industrialized workflow that’s built to uncover new targets, design molecules quickly, and validate them in a closed feedback loop.
The platform is one of the most comprehensive in TechBio. Recursion runs millions of experiments per week, converting every cellular image and biological output into structured digital data, and feeding it back into models that improve over time.
They’re also one of the few AI-native biotechs with clinical-stage programs across oncology and rare diseases, showing real-world traction beyond theoretical discovery.
Key Features
Pros:
Cons:
Score: 4.5/5
View Now5. Evotec
Best for integrated AI plus automated biology in early discovery
Evotec sits at the intersection of advanced AI, automated biology, and large-scale drug discovery infrastructure. Unlike point-solution AI tools, Evotec operates more like an extension of a pharma R&D engine: target ID, screening, hit-to-lead, optimization, ADME-tox, and IND-enabling all sit under one roof.
Their strength comes from pairing in-silico modeling with high-throughput lab automation and deep therapeutic expertise across neuroscience, oncology, metabolic disease, kidney disorders, and anti-infectives.
Their portfolio of next-gen platforms – from machine learning–guided chemistry to automated wet-lab systems and PanOmics data layers – makes them a powerful option for pharma teams looking to compress early discovery timelines without giving up scientific rigor.
Add in partnerships with Novo Nordisk, Bristol Myers Squibb, Pfizer, Sandoz, and Just–Evotec Biologics, and they’ve built one of the most connected ecosystems in the industry.
Key Features
Pros:
Cons:
Score: 4.4/5
View Now6. Aizon
Best for AI-driven manufacturing optimization and GxP-compliant pharma production
Aizon is laser-focused on the regulated, day-to-day reality of pharmaceutical manufacturing. Their platform replaces paper-based decision-making with real-time, GxP-compliant AI models that help teams improve yield, spot deviations early, reduce batch failures, and tighten process control across biologics and small-molecule production.
What makes Aizon stand out is the combination of pragmatic AI, industry-specific workflows, and regulatory alignment. The platform is built for FDA-regulated environments, not retrofitted for them.
Their digital twin technology for bioreactors, multi-site data unification, and batch-level predictive analytics give manufacturing teams a practical path toward smarter, more consistent execution without months of custom engineering.
Key Features
Pros:
Cons:
Score: 4.2/5
View Now7. Scilife
Best for digital quality management and compliance automation in life sciences
Scilife is pragmatic and foundational: helping pharma, biotech, medical device, and ATMP companies build a quality system that keeps up with the scale and speed of modern operations.
Where most QMS tools are static and document-heavy, Scilife adds automation, traceability, risk management, training compliance, controlled printing, and design controls in one validated, 21 CFR 11–compliant platform. It’s built specifically for regulated life sciences teams that need audit readiness, cross-team accountability, and smart workflows without drowning in manual tasks.
For organizations struggling with siloed quality processes, recurring compliance gaps, or multi-site coordination issues, Scilife becomes a digital backbone – structured, traceable, and easy to scale.
Key Features
Pros:
Cons:
Score: 4.1/5
View NowComparison: Best AI Solutions for Pharma
How to Choose the Right AI Solution for Pharma
Choosing the right AI platform is all about finding the tool that fits how your teams work, what your data looks like, and where your biggest bottlenecks sit.
Here are the criteria that matter most and how each of the 7 stacks up:
1. Match the Solution to Your Real Use Case
Getting this wrong is the fastest way to waste time and budget.
Pharma has wildly different problems depending on the team: discovery needs generative chemistry; QC needs consistency; clinical needs data interpretation; manufacturing needs deviation prediction.
Best Fit by Use Case:
Less Good:
Anything outside each platform’s specialization. For example, Aizon won’t help with R&D, and Scilife won’t design molecules.
2. Check Data Readiness and Compatibility
AI only performs as well as the data it receives.
Discovery tools need multi-omics depth. QC tools need stable imaging. Clinical tools need clean unstructured data ingestion. Manufacturing AI needs connected plant data.
Platforms Good at Handling Complex Data:
More Sensitive to Data Conditions:
3. Evaluate Interpretability and Scientific Trust
Teams adopt AI faster when they can understand why the model made a call.
Strongest in Interpretability:
More Complex:
Insilico and Recursion – very powerful, but their generative/discovery models can feel abstract without strong internal science teams.
4. Consider Regulatory Alignment
Regulatory expectations shape every decision in pharma, so validation, traceability, and auditability matter more than raw speed.
Regulatory-Ready Leaders:
Less Suited Here:
Recursion and Evotec are strong scientifically but not QMS tools.
5. Integration and Workflow Fit
Pharma workflows are fragile. AI should slide in, not force a rebuild.
Best for Quick, Low-Disruption Integration:
More Involved:
Recursion, Evotec, and Insilico require deeper technical onboarding.
6. Scalability and Long-Term Fit
The best solutions grow with your pipeline, SKUs, or sites.
Highly Scalable:
Averroes.ai, Aizon, IBM Watson, Recursion, Evotec
More Use-Case Bounded:
Looking To Stabilize Manufacturing Outcomes?
Get 99%+ accuracy with fewer disruptions.
Frequently Asked Questions
How do you measure ROI when implementing an AI solution in pharma?
ROI usually comes from reduced cycle times, fewer failed batches, faster discovery loops, or lower compliance burden. Teams track before/after metrics like throughput, deviations, or lead-selection timelines to quantify real impact.
Do AI platforms replace scientists, operators, or QA teams?
No. These tools augment expertise rather than replace it. Scientists, engineers, and QA specialists still make decisions; AI simply accelerates analysis, flags issues earlier, or handles repeatable tasks with higher consistency.
How long does it typically take to deploy a pharma-ready AI system?
Deployment ranges from weeks to several months depending on the platform and use case. Manufacturing-focused solutions like Aizon or Averroes deploy fastest, while deep discovery engines such as Recursion or Insilico require more setup and integration.
Can these AI systems handle multi-site or global operations?
Most platforms scale across multiple sites, but the readiness varies. Averroes, Aizon, IBM Watson, and Scilife support multi-site orchestration well, while heavy scientific platforms may require additional infrastructure and governance to scale globally.
Conclusion
Pharma teams are under pressure to make smarter, faster decisions without sacrificing compliance, yield, or patient safety.
Each of these 7 AI solutions plays a different role:
If you’re exploring AI to improve inspection accuracy, stabilize processes, or reduce batch risk, book a free demo to see how Averroes supports pharma manufacturing teams.