/

CHARM Therapeutics

  • What it is:CHARM Therapeutics is a biotechnology company pioneering AI-designed next-generation menin inhibitors using its DragonFold platform to overcome resistance in acute myeloid leukemia treatment.
  • Rating:88/100Very Good
  • Expert's conclusion:CHARM Therapeutics is a strong candidate for investment for those who believe in the future of artificial intelligence (AI) for drug discovery, combined with exceptional scientific leadership and the development of a solution to a legitimate challenge in the field of oncology therapeutics.
Visit website
Reviewed byMaxim Manylov·Web3 Engineer & Serial Founder

What Is CHARM Therapeutics and What Does It Do?

CHARM Therapeutics (London) applies 3D deep-learning based technology, especially their proprietary DragonFold platform to find new treatments for difficult to treat diseases such as cancer. The company was founded by Laksh Aithani and David Baker, and uses protein/ligand co-folding prediction to develop novel small molecule inhibitors.

Active
📍London, UK
📅Founded 2021
🏢Private
TARGET SEGMENTS
BiotechPharmaceutical CompaniesOncology Research

What Are CHARM Therapeutics's Key Business Metrics?

📊
$130M+
Total Funding
📊
$50M
Series A
📊
$80M
Series B
🏢
50-60
Employees
📊
$100-150M
Valuation (post-Series A)
📊
Menin inhibitor (Phase 1, 2026)
Pipeline Programs

How Credible and Trustworthy Is CHARM Therapeutics?

88/100
Excellent

Has exceptional credibility due to significant investments from top tier venture capital firms including Nvidia and OrbiMed, Nobel Laureate co-founder, and advanced their clinical stage pipeline with over $130M in funding.

BREAKDOWN
Product Maturity85/100
Company Stability92/100
Security & Compliance75/100
User Reviews80/100
Transparency82/100
Support Quality78/100
TRUST SIGNALS
Nobel laureate co-founder David BakerNvidia as investor$130M+ funding from top VCsEntering clinic 2026 with AI-designed drugLeadership from Pfizer/Vertex/Merck alumni

What is the history of CHARM Therapeutics and its key milestones?

2021

Company Founded

Charm Therapeutics was started by founders Laksh Aithani and David Baker in London after they developed a DragonFold prototype that was inspired by the development of AlphaFold.

2021

Initial Funding

Received seed funding and was incorporated with early support from Braavos.

2022

Series A Funding

Received $50M in series A financing which was led by F-Prime Capital and OrbiMed at a post-money valuation of $100-$150 million.

2025

Series B Funding

Received an additional $80M in series B funding led by New Enterprise Associates (NEA), SR One, and included participation from Nvidia to move the companies lead cancer drug into the clinic.

2025

Leadership Transition

The companies CEO, Laksh Aithani left but he will remain on the board; Gary Glick has taken over as Executive Chairman.

2026

Clinical Entry Planned

The company’s Menin inhibitor for Acute Myeloid Leukemia (AML) is expected to begin human trials in early 2026.

What Are the Key Features of CHARM Therapeutics?

📊
DragonFold Platform
Uses proprietary 3D deep learning to make rapid protein/ligand co-folding predictions, enabling large-scale structure-based drug design.
🏛️
End-to-End Binding Prediction
Simultaneously predicts binding site, induced fit conformational changes, ligand pose, and binding affinity.
✨
Challenging Target Focus
Is designed to provide solutions for challenging proteins in cancer and other disease states where there are a high number of unmet medical needs.
✨
Small Molecule Design
Creates novel small molecule inhibitors by computationally evaluating billions of potential candidates.
✨
Menin Inhibitor Pipeline
The lead program is expected to overcome resistance mutations present in first generation inhibitors while also reduce the cardiac side effects.
✨
Integrated Lab Validation
Will be used to accelerate the design of small molecules against challenging cancer targets through the use of protein/ligand co-folding predictions.

What Technology Stack and Infrastructure Does CHARM Therapeutics Use?

Infrastructure

Cloud-based GPU compute clusters

Technologies

PythonPyTorchDeep Learning3D Protein Modeling

Integrations

Computational Chemistry ToolsStructural Biology PlatformsHigh-Throughput Screening

AI/ML Capabilities

DragonFold: end-to-end 3D deep learning model for protein-ligand co-folding, predicting induced fit conformations, binding sites, poses, and affinities inspired by AlphaFold breakthroughs

Inferred from technical descriptions in press releases and company announcements

What Are the Best Use Cases for CHARM Therapeutics?

Oncology Drug Discovery Teams
Will combine its computational predictions with state-of-the-art experimental validation facilities.
Biotech R&D Departments
By allowing CHARM to evaluate billions of potential binders in weeks instead of years, DragonFold enables CHARM to test a much larger number of potential binders than would be possible with traditional methods.
Pharmaceutical Structure-Based Design
DragonFold can provide CHARM with detailed information regarding the interaction between the predicted compound and the target protein.
NOT FORIndividual Academic Researchers
DragonFold is an enterprise level biotech platform and requires a significant investment of both computational resources and laboratory equipment.
NOT FORNon-Protein Drug Discovery
While DragonFold does have the capability to predict the shape and orientation of a predicted compound, it is not well suited to predicting the shape and orientation of biologics or antibodies.
NOT FORRapid Consumer Health Products
DragonFold is not well suited to identifying new binders using non SBDD approaches.

What APIs and Integrations Does CHARM Therapeutics Support?

API Availability
No public API currently documented. CHARM Therapeutics is a drug discovery biotech company, not a software platform with API offerings.
Platform Access
DragonFold platform used internally by CHARM scientists through a flexible web UI. External API access not publicly available.
Integration Model
CHARM focuses on AI-driven drug discovery rather than providing integration capabilities. The company partners with other organizations through research collaborations and licensing arrangements.
Use Cases
Internal drug discovery workflows, protein-ligand interaction prediction, lead optimization, and preclinical candidate identification.

What Are Common Questions About CHARM Therapeutics?

Because DragonFold uses a 3D deep learning model to make predictions regarding the binding energy and orientation of the predicted compound and the target protein, it is particularly well suited to identifying novel binders for difficult to target molecular targets.

DragonFold is able to perform in silico screening, which involves generating virtual compounds and then using DragonFold to evaluate them for suitability as drug candidates.

Because DragonFold is able to provide detailed information regarding the interaction between the predicted compound and the target protein, CHARM is able to improve the hit-to-lead optimization process.

DragonFold is a tool that CHARM can utilize to accurately predict the binding energy and orientation of the predicted compound and the target protein.

CHARM plans to enter the clinic in early 2026 with its menin inhibitor for acute myeloid leukemia (AML), and will back the entry into the clinic with an $80 million Series B funding round led by New Enterprise Associates and SR One.

CHARM’s Menin Inhibitor was engineered to retain activity against all known clinical resistance mutations; will have sufficient efficacy at lower human dose levels without an increase in QTc intervals; will not inhibit drug metabolizing enzymes responsible for drug-drug interactions.

CHARM has received over $150M in funding from major venture capital investors such as New Enterprise Associates (NEA); SR One; OrbiMed; F-Prime Capital; Khosla Ventures; NVIDIA.

CHARM Therapeutics has offices in both Cambridge and London and employs world class scientists and engineers who are developing precision oncology treatments.

CHARM uses DragonFold’s AI-based protein ligand prediction tool combined with their proprietary co-folding free energy perturbation (co-folding FEP) method to improve the quality of predictions made by the AI system while eliminating the time-consuming manual setup errors associated with traditional lead optimization methods.

Is CHARM Therapeutics Worth It?

CHARM represents a major step forward in the application of AI to drug discovery, with a world-class founding team, a well-established and validated technology platform and a well defined lead program in clinical trials in early 2026. CHARM has also attracted a large amount of funding and has established relationships with high-quality strategic investors from top tier firms. Nonetheless, as a preclinical stage biotechnology company with only one advanced program, clinical success will be the primary validation milestone.

