Intelligent Virtual Assistants

Voice agents are having their moment in 2025: an open-source breakthrough just redefined real-time multimodal AI by slashing interaction latency to 1.5 seconds, challenging the recently released proprietary real-time APIs from OpenAI and Google. VITA-1.5, the latest iteration of the open-source interactive omni-multimodal LLM, brings three major improvements that push the boundaries of multimodal AI: (1) Speed transformation - reduced end-to-end speech interaction latency from 4 seconds to 1.5 seconds, enabling true real-time conversations (2) Speech processing leap - decreased Word Error Rate from 18.4 to 7.5, rivaling specialized speech models (3) Multimodal excellence - boosted performance across MME, MMBench, and MathVista from 59.8 to 70.8 while maintaining robust vision-language capabilities One novel method from the paper is VITA’s progressive training strategy that allows speech integration without compromising other multimodal capabilities - a persistent challenge in the field. The image understanding performance only drops by 0.5 points while gaining an entirely new modality. As we move towards agentic AI systems that need to process and respond to multiple input streams in real time, VITA-1.5's achievement in reducing latency while maintaining high accuracy across modalities sets a new standard for what's possible in open-source AI. This release signals a shift in the multimodal AI landscape, demonstrating that open-source alternatives can compete with proprietary solutions in the race for real-time, multi-sensory AI interactions. VITA-1.5 https://lnkd.in/gj7pd77P More tools, open-source models, and APIs for building voice agents in my recent AI Tidbits post https://lnkd.in/g9ebbfX3

Your next colleague might be an AI with its own login credentials and autonomy. Anthropic's security chief explains why that's both exciting and terrifying. Jason Clinton, Chief Information Security Officer at Anthropic, predicts AI virtual employees with their own identities, memories, and corporate accounts could arrive within a year. This isn't about chatbots. These are intelligent systems with broader responsibilities, deeper workflow integration, and independent decision-making capabilities. The business case is already driving rapid adoption.  📌Shopify CEO Tobi Lütke told employees AI tools must be tried first before requesting new headcount. 📌Klarna halted all hiring after AI assistants absorbed work equivalent to 700 customer service staff, cutting resolution times from 11 minutes to two. But Clinton warns most organizations haven't addressed the cybersecurity risks. Virtual employees will need user accounts, system access, and autonomy. What happens when credentials get compromised? How do you prevent them from going rogue? How do you audit their decisions when something goes wrong? Instead of rushing deployment, here's what you should do: ✅ Build incident response protocols specifically for AI employee failures  ✅ Implement behavioral monitoring designed for non-human patterns  ✅ Create clear accountability chains before deployment, not after incidents The companies that solve these problems first gain competitive advantage. Those ignoring them will scramble to contain breaches and operational failures. AI virtual employees will transform work. But transformation without preparation is expensive chaos. How is your organization preparing for AI employees with system access and decision-making autonomy?

AI + HI = Improved CX In today’s digital world, businesses strive to deliver exceptional customer experiences (CX) to stand out. While artificial intelligence (AI) has revolutionized CX by enabling automation, personalization, and efficiency, it cannot fully replace the human touch. AI enhances CX by processing vast amounts of data in real time, predicting customer preferences, and providing instant responses through chatbots, recommendation engines, and self-service options. It reduces wait times, offers 24/7 support, and ensures consistency across interactions. However, AI alone has limitations—it lacks emotional intelligence, creativity, and the ability to handle complex, nuanced customer concerns. Human agents bring empathy, critical thinking, and problem-solving skills that AI cannot replicate. When combined with AI, human agents become more efficient, as AI handles routine tasks, provides insights, and allows them to focus on high-value interactions. Impact on BPO KPIs 1. First Call Resolution (FCR) Improvement: • AI-driven knowledge bases and predictive analytics equip human agents with real-time solutions, reducing repeat calls. • Virtual assistants handle routine inquiries, allowing human agents to focus on complex issues. 2. Reduction in Average Handling Time (AHT): • AI-powered tools like speech analytics and automated summaries minimize the time agents spend on after-call work (ACW). • Virtual assistants can gather customer information before handing over to a live agent, speeding up resolutions. 3. Increased Customer Satisfaction (CSAT): • AI ensures faster response times and personalized interactions based on past behavior. • Human agents, equipped with AI-driven insights, can provide more empathetic and accurate solutions, improving overall satisfaction. 4. Enhanced Agent Productivity and Utilization: • AI automates repetitive tasks such as data entry, ticket classification, and FAQs, freeing up agents for complex interactions. • Sentiment analysis tools help agents adjust their approach in real time for better engagement. 5. Lower Cost Per Contact: • AI-driven self-service options reduce the volume of inbound calls and chats, lowering operational costs. • Intelligent routing ensures the right agent handles the right query, optimizing workforce efficiency. 6. Improved Net Promoter Score (NPS): • Personalized AI-driven recommendations and proactive outreach enhance customer engagement. • The combination of AI efficiency and human empathy fosters long-term customer loyalty. The synergy of AI and HI leads to an improved CX by ensuring speed, accuracy, and emotional connection. AI-driven insights empower human agents to offer proactive solutions, while human empathy ensures customers feel valued. AI and HI are not competitors but collaborators. Businesses that successfully integrate both will deliver superior CX, optimize BPO performance, and achieve sustainable growth in an increasingly digital world.

Contact centers may not be the most exciting application for AI, but as our team has been digging into the category, I’ve been impressed by how far things have come — even since we last looked at it a few months ago. One area in particular is AI agent assistants. These copilot solutions are advancing rapidly, with capabilities such as: • Call summarization, classification, and structured data collection (i.e. filling out CRM fields) • Agent response and next-best-action support (for both chat and phone conversations) • Real-time caller sentiment analysis • Real-time QA and agent feedback • Automatic surfacing of relevant information (e.g. SOPs, help content, and customer info) Unlike many of the other areas we cover, the AI agent assistant category is primarily composed of vendors who are not specific to the healthcare industry. These products frequently show up as part of more comprehensive omnichannel Contact Center as a Service (CCaaS) platforms, such as: • Bright Pattern • Dialpad • Five9 • Genesys • NICE • Talkdesk • ujet.cx Additionally, there are a handful of industry-agnostic vendors who offer agent assistants as a standalone product or paired with broader intelligence features, like QA insights and performance analytics. These include: • Abstrakt • Balto • Convin • JustCall • Level AI Where the vendors above offer solutions that will work across all contact center use cases, there are situations where solutions for specific healthcare workflows — such as instances where clinical care and digital communication overlap — are needed. While these solutions may not work for your entire contact center, they can drive meaningful value for specific aspects of your operation. Examples include: • Birch.ai - healthcare-specific AI-powered agent assistants and call center intelligence • Laguna Health - AI-enabled conversational AI care management platform • Rotera Alyks - digital assistant for revenue cycle call center operations • Verbal - AI-enabled assistance and QA platform for virtual care clinicians We're interested to see whether organizations will be willing to implement multiple specialized solutions or will sacrifice specificity for efficiency with one-size-fits-all options. Like everything else in AI these days, this space is evolving rapidly.

My AI Journey, Chapter 1: From Ambitