Brian Benedict, Arcee.ai | theCUBE + NYSE Wired: Media Week - Cyber & AI Innovators Summit
summary
Company launches agentic AI workflows at re:Invent and claims production-ready small language models that run on CPUs as well as GPUs, with GA planned in January and a live B round underway. Focus is on no-code workflow orchestration, model routing that picks specialized SLMs, post-training to reduce hallucinations and latency, portability into customer VPCs, and getting customers from POC to production in weeks. Cites 77% of enterprises using SLMs and internal benchmarks showing routing systems at 14–15% accuracy.
AI-written summary · about
chapters (17)
- 0:00Introduction and welcome
- 1:09Agentic AI and workflows
- 3:16No-code agentic workflows
- 8:54Arcee's product and mission
- 10:16Search vs. tasks in AI
- 11:41Model routing explained
- 13:24Adoption progress and challenges
- 15:31Customer journey with Arcee
- 16:35Business momentum and growth
- 17:56Inbound demand and enterprise needs
- 20:43AI cloud and infrastructure costs
- 22:13Origination story of Arcee
- 23:58Future of RAG and memory
- 25:22Infrastructure selection advice
- 26:47Preserving humanity in AI
- 28:32Human data and AGI
- 29:57Hiring and closing remarks