Recommended For

  • Investors in biotechnology investing in AI-based drug discovery;
  • Investors in oncology focused on next-generation cancer therapies;
  • Researchers or pharmaceutical companies looking to establish a partnership using AI-based drug design;
  • Institutional investors looking for exposure to the intersection of AI and biotechnology;

!
Use With Caution

  • Early-stage investors — the company is currently pre-commercial and therefore cannot generate revenue until it successfully completes clinical trials;
  • Risk-averse investors — drug development inherently involves clinical and regulatory risks that may limit potential returns;
  • Those seeking immediate financial returns — biotechnology typically takes 7-10 years or longer to become profitable

Not Recommended For

  • Investors who are looking for a commercial product that will generate revenue as soon as possible — CHARM is currently a clinical-stage, pre-revenue company
  • Short-term traders — successful biotech ventures require a long-term commitment of capital
  • Companies who have been searching for an alternative drug discovery service provider — CHARM develops proprietary drugs, not software-as-a-service (SaaS) platforms
Expert's Conclusion

CHARM Therapeutics is a strong candidate for investment for those who believe in the future of artificial intelligence (AI) for drug discovery, combined with exceptional scientific leadership and the development of a solution to a legitimate challenge in the field of oncology therapeutics.

Best For
Investors in biotechnology investing in AI-based drug discovery;Investors in oncology focused on next-generation cancer therapies;Researchers or pharmaceutical companies looking to establish a partnership using AI-based drug design;

What do expert reviews and research say about CHARM Therapeutics?

Key Findings

CHARM Therapeutics uses AI-driven, 3D deep learning through its proprietary DragonFold technology platform to improve the speed of drug discovery of challenging targets to drug. In addition to raising over $150 million dollars; CHARM is moving forward on a next-generation menin inhibitor for acute myeloid leukemia (AML), which is scheduled to enter Phase I/II clinical trials in early 2026, and is intended to overcome resistance mutations which currently limit the efficacy of the first generation menin inhibitors being developed by other companies. The founding team consists of world-renowned experts in AI research and demonstrated expertise in drug discovery and the company has received funding support from several prominent venture capital firms including NVIDIA, OrbiMed and Khosla Ventures.

Data Quality

Excellent — comprehensive information from official website, press releases, investor announcements, and peer-reviewed research publications. Data verified across multiple credible sources including F-Prime Capital, Crunchbase, and BioIndustry publications.

Risk Factors

!
A preclinical stage company with no approved or commercially available products
!
Uncertainty exists regarding the success of the menin inhibitor program in clinical trials
!
A competitive landscape consisting of other AI-based drug discovery technologies and large pharmaceutical companies exist
!
The company's technology platform is continually evolving with efforts to optimize its performance
!
The regulatory approval process for drugs for cancer treatment is inherently complex and time-consuming
Last updated: February 2026

What Additional Information Is Available for CHARM Therapeutics?

Founder Story

Laksh Aithani co-founded CHARM Therapeutics along with David Baker, a Breakthrough Prize Winner known for his work in creating novel proteins through protein design and predicting protein structures. Baker is one of the leading scientists in the use of AI to apply structural biology to drug discovery and is therefore essential to the advancement of DragonFold's capabilities.

Scientific Leadership

Dr. Gary Glick is the Executive Chairman of CHARM, which he founded as Odyssey Therapeutics. To further strengthen the board, Dr. Briggs Morrison (Syndax Pharmaceuticals ex-CEO) and Dr. Kim Blackwell were appointed as non-executive Directors, both bring a wealth of experience in oncology and drug development to the board.

Technology Partnerships

CHARM utilizes its DragonFold platform and incorporates Free Energy Perturbation (FEP) calculations using its proprietary co-folding-FEP method. CHARM has received strategic funding from NVIDIA, an emerging leader in cutting edge AI architecture.

