InsAIght

Growing Intelligently

AI | Biotechnology | Pharmaceuticals | Synthetic Biology

Description

InsAIght specializes in custom protein design through Artificial Intelligence. It is currently working to bring to market tailored protein design services for the pharmaceutical and biotechnology industries.

It also aims to complement its services with regulatory support and guidance throughout the biomolecule patenting process.

Its services focus on enabling their clients to manage their resources more efficiently, significantly reducing the costs and timelines of R&D processes for their protein-based products, and strengthening their business models by helping them bring biopharmaceuticals to market faster, at lower cost, and protected by patents with longer effective exploitation periods.

Since its founding in 2024, InsAIght has been developing its Artificial Intelligence platform. At present, it consists of several key components: StructureGPT and the XSeq systems.

StructureGPT is a Generative Pretrained Transformer that maps a protein’s structure to its sequence. It has two main capabilities: generativity and inpainting. Generativity allows it to associate a single protein structure with multiple sequences that either preserve the original function but have improved manufacturability (stability, solubility, expression in cells), or offer enhanced functional properties (e.g., higher catalytic activity in enzymes, greater affinity and/or specificity in antibodies, etc.). Inpainting allows StructureGPT to redesign parts of a protein, or even design it from scratch, accelerating the development of complex molecules like chimeric receptors (CAR-T), or enabling the design of ad hoc proteins for challenges where no natural candidates exist.

The XSeq systems are a family of AI models that evaluate the sequences generated by StructureGPT and filter them, enabling the selection of those best suited to the client’s problem. This is possible because each XSeq model predicts a property related to the sequence, which defines its name. For example, ΔΔGSeq predicts the thermodynamic stability (ΔΔG) of a protein based on its sequence.

Together, this platform enables the generation and selection of improved versions of its clients’ proteins. However, InsAIght knows this is not enough. Its clients need assurance that its designs will work in the lab. For this reason, it is working to establish an experimental laboratory that will allow the company to test its designs and transfer only those that are proven to work.

Featured projects

Founding of InsAIght in early 2024:
After a short period of developing the first versions of StructureGPT (StructureGPT-1.0 and 1.5), and with promising theoretical results, the InsAIght team succeeded in convincing its first investors to establish the company with an initial budget of €500,000.
Admission to the CEIN business incubator in mid-2024:
A few months after its founding, we managed to convince the CEIN team of the value of our business proposal, and we were granted access to their facilities and training programs. The interaction with other incubated companies and the training provided by CEIN have greatly accelerated our activities.
Development of StructureGPT-1.6 and 1.7 in 2025:
Just a year and a half later, at InsAIght they have successfully developed two more powerful versions of StructureGPT, both with 226 million parameters (about 1/7 the size of GPT-3, the first version of ChatGPT).

INSAIGHT

Description

Featured projects

WITH THE SUPPORT OF

Ajustes de privacidad

Con el deslizador, puede habilitar o deshabilitar los diferentes tipos de cookies:

Este sitio web hará:

Este sitio web no: