Alarm.com is enhancing its video security system with new AI features, including an AI Video Event Search that allows users to find specific clips using natural language, such as "red pickup truck last week." They also introduced new hardware like the battery-powered ADC-V731B spotlight camera and the ADC-VDB775 video doorbell, which features familiar face recognition and improved detection for people and packages. Concurrently, Oracle NetSuite is boosting its AI capabilities for businesses, with the NetSuite AI Connector Service Companion designed to help AI assistants better understand NetSuite data and workflows, offering over 100 finance-specific prompt templates.
Salesforce has significantly updated Slack, introducing 30 new AI features. These include reusable AI skills for Slackbot to automate tasks like budget creation and meeting scheduling. Slackbot now also functions as an MCP client, connecting with external services such as Salesforce's Agentforce platform, and can transcribe meetings or suggest actions based on desktop activity. On the security front, DefenseClaw, an open-source project, launched to secure AI agents with three layers of defense, including supply chain and runtime scanning, while the RSA Conference 2026 highlighted AI security, focusing on EDR for AI agents and a "5 Layers of AI Security" framework.
The effectiveness of artificial intelligence heavily relies on data quality, with many AI projects failing due to messy or disconnected data. Experts emphasize the need for a "semantic highway" to provide AI with contextual data across business systems like CRM and ERP. A recent Anthropic report suggested AI could theoretically perform over 80 percent of job tasks across many occupations, though this measure is based on speculative guesses rather than actual performance data, leading to questions about its broad conclusions. Congress is also urged to address critical issues for AI-powered wearable devices, focusing on data privacy, regulatory frameworks, and ensuring accessibility.
Cognizant AI Lab's Risto Miikkulainen proposes rethinking the approach to AI hallucinations, suggesting that creativity and hallucinations might be linked, and removing both could discard valuable aspects of AI. His team is exploring evolutionary AI to help models understand their own confidence levels, allowing them to signal when they are guessing. This focus on metacognition aims to improve AI reliability. Ultimately, successful AI implementation is seen as a "team sport," as noted by John Cedillo from Vast Data, underscoring the collaborative effort required for its effective utilization.
Key Takeaways
- Alarm.com introduced AI Video Event Search for text-based video clip retrieval and new cameras, including the ADC-V731B battery spotlight and ADC-VDB775 video doorbell with familiar face recognition.
- Salesforce enhanced Slack with 30 new AI features, including reusable AI skills for Slackbot, which can automate tasks, transcribe meetings, and connect to external services like Agentforce.
- Oracle NetSuite improved AI integration with the NetSuite AI Connector Service Companion, offering over 100 finance-specific prompt templates and integrating with NetSuite Analytics Warehouse.
- AI success hinges on high-quality, unified data, requiring a "semantic highway" to integrate data from systems like CRM and ERP for consistent decision-making.
- DefenseClaw, an open-source project, launched to provide security and governance for AI agents through three layers of defense, including supply chain and runtime scanning.
- Cognizant AI Lab's Risto Miikkulainen suggests rethinking AI hallucinations, proposing that models should understand their confidence levels (metacognition) rather than eliminating creativity.
- An Anthropic report claiming AI could theoretically perform over 80 percent of job tasks is questioned due to its reliance on speculative guesses and methodological limitations.
- The RSA Conference 2026 focused on AI security, highlighting EDR for AI agents, vulnerability management, and presenting a "5 Layers of AI Security" framework.
- Congress is urged to proactively address data privacy, regulatory frameworks, and accessibility for AI-powered wearable devices.
- Successful AI implementation is viewed as a "team sport," emphasizing the necessity of collaboration, as stated by Vast Data's John Cedillo.
Alarm.com uses AI to make finding video footage easier
Alarm.com has updated its video security system with new AI features. The AI Video Event Search lets users find specific video clips by typing what they are looking for, like "red pickup truck last week." This helps people find important moments faster without watching hours of video. Alarm.com also added a new battery-powered spotlight camera, the ADC-V731B, for areas without power. Additionally, the new ADC-VDB775 video doorbell uses AI to recognize familiar faces and detect packages or people. These updates aim to improve how customers search and use their video security.
Alarm.com enhances video security with AI search and new cameras
Alarm.com has introduced new AI capabilities and camera options to its video security platform. The AI Video Event Search feature allows users to find specific video clips using simple text descriptions, saving time and effort. They also launched the ADC-V731B Battery Spotlight Camera, which can be installed in locations without wired power and supports solar charging. The ADC-VDB775 Video Doorbell now includes features like Familiar Face recognition and improved detection for people, packages, and vehicles. These updates are designed to help customers find important video information faster and extend their security coverage.