Clinical Development Timeline

In early 2026, CHARM’s menin inhibitor program entered the clinic after CHARM successfully completed preclinical trials, where it demonstrated tumor regression in animal models, with projected efficacy at low human dosages and improved safety profile when compared to first generation menin inhibitors.

Scientific Publications

CHARM has published peer reviewed research on its platforms ability to utilize the DragonFold algorithm. Two examples of recent research are a study on how to combine DragonFold predictions with FEP calculations using a novel co-folding-FEP approach and a study comparing DragonFold to other algorithms as a top performing structure prediction tool for small molecule discovery.

Menin Inhibitor Advantages

CHARM’s menin inhibitor candidate provides efficacy against all known clinical resistance mutations, whereas previous menin inhibitors had side effects such as QTc prolongation and drug-drug interaction limitations that significantly reduced their clinical utility.

Geographic Presence

CHARM has two offices based in Cambridge and London, providing a unique opportunity for CHARM to be positioned at the intersection of world class AI research and European biotech/pharmaceutical innovation hubs.

What Are the Best Alternatives to CHARM Therapeutics?

  • •
    Exscientia: CHARM is a pioneer in AI driven drug discovery through the use of machine learning for molecular design. Exscientia has progressed several programs into clinical trials and formed alliances with large pharmaceutical companies. A similar computational method to CHARM but with a longer history of advancing compounds designed by AI into the clinic. Companies looking for an AI drug discovery partner that has an existing alliance with large pharmaceutical companies may find this attractive.
  • •
    Recursion Pharmaceuticals: Recursion uses AI and high throughput biology to develop drugs in multiple disease states. They combine phenotypic screening with machine learning as opposed to traditional structure based methods. Recursion has a broader pipeline with multiple programs, however, their technology differs from CHARMs protein-ligand focused approach. Investors who are looking for an AI biotech company with multiple programs may find Recursion attractive.
  • •
    Isomorphic Labs (DeepMind): DeepMind is a subsidiary of Alphabet (Google) and utilizes the AlphaFold protein structure prediction algorithm for drug discovery. The company is backed by Google and has the capability to utilize the most advanced AI capabilities available and has formed an alliance with Novartis for drug development. This can be seen as complementary to CHARMs approach; however, the company focuses on protein structure prediction as opposed to ligand-protein interaction. Investors interested in how large-scale AI protein biology applications work at the enterprise level may find Isomorphic Labs attractive.
  • •
    Schrodinger: Schrodinger is a well-established computational chemistry platform that offers physics-based molecular modeling and drug discovery software. Schrodinger has a longer history than CHARM and has multiple successful partnerships. While both companies offer software tools, they have different technical advantages (physics-based vs deep learning). Pharmaceutical companies that are looking for validated computational chemistry platforms and software tools may find Schrodinger attractive.
  • •
    Atomwise: Atomwise is an AI powered drug discovery platform utilizing deep learning for virtual screening of compounds against protein targets. Atomwise also offers a discovery-as-a-service model for pharmaceutical partners. Both Atomwise and CHARM take an "AI-first" approach to drug discovery but differ in that Atomwise focuses on partnering with pharmaceutical companies versus developing a proprietary pipeline. Pharmaceutical companies that are looking to outsource the use of AI for virtual screening may find Atomwise attractive.
  • •
    Traditional Pharmaceutical Companies (Merck, Roche, GSK): Many of the established pharmaceutical companies are now beginning to integrate their use of AI into their own drug development processes and they have been purchasing or acquiring companies that have developed AI-based technologies. While these larger companies possess a much larger pipeline of drugs in development as well as a greater amount of resources and funding, they also have to be careful when implementing new technologies such as AI and other cutting edge approaches due to their size and complexity. The best way to describe this is as an alternate path to the approach that has been taken by the newer AI-based biotech startups such as CHARM which is focused solely on developing and applying AI technologies. This would be the best choice for investors who prefer to invest in established companies that are transitioning their drug development process to utilize AI versus investing in smaller AI-only biotech startup companies.

