Original source: Helen Yu
This article is an editorial summary and interpretation of that content. The ideas belong to the original authors; the selection and writing are by Streamed.News.
This video from Helen Yu covered a lot of ground. 6 segments stood out as worth your time. Everything below links directly to the timestamp in the original video.
Imagine a world where critical certifications and financial risks are monitored not once a year, but continuously. This example shows how AI can provide that vigilance, transforming compliance and risk management across diverse sectors.
Agentic AI System Cuts Costs and Risk by Continuous Personnel Certification
Chris Hollandback highlighted an example from Ciro, a company that provides services to governments, demonstrating how agentic workflows simultaneously deliver growth, cost savings, and risk reduction. Ciro utilizes an agent system to continuously re-check and certify government service personnel, such as ambulance drivers and firefighters, by monitoring publicly available records every few days to ensure ongoing compliance and identify issues like DUI offenses, a process normally conducted only once a year.
Hollandback extended this principle to other industries, noting that similar continuous monitoring could be applied to credit risks, such as re-evaluating creditworthiness for payables over $1,000, or to contract variability in manufacturing. By automating these checks, companies can reduce costs, mitigate risks by acting on timely information, and improve gross margins, making processes that were previously too expensive to perform frequently now feasible and impactful.
"I reduced my cost because I'm doing it in an automated way. I reduced my risk because I'm doing something I could only afford to do once a year. I'm doing it all the time whenever warranted and I'm increasing my impact, my gross margins in doing that."
Unified Platforms Essential for Production-Ready AI Agents, Boommy Executive Says
Chris Hollandback advocates for a unified platform approach over point solutions for agentic transformation, asserting that it offers crucial capabilities for production-ready AI systems. A unified platform provides sophisticated connectivity management, allowing businesses to simplify complex APIs for agents while also narrowing powerful API access to prevent misuse. This approach is vital for moving business process definitions into an agentic fabric, combining deterministic code with AI agents for seamless workflow execution.
Such platforms are indispensable for debugging, performance tuning, and ensuring complete observability of AI agents in production, guaranteeing security and proper function. Unlike custom builds, a unified platform provides essential governance and audit trails—including detailed records of AI agent 'thinking'—which are critical for compliance with regulations like the EU AI Act and for overcoming the challenges identified by a McKinsey CEO paper regarding the struggle to move AI initiatives from pilot to production.
"If you build things by hand and from scratch, the security, the observability, and the governance simply isn't there."
AI Security and Governance Deemed Non-Negotiable for Enterprise Adoption
Chris Hollandback underscores the non-negotiable importance of security and governance for AI, highlighting Boommy’s comprehensive approach, including its ISO 42001 certification—the specific standard for AI governance. Beyond certifications like FedRAMP and SOC, Boommy integrates multi-layered security measures throughout its AI system, such as delegated authorization for tools to ensure data access is limited to what users are permitted to see, preventing agents from exceeding their authority.
The company also employs third-party penetration testing specifically targeting AI vulnerabilities like prompt injection, which could trick an agent into revealing sensitive information. This extensive security framework allows customers to rapidly scale AI initiatives from pilot to production with confidence, bypassing the need to hire specialized security experts and ensuring full observability, governance, audit trails, and scalability for potentially billions of invocations.
"What that does for our customers though is it means they can move so quickly. They're like, 'Wow, so you've done all the security at multiple layers… so I don't have to hire custom security experts and all that.'"
AI Agent Integration Crucial to Prevent Hallucination and Ensure Security
Chris Hollandback states that deep integration is critical for AI agents because a lack of proper context and structured prompts can lead to "hallucination," where AI invents information. A robust platform enables the construction of structured agents with carefully defined processes, specific success instructions, and controlled access to tools and systems. This ensures that agents operate within boundaries, stopping if they cannot complete a task and adhering to user-specific authorizations through delegated authentication.
This meticulous approach yields high-quality, accurate answers that often surpass human capabilities for specific processes, while also managing inherent risks. Proper guardrails and protections against threats like prompt injection, which attempts to manipulate AI, are essential to prevent the system from generating unauthorized or biased responses, safeguarding sensitive information and maintaining ethical standards.
"If you build that really well and you say, 'Look,' and it's structured to tell it this is the process I want you to follow... if that is done, the agent behaves well."
Boommy Addresses Growing 'Agent Sprawl' as Companies Lose Control Over Disparate AI Tools
Chris Hollandback confirms that “agent sprawl,” where companies build AI agents from different vendors without central control, has rapidly escalated in the past four months, mirroring previous challenges with unmanaged data marts and APIs. This uncontrolled proliferation leads to significant data security risks and a loss of governance, as companies become unaware of what data, including internal IP or PII, is flowing into these disparate agent systems and where it is going.
He warns that without oversight, agents could gain access to sensitive systems like HR or finance, potentially exposing confidential information such as executive salaries or home addresses. Boommy addresses this by offering a Control Tower feature, which allows companies to register agents from various major vendors and even custom builds. This central registry provides unified observability and governance, enabling businesses to monitor agent behavior, access levels, and regain control over their AI deployments.
"What agent sprawl is is when everyone in your company starts just building agents from different vendors… and we lost control."
Chatbots Fall Short for True AI Transformation, Lacking Context and Connectivity
Chris Hollandback differentiates between basic chatbots and advanced AI agents, explaining that chatbots are often insufficient for true AI transformation due to their lack of context, connectivity, and memory. While some are AI-based, they typically cannot handle multi-faceted issues—like a customer underpaying an invoice due to damaged goods, where details are spread across different channels—because they don't integrate distinct business processes or retain long-term customer history and preferences.
Hollandback emphasizes that this limitation leads to "unmeasurable ROI" for simple horizontal AI systems, a finding supported by research from McKinsey and MIT. In contrast, robust AI agents excel by understanding dynamic conversations, accessing comprehensive customer history, transactional records, and support tickets, and proactively triggering business processes like sending emails or notifying staff, thereby delivering significant business impact and measurable return on investment.
"Chat bots don't solve that... Good agents have no problem with this."
Also mentioned in this video
- His role at Boommy, emphasizing that AI and platform are intrinsically linked,… (0:43)
- Agentic transformation as a fundamental rethinking of business processes beyond… (2:29)
- While APIs remain crucial, their role in an agentic world emphasizes gaining… (24:41)
- The hardest part of becoming a fully agentic company is achieving high-quality,… (29:30)
Summarised from Helen Yu · 37:45. All credit belongs to the original creators. Streamed.News summarises publicly available video content.
Streamed.News
This publication is generated automatically from YouTube.
Convert your full video library into a digital newspaper.
Get this for your newsroom →
