Do all AI drug discovery platforms do the same work?

How much of your research can an AI drug discovery platform handle?

Drug researchers usually have far more candidates to investigate than they can test experimentally. Computation helps decide which molecules to make and which properties to examine first. Research that predicts molecular structures and properties or designs and evaluates candidates computationally is called in silico research. It includes physics- and statistics-based methods as well as AI.

Finding a good model does not make it immediately usable. Inputs need preparation, software needs a compatible environment, and results need to be compared with colleagues. Installation problems and computing resources add a substantial burden before research can even begin.

In silico R&D platforms take on different parts of this work. While building Nadoo Bio, we wanted to understand what researchers handle themselves and what they delegate to a platform. We examined official materials for 13 Korean and international examples. The label “AI drug discovery” covers a surprisingly wide range of roles.

Look at the work a platform handles before counting its models

Four roles: specialist computation, AI design, execution workflows, and collaborative research

Specialist computational environments focus on examining molecular structures and tuning calculation conditions. AI design tools propose new candidates based on desired criteria. Execution platforms connect tools with cloud resources to run calculations. Collaborative research services work alongside research teams on goals, candidate design, and experimental validation.

These distinctions change how a feature table should be read. “Candidate design” may mean software used directly by researchers or a service delivered with specialists. The required inputs, costs, and deliverables differ. Some specialist software runs in the cloud, and some companies offer both software and collaborative research.

Approaches available in Korea

Company or platform

Role described in official materials

What to examine when comparing

HITS · HyperLab

Small-molecule tools and protein design/structure prediction through Bio Co-Scientist

Web tools coexist with expert services. The small-molecule Core plan, K-Fold, and Bio Co-Scientist have different offering terms. [1]

GALUX · GaluxDesign

Collaborative research linking protein candidate design to experimental validation

Public partnership guidance proceeds from proposing a goal to feasibility assessment and joint design and validation. [2]

Standigm · STELLA Workspace

SaaS and APIs for generating and comparing candidates from a reference molecule

Direct-use software is offered alongside collaborative research. Examine candidate comparison and project-history coverage. [3]

Syntekabio · DEEPMATCHER / LM-VS / STB CLOUD

Structure-based candidate discovery and computational/analysis services

Target-specific discovery services and cloud access are distinct offerings. Terms and contract scope vary by product. [4]

Pharos iBio · Chemiverse

A research platform supporting target discovery and molecular design

Public materials describe internal research and open innovation. Direct access for external teams requires separate confirmation. [5]

A small molecule is a molecule smaller than a protein. A target is the biological entity researchers intend a drug to act on. Designing small molecules and designing protein candidates require different inputs and evaluation methods. The label “AI design” alone does not establish suitability for the same research problem.

Even within Korea, software access and collaborative research overlap. Comparing the actual product and contract your team would use is more accurate than classifying all Korean companies as collaborative services and all international companies as software providers.

What international platforms handle

Company or platform

Role described in official materials

What to examine when comparing

Schrödinger · Maestro / LiveDesign

Molecular modeling and collaboration integrating computational and experimental data

Specialist computation is combined with an environment for team data review. [6]

BIOVIA · Discovery Studio

Small-molecule/protein modeling and simulation

Discovery Studio Simulation also describes cloud collaboration. Scope varies by product and deployment form. [7]

Chemical Computing Group · MOE

A specialist environment combining visualization, design, computation, and analysis

Research methods can be assembled using its functions and extension language. [8]

OpenEye / Cadence · Orion

Cloud workflows connecting molecular computation and analysis

Examine both browser use and Python workflow extensibility. [9]

Insilico Medicine · Pharma.AI

Products for target discovery, molecular design, and clinical-trial prediction

Both standalone software and collaborative research are described. Products address different research stages. [10]

Iktos · Makya / Spaya

Molecular design and synthesis-route planning

Distinguish software from robotic synthesis and testing offered through separate collaborations. [11]

Tamarind

Web, API, batch execution, and pipelines for computational tools

Examine how tools connect, supported versions, and result delivery. [12]

Neurosnap

Model execution, result analysis, and visual pipelines

Provides both direct tool selection and ways to connect multiple tasks. [13]

Side-by-side, these examples show products combining specialist modeling, AI design, execution management, and experimental collaboration. Established platforms already continue to develop cloud access, connected tools, and teamwork. That makes our task clear: with many capable tools available, Nadoo Bio must show concretely which burdens it removes for researchers.

Research continues after the calculation

From a research question through input preparation and computation to the next experiment

Calculations inform the next experiment. Structure predictions and computational scores do not independently establish binding, biological function, or clinical effects.

After a calculation, a team still needs to decide which candidates to investigate, whether to change conditions, and which experiments to run first. Understanding and revisiting results matters as much as making computation easy to start.

Services that handle installation and GPU operations can reduce preparation work. That makes clear guidance on inputs, models, settings, and limitations all the more important. It is also why researchers who understand the biological question need to work with computational and software specialists. Joint interpretation lets a calculation become evidence for the next research step.

This is also why we started Nadoo Bio. We wanted to connect Nadoo Cloud's computing technology with research tools so that researchers unfamiliar with computational environments could use the calculations they need. The next article compares platform responsibilities with research-team responsibilities in protein design and small-molecule screening.

References

These official product and usage materials were checked on October 26, 2026. Feature availability depends on the product and contract.

  1. HITS. HyperLab, Bio Co-Scientist, Pricing.

  2. GALUX. Partner with us.

  3. Standigm. STELLA Workspace, AI SaaS.

  4. Syntekabio. DEEPMATCHER, LM-VS, STB CLOUD.

  5. Pharos iBio. Chemiverse.

  6. Schrödinger. LiveDesign, Maestro.

  7. BIOVIA. Discovery Studio Simulation.

  8. Chemical Computing Group. MOE Products.

  9. OpenEye. Orion.

  10. Insilico Medicine. Pharma.AI.

  11. Iktos. Products and research workflows.

  12. Tamarind Bio. Features.

  13. Neurosnap. Products and pipelines.

Structure sources for the illustrations

Explanatory illustrations based on coordinates from RCSB PDB 1UBQ and 1STP, with colors and arrangements adapted.