TRUSTED BY THE BEST
Scientific teams at Novo Nordisk, Johnson & Johnson, and IFF all use Cradle to engineer proteins faster.
"Cradle’s AI-based protein design platform provides easy access and acceleration to protein optimization for the discovery pipeline."
Eswar Narayanan
Director, Corteva Catalyst
Case studies
PROTEIN ENGINEERING 2.0
Compounding results
Move from incremental progress to compounding results with models that learn from every round of your wet lab data. Teams that use Cradle report 1.5-12x faster development timelines.
Breakthroughs on repeat
Following a single path puts you at risk of hitting a dead end. Instead, confidently explore multiple design strategies to bring more programs over the finish line and strengthen your IP portfolio.
Co-optimize all your properties
Activity, binding, stability, expression, and more. Co-optimize the properties you care about, turning complex trade-offs into optimized solutions in fewer rounds.
WORK ON ANY PROTEIN
Accelerate your journey from hit identification to lead optimization and beyond. If you can measure it, you can optimize it with Cradle.
This makes Cradle fit-for-purpose for teams developing therapeutics, agricultural solutions, food ingredients, and more.
HOW CRADLE WORKS
Design in Cradle.
Validate in the lab.
Quickly generate lab-ready protein candidates to test in your own lab or with a CRO. Track model and candidate performance in your reports. Learn every time you add assay data.
To begin, upload wet lab data from screening or optimization experiments. If you're starting a project from scratch, simply define your goals and assays to get going.
Cradle automatically analyses your experimental data to determine data quality and AI’s ability to learn from it. This helps you to understand areas to improve your workflows and assays to get to optimal sequences faster.
Every time you upload lab results, Cradle gets better at understanding your protein. This means increasingly performant variants that get you closer to your goals— every round.
The game-changer of Cradle: new, optimized candidates at the click of a button. Choose a design strategy, set sequence- or lab based constraints, bring in your knowledge of the protein and hit ‘Generate’. No PhD in computer science needed.
Cradle learns the complex interactions between all properties of your protein. During generation, this translates into candidates that balance improved performance across properties with high scores on diversity and novelty
The sequences that you download from Cradle are a carefully curated set, selected from millions of in silico designs to maximize your chances of success.
Generation Reports give you full control over what candidates you bring to the lab and what to expect in performance. See predictive performance scores for your entire plate and explore specific sequences and mutations in 3D.
Now it’s time to turn bits into atoms. Download sequences from Cradle, then order DNA and run assays in your own lab. Or send the sequences to your favorite CRO.
Track lab progress in Cradle using Round Statuses and see live metrics per protein property as new assay data comes in from testing.
PRIVATE. SECURE. YOURS
You experimental data is only used to train your organizations custom models.
Cradle is fully SOC 2 Compliant with active monitoring and tight security protocols in place.
Secure and simple account management with SSO support for Google and Microsoft.
Use internally developed predictors right in Cradle and see outputs inside Reports.
Fully managed AI infrastructure with optimized GPU that scales with your need.
Our team of scientists and ML experts are ready to assist with any issues.
THE CRADLE LAB
AI that gets better
every two weeks.
Cradle isn’t a CRO, so why do we run our own wet lab in Amsterdam?
It's simple. So you get reliable and high-performing models from day one, while your proprietary data remains completely private.
Much like a self-driving car company: they wouldn’t just design algorithms and cross their fingers. They need to test real-world scenarios and collect data in order to fine-tune performance. We take the same approach, ensuring our models are performant even before seeing your data.
