Science

We read what a cell does.

QMI measures hundreds of protein interactions inside living cells at once. It is how we see what a CAR design will really do.

A CAR can bind its target and still fail.

AI can now design binders in days, and binding is easy to test. What the cell does next is not.

It depends on which proteins gather around the CAR inside the cell. Standard lab tests often miss those differences. QMI measures them directly.

“…the only preclinical measure that reliably differentiated the two products.”

Summers et al., on the protein interactions QMI measured in two CAR-T cell products. Blood Immunology & Cellular Therapy, 2026.

How QMI works

Hundreds of interactions. One small sample.

01 Collect

We take a small sample of engineered cells, resting or just after they meet their target, and open them gently so protein complexes stay intact.

02 Capture

Color-coded beads, each carrying an antibody, pull out one protein apiece along with whatever is bound to it.

03 Detect

A second antibody shows which partner came along. Each bead and probe pairing reads one protein interaction.

04 Map

A flow-based reader measures hundreds of pairings at once. Together they form a quantitative map of the cell’s signaling network.

QMI stands for Quantitative Multiplex co-Immunoprecipitation.

  • 400+protein interactions read in every sample
  • 4peer-reviewed papers
  • 10 yearsof QMI research, since 2016

Every reading makes the AI smarter.

Each QMI run turns a design into a detailed signaling profile. Our models learn from every profile, predict which designs will be safe and potent, and pick the next ones to build.

Built for any living cell.

QMI has already read protein networks in T cells and neurons. The same method works in NK cells, macrophages and therapies made from stem cells. CAR-T is where we start.

  • T cells & CAR-T
  • NK cells
  • Macrophages
  • Neurons
  • iPSC-derived cells

Let’s talk.