Building the self-driving lab for drug discovery

Navexis lets drug discovery teams screen and optimise candidates on a computer first. Laboratory budgets then go to the molecules most likely to succeed.

Request the deck See how it works
Every molecule, screened on a computerThe few worth making

AI is changing drug design. Labs are still catching up.

AI now generates potential drug candidates at record rates. Scientists still review, synthesise and test them by hand, across separate tools and outsourced labs. Every candidate that reaches the bench costs time and money.

  • A$3.1BAverage cost of bringing one successful drug to market
  • 15 yearsAverage time from target to approval
  • 1 in 10Candidates entering human trials are approved

Sources: Deloitte, Measuring the Return from Pharmaceutical Innovation (2025); Paul et al., Nature Reviews Drug Discovery (2010); BIO, Clinical Development Success Rates (2021).

One workflow, from molecule to decision

Vertically integrated software for early-stage drug discovery. Scientists screen, compare and optimise candidates in one place, and it is live today.

  1. The Navexis screening view, showing toxicity readouts and a prediction radar for one molecule

    1Screen

    Predict how the body absorbs, distributes and clears each molecule, and how safe it is.

  2. The Navexis comparison view, showing endpoints selected for a parent molecule

    2Compare

    Rank a whole batch side by side and see which candidates lead.

  3. The Navexis optimisation view, showing a batch comparison chart and chemical space projection

    3Optimise

    The platform proposes improved versions of the strongest molecules.

Ranked first globally among models built without proprietary data

In the 2026 OpenADMET ExpansionRx blind challenge, Navexis placed in the top five overall of more than 370 teams, and first among models trained without proprietary data.

  • 10chemists testing the platform
  • 3active pilot labs
  • 10,000+molecules predicted

From prediction to a self-driving lab

Each level brings another stage of discovery into the same system, until Navexis designs, makes and tests molecules itself.

  1. Level 1

    Screen and optimise

    ADMET, toxicity and optimisation.

    Live today
  2. Level 2

    Simulate and learn

    Virtual animal modelling (PBPK), peptides and active learning.

  3. Level 3

    Orchestrate the tools

    An agentic layer runs the right engine for each chemistry. Biologics.

  4. Level 4

    Close the loop with robots

    Partner robotic labs make and test what Navexis designs.

  5. Level 5

    Autonomous discovery

    A Navexis-run robotic lab, with scientists setting the goals.

Better with every customer

  • Accurate predictions from less data

    The companies with the most data used to have the advantage. Our proprietary methodology makes limited data go further.

    Today

  • Federated learning

    The model improves with every customer, while their underlying IP stays with them.

    Being added

  • Physics and biology, integrated

    Physics-based and biology-based prediction, each informing and improving the other.

    Over time

We are raising a pre-seed round

Help us accelerate drug discovery, so new medicines can reach patients sooner.

Request the deck hello@navexislabs.com