AI needs a strong data foundation to succeed
Artificial intelligence (AI) can drive business success, but its effectiveness depends heavily on the quality of data it uses. Many AI projects fail because the underlying data is messy, incomplete, or disconnected across different systems. A "semantic highway" is needed, which is a reliable data layer that provides AI with context across the entire business. This involves integrating data from systems like CRM, ERP, and HRIS into a unified layer. Building this data infrastructure allows AI to make consistent decisions, leading to deeper automation and a competitive edge, while neglecting it risks falling behind competitors.
Cognizant AI expert Rethinks AI Hallucinations
Risto Miikkulainen from Cognizant AI Lab suggests that trying to eliminate AI hallucinations might be the wrong approach. He believes creativity and hallucinations are linked, and removing both could discard valuable aspects of AI. Miikkulainen's team is using evolutionary AI to help models understand their own confidence levels, allowing them to signal when they are guessing or creating information. This focus on metacognition aims to improve AI reliability by making models aware of their own certainty. The research also explores how AI can support real-world applications like water networks and brain modeling.
DefenseClaw launched to secure AI agents
DefenseClaw, an open-source project, has been released to provide security and governance for AI agents like OpenClaw. It offers three layers of defense: securing the supply chain by scanning installed components, securing the runtime by checking prompts and code for threats, and securing the system boundary with OpenShell. DefenseClaw also provides observability by streaming scan results and decisions to systems like Splunk. The goal is to automatically enforce security decisions and protect against potential risks associated with AI agents.
NetSuite enhances AI integration for businesses
Oracle NetSuite is improving its AI capabilities to help businesses use artificial intelligence more effectively. The NetSuite AI Connector Service Companion allows AI assistants to better understand NetSuite data, permissions, and workflows, enabling more reliable AI use in finance and operations. It includes a library of over 100 finance-specific prompt templates and skills to guide AI models. NetSuite MCP Apps will bring familiar NetSuite interfaces into AI assistants for easier interaction with data. Additionally, the NetSuite AI Connector Service now works with NetSuite Analytics Warehouse for broader AI-powered analysis.
Congress urged to consider AI wearable device issues
A recent meeting highlighted three key areas for Congress regarding AI-powered wearable devices. First, data privacy and security are crucial due to the sensitive information these devices collect. Second, the regulatory framework needs updating to address AI-specific challenges like bias and accountability. Third, ensuring accessibility and equity for all users is important to prevent a digital divide. Policymakers are encouraged to take a proactive approach to shape the future of these technologies responsibly.
Anthropic's AI job capability study questioned
A recent Anthropic report suggested AI could theoretically perform over 80 percent of job tasks across many occupations. However, a closer look reveals this 'theoretical capability' measure is based on speculative guesses from AI experts, not actual job performance data. The study's methodology, which involved annotators unfamiliar with specific jobs, had limitations leading to subjective labels and unclear logic. While AI can perform certain tasks like text generation and summarization, the report's broad conclusions about AI taking over jobs are based on outdated assumptions and may not reflect current AI capabilities.
RSAC 2026 focuses on AI security and innovation
The RSA Conference 2026 shifted from spectacle to substance, with a focus on AI security and innovation. Key trends included EDR for AI agents and new approaches to vulnerability management. A notable issue was the 'Collaboration Paradox,' where vendors possess valuable threat data but often withhold it due to competition. The conference highlighted a 50/50 market split between securing AI systems and using AI in security tools. The "5 Layers of AI Security" framework was presented to guide the protection of modern AI systems, emphasizing practical implementation over hype.
Vast Data exec: AI requires teamwork
John Cedillo, VP of Vast Data's Global Partner Organization, stated that artificial intelligence is a 'team sport.' This perspective was shared at the Vast Forward event. The comment suggests that successful AI implementation and utilization involve collaboration among various individuals, teams, or even companies.
Salesforce revamps Slack with 30 AI features
Salesforce has introduced significant AI-driven updates to Slack, adding 30 new features. A key addition is reusable AI skills for Slackbot, allowing users to define tasks that can automate processes like budget creation and meeting scheduling. Slackbot now also acts as an MCP client, connecting with external services like Salesforce's Agentforce platform. Additionally, Slackbot can transcribe and summarize meetings, and operate outside Slack to monitor desktop activities and suggest actions. These changes aim to transform Slack into a more versatile platform for business tasks.
Sources
- Searching security video gets easier with Alarm.com's new AI tool
- Alarm.com Expands Video Security Portfolio with New AI Capabilities and Flexible Camera Solutions
- AI Is Only As Good As The Semantic Highway It Runs On
- When AI Doesn't Know What It Doesn't Know
- DefenseClaw is Live!
- NetSuite Extends Commitment to Helping Businesses Use AI Their Way
- Three Considerations for Congress on AI-Powered Wearable Devices
- How did Anthropic measure AI's "theoretical capabilities" in the job market?
- Beyond the Spectacle - RSAC 2026 and The 5 Layers of AI Security
- Vast Data exec explains why ‘AI is a team sport’ at Vast Forward
- Salesforce announces an AI-heavy makeover for Slack, with 30 new features
Comments
Please log in to post a comment.