The Platform
One module, added to a CAR — built to address all three problems at once. NeXell's AI-driven platform integrates two layers of intelligence directly into the CAR system, in a single ultra-compact regulatory module of less than 600 base pairs. It is an all-in-one, CAR-agnostic approach, designed to improve safety, efficacy and durability simultaneously rather than trading one against another.
| Conventional CARs | CARs with NeXell Self Regulating Module | |
|---|---|---|
| Activity | Same activity everywhere, regardless of context | Scales with antigen context — low in normal tissue, full strength in the tumor |
| Toxicity | Higher CAR sensitivity can increase on-target, off-tumor toxicity | Graded by antigen density — spares healthy tissue with low antigen levels |
| Durability | Tonic signaling drives exhaustion over time | Off at rest — no tonic signaling, no baseline exhaustion |
| Tumor microenvironment | Immune-cell activity suppressed once inside the tumor | Enhanced immune-cell activity within the suppressive microenvironment |
NeXell Resolves the Safety-Efficacy Trade-Off
NeXell’s module autonomously adjusts CAR-T activity in response to antigen encounter, not a drug dose, helping resolve the trade-off shown above without manually turning a dial.
Validated Across CAR Constructs
Validated in FDA-approved CAR constructs in blood malignancies, and across multiple solid tumor indications. We hold specifics — targets, indications, model systems, and readouts — for discussion under a mutual confidentiality agreement; what's established here is that the module works across settings, not only in the one it was designed for.
NeXell Differentiating Factors
NeXell Design Engine:
Data Driven and AI-Guided
NeXell's pipeline is data-driven and AI-guided. Candidates are generated computationally from large biological datasets and machine learning–guided design, then prioritized against defined criteria before anything reaches the bench.
Every experimental result feeds back into the design rules, so the engine improves as it runs — including from the candidates that don't work. That loop is what makes the platform a repeatable engine rather than a single discovery.