AI21 post-trains specialized models on Nvidia's new Nemotron 3.5 Super VL for Gong, Deel, Alice
2026-10-15
AI21 announced on Oct 15, 2026 that its team of 'Forward-Deployed Researchers' spent three weeks post-training the recently released NVIDIA Nemotron 3.5 Super VL model to build specialized models for partners Gong, Deel, and Alice.
Significance 3: A named vendor's forward-deployed-researcher program customizing a brand-new Nvidia foundation model for named enterprise partners is a concrete data point on both the spread of the FDE staffing model and Nvidia's ecosystem reach beyond chip sales.
Implications · machine-drafted, not owner judgment
AI21's own forward-deployed-researcher model landing named customers on a brand-new Nvidia model is a second, smaller data point (after Anthropic/AWS/Atlassian) that the forward-deployed-engineer staffing pattern is spreading rather than contracting — directly relevant to the owner's own fde-overlay-model-durability thesis about a model they built and led. It also shows Nemotron reaching beyond chip sales into being a base model other labs tune, reinforcing Nvidia's ecosystem gravity.
- Whether AI21 discloses outcome/ROI data from these post-trained models, not just the partnership
- Whether Nemotron 3.5 Super VL shows up at other labs' FDE programs
Sources
- primary AI21 Partners with Gong, Deel and Alice to Post-Train Specialized Models on NVIDIA Nemotron 3.5 Super VL retrieved 2026-10-15