Skan AI Raises $63M to Build the "Context Layer" Enterprise AI Is Missing
Skan AI closed a $63M Series C led by Cathay Innovation and Dell Technologies Capital, bringing total funding to ~$120M. The thesis: AI agents fail in production because they're grounded in process docs and system logs — not in the work itself. Skan's software observes how employees actually move work through enterprise applications (CRM, email, mainframes), then distills patterns into a continuously updated "context graph of work." At one top U.S. bank, Skan claims to have uncovered $37M in operational friction and cut cost per transaction by 32%. The company says revenue grew 300%+ YoY for a second consecutive year, with 7 of the 10 largest U.S. banks and a quarter of the Fortune 50 as customers.
Read on Unite.AI →HERE Research: Enterprise AI Often Creates as Much Work as It Saves
New research from HERE Enterprise surveyed 1,000 U.S. finance, healthcare, and public sector professionals — and found enterprise AI is creating a "Toggle Tax": the cognitive cost of constantly switching between AI tools and other systems. Key findings: 30% of AI users spend at least half their workday re-entering information between AI and other systems. 40% handle higher task volumes because they're overseeing AI output. 48% say AI has made jobs harder or created as much work as it saved. 72% admit to circumventing their organization's AI restrictions. Six in 10 say productivity expectations have increased since AI was introduced, even though workloads haven't decreased.
Read on GlobeNewswire →SpaceXAI's Grok Bot: Persistent AI Agents That Sign Into Your Apps
SpaceXAI opened early beta access to Grok Bot — a system of persistent AI agents where each bot runs on a dedicated cloud computer and can sign into existing applications and websites, even those without APIs. The agents continue multi-step jobs after the user disconnects, coordinate with peer bots through shared context, and learn reusable workflows from a single demonstration. Access is tied to premium subscriptions like SuperGrok Heavy and Cursor Ultra. This moves agents from toy automations to always-on teammates handling email, CRM updates, and spreadsheets across multiple tools without constant supervision.
Read on AI Agent Store →River AI Raises $1.1B to Power Trainable Personal Agent Stacks
River AI, founded by xAI co-founder Igor Babuschkin, closed a $1.1B round led by General Catalyst and AMP PBC, with strategic backing from Nvidia, AMD Ventures, Y Combinator, and Temasek. River offers a training API that performs LoRA fine-tuning and RL runs on frontier open-weight models — completing complex RL jobs in 15-20 minutes without a dedicated ML team. The company claims 2-4x lower training costs than closed alternatives while keeping models open-weight. This is the infrastructure that lets teams continuously fine-tune agents on proprietary workflows without building ML infrastructure.
Read on AI Agent Store →Nvidia Ships Nemotron 3.5 Lightning and NeMo Switchyard for Smarter Agent Routing
Nvidia released Nemotron 3.5 Lightning, an open 30B-parameter mixture-of-experts model with only 3B parameters active at any time, built on a hybrid Mamba-Transformer design and tuned for high-volume agent tasks. On agent benchmarks, it delivers up to 4x faster token generation and ~30% faster task completion while matching accuracy on coding, research, and file-management workloads. Alongside it: NeMo Switchyard, an open-source routing library that dynamically chooses between open, proprietary, and Nvidia models at each step of an agent workflow — optimizing for quality, latency, or cost without hand-wiring every decision.
Read on AI Agent Store →💡 My Take
The HERE Research data is uncomfortable but essential reading. Half of AI users say AI has made jobs harder or created as much work as it saved. 72% are going around their company's AI restrictions. This is the gap between executive AI optimism and frontline reality. The message opportunity is clear: don't promise "AI does the work" — promise "AI that actually integrates." Skan's $63M raise and 300% YoY growth shows the market agrees: the missing piece isn't smarter models, it's understanding how work actually happens. Meanwhile, Grok Bot's persistent agents and River AI's $1.1B raise signal that personalized, trainable agents are the next battleground. The winners will be companies that solve integration, not just intelligence.