The Bad PlaceBusiness Latest | Former Google and Apple Researchers Launch a Startup to Build AI’s Missing Feedback Loop by Maxwell Zeff
AI generated summary, Read the full article for complete information.
Former Google DeepMind, Apple, OpenAI, and Meta researchers have launched Trajectory, a startup aimed at creating a platform for continuous AI learning by regularly training models on real‑world user interactions. Co‑founders Ronak Malde, Arjun Karanam, and Michael Elabd—who previously worked on Google DeepMind’s coding initiatives, Apple’s Vision Pro, and DeepMind robotics—raised a $15 million seed round at a $115 million post‑money valuation, backed by Conviction, Bessemer, Radical VC, BoxGroup, Jeff Dean, and Fei‑Fei Li. Trajectory’s approach starts with open‑source models that are post‑trained on domain‑specific data (e.g., AI customer‑support agents) and updated as frequently as weekly, aiming to outperform static models from leading labs on narrow, business‑critical tasks. By automating this feedback loop, the company hopes to eliminate the need for dedicated “forward‑deployed” engineers, already serving AI‑native firms such as Clay and Harvey and eventually targeting larger enterprises. Critics note the current update cadence is still periodic rather than truly real‑time, but the founders envision a future where AI models can be refreshed hourly—or even after each interaction.
#Trajectory #Conviction #Bessemer #business_artificialintelligence #modelbehavior #RonakMalde #ArjunKaranam #MichaelElabd #JeffDean #Fei_FeiLi