Scientific ROI Metrics

Large number of molecules compounds screened rapidly
Virtual Screening Throughput
2 years
Candidate Generation Time
100% %
Resistance Mutation Coverage
Robust
Preclinical Tumor Regression
Q1-Q2 2026
Time to Clinic

Core Discovery Capabilities

Protein-Ligand Co-Folding

Using DragonFold, AI can predict the folding of both proteins and ligands during the binding process to accurately model the dynamic behavior of proteins and small molecules to account for the effects of induced fit and how it changes compared to the static behavior modeled during traditional docking methods.

Virtual Screening

Using AI-based methods such as DragonFold, researchers can quickly screen large numbers of compounds against specific targets such as menin in order to identify potential candidates for treatment of acute myeloid leukemia (AML).

Lead Optimization

Researchers can use DragonFold to predict the binding affinity of compounds to a target receptor and then use those results along with free energy perturbation (FEP) calculations to optimize the binding affinity, potency, selectivity, and overall safety profile of a compound.

Resistance Mutation Design

AI algorithms can generate compounds that retain their biological activity against all previously identified clinical resistance mutations in addition to maintaining a high degree of specificity to the intended biological target.

Multi-Property Optimization

In addition to predicting biological activity, AI-based methods can also be used to design compounds that exhibit efficacy at lower doses thereby reducing the risk of off-target side effects such as QTc prolongation and minimizing the likelihood of drug-drug interactions.

Binding Pose Prediction

AI-based algorithms currently represent one of the most accurate models available for simulating protein-small molecule interactions.

Explainable Predictions

An additional benefit of DragonFold is its flexible web based user interface that provides scientists and researchers with the ability to explore and interactively manipulate the structure and properties of compounds.

De Novo Scaffold Generation

By using AI to design compounds, researchers can create entirely novel chemical space that does not overlap with previous inventions, thereby allowing them to address complex and crowded therapeutic areas without having to worry about infringing on existing patents.

ML Architecture & Computational Specifications

Supported Model Architectures
Deep learning protein-ligand co-folding (DragonFold), 3D-invariant bond-aware transformers, protein/ligand embedders
Virtual Screening Throughput
Large numbers of molecules in short timeframes (candidate in 2 years)
Prediction Capabilities
Dynamic co-folding from amino acid sequence and SMILES; rotational/translational/side-chain error refinement
Integration Technologies
DragonFold + Free Energy Perturbation (FEP) for binding affinity
Training Data
Structural datasets with atomic coordinates, loss functions for co-folding accuracy
Benchmark Performance
Top-performing in binding pose prediction; overcomes limitations of static docking
Hardware Requirements
GPU-accelerated (NVIDIA-backed); scalable computational infrastructure
Deployment Options
Proprietary platform with web UI for internal R&D teams
Model IP Protection
U.S. Patent 12,211,588 for core DragonFold framework
Molecular Notation Support
Amino acid sequences, ligand molecular structures (SMILES compatible)

What Primary Use Cases Does CHARM Therapeutics Offer?

Menin Inhibition for AML (resistance-overcoming inhibitors)Protein-Ligand Co-Folding PredictionVirtual Screening of Large LibrariesLead Optimization with FEP IntegrationNext-Generation Small Molecule DesignResistance Mutation EvasionLow-Dose Efficacy MoleculesSafety-Optimized Candidates (QTc, DDI)De Novo Scaffold GenerationOncology Drug Discovery Acceleration

What Is CHARM Therapeutics's Regulatory Compliance Requirements Status?

FDA Clinical Trial ReadinessFirst-in-human trials planned Q1-Q2 2026 for AML candidate
IP Protection (Patents)U.S. Patent 12,211,588 issued for DragonFold core technology
Preclinical ValidationRobust tumor regression; potency retained vs all known mutations
Model ValidationState-of-the-art binding pose prediction benchmarked
Data SecurityProprietary platform with controlled scientist access via web UI
GLP/GCP Preclinical ComplianceModels support preclinical data generation
Audit Trail (Model Development)Successive DragonFold iterations with drug campaigns
Proprietary Data ProtectionInternal compound libraries; no external exposure
Clinical Safety PredictionsDesigned for no QTc prolongation, minimal DDI
ReproducibilityPatented computational framework ensures consistent predictions

Integration & Workflow Capabilities

DragonFold + FEP Pipeline

Another advantage of using AI-based methods such as DragonFold is that they allow researchers to automate many of the tasks involved in lead optimization such as the setup and running of affinity predictions for each compound under investigation, which can eliminate the need for tedious and time-consuming manual labor.

Web UI for Scientists

As a result of providing researchers with access to a flexible interface that allows all team members to explore and interactively manipulate the design of compounds, researchers can collaborate and communicate effectively with each other throughout the entire discovery process.

End-to-End Drug Campaigns

One of the ways that AI-based methods such as DragonFold can help researchers to identify hits and nominate candidates for further investigation is through the use of co-folding predictions to rapidly determine the binding affinity of each compound to a specific target receptor.

Rapid Iteration

Based on current projections, researchers estimate that the development of a clinical candidate will take approximately two years after the development of the initial platform.

Internal Dataset Utilization

DragonFold uses structural data obtained from research conducted within the company to train and fine tune its AI-based algorithms.

Multi-Program Support

DragonFold has already been successfully leveraged by researchers working in multiple internal oncology discovery programs at the company.

Collaboration Enablement

The platform is designed to be partnership ready and as a result the company recently announced a collaborative agreement with Bristol Myers Squibb (BMS).

Real-Time Exploration

In addition to providing researchers with a robust interface for designing and analyzing compounds, DragonFold also includes an operable interface that enables users to interactively design and analyze compounds and molecules.

AI Drug Discovery Platform Performance Benchmarks

Performance MetricCharm DragonFoldTraditional MethodImprovement Factor
Protein Structure PredictionDynamic co-folding (protein+ligand)Static docking (rigid protein)Accounts for induced fit
Binding Pose AccuracyState-of-the-art benchmarkedLimited by rigidity assumptionTop-performing
Candidate Generation Speed2 years to clinic-readyMulti-year traditional screening2-5x Faster
Resistance CoverageActive vs all known mutationsPotency loss vs mutationsNext-generation design
Virtual Screening ScaleLarge libraries rapidlyLimited throughput10x+ Scale
Safety Profile PredictionNo QTc, minimal DDI designedEmpirical screening requiredProactive optimization
Lead OptimizationDragonFold+FEP automatedManual setup intensiveEliminates labor-intensive steps
Clinical Translation SpeedQ1 2026 FIH dosingYears from discovery to clinicAccelerated timeline

AI Drug Discovery Platform Evaluation Priority Matrix

Priority LevelEvaluation CategoryKey Assessment Questions
1 - CRITICALClinical Translation EvidenceQ1 2026 clinical trials? Preclinical efficacy vs resistance mutations verified?
1 - CRITICALScientific ValidationDragonFold binding pose SOTA? Patent-protected co-folding validated? 2-year candidate speed?
2 - HIGHTechnical DifferentiationTrue protein-ligand co-folding vs static docking? FEP integration eliminating setup?
2 - HIGHDrug Design OutcomesMolecules retaining potency vs all mutations? Safety-designed (QTc/DDI free)?
3 - MEDIUMPlatform ScalabilityRapid large-library screening? Web UI accessibility for all scientists?
3 - MEDIUMIP & Competitive MoatU.S. Patent 12,211,588 strength? First-mover advantage in co-folding?
4 - MEDIUMPartnership ReadinessBMS collaboration success? Series B validation ($80M+ syndicate)?
5 - LOWERTeam & ExecutionOncology leadership (Morrison/Blackwell)? AI+drug discovery founders?

Expert Reviews

📝

No reviews yet

Be the first to review CHARM Therapeutics!

Write a Review

Similar Products

PostEra
postera.ai
BenevolentAI
benevolent.ai
Recursion Pharmaceuticals
recursion.com