"Our special sauce is the prompt libraries, the eval suites, and the thousands and thousands of real-world examples that the AI gets trained on to ensure it actually does what it's expected to do," Solberg told TechCrunch."
""With 95% of our customers, humans still solve tickets the traditional way," Solberg said. "But we see the newer companies shifting to have the primary interface between humans and the technology be an LLM.""
Risotto raised a $10 million seed round led by Bonfire Ventures with participation from 645 Ventures, Y Combinator, Ritual Capital, and Surgepoint Capital. The product autonomously resolves help-desk tickets by sitting between ticketing systems like Jira and internal resolution tooling. The system uses a third-party foundation model plus an infrastructure layer that controls the model's nondeterminism through prompt libraries, evaluation suites, and thousands of real-world examples. A payroll customer, Gusto, saw 60% of support tickets automated. Most customers still use humans for tickets, while newer companies are shifting toward LLMs as the primary human-technology interface.
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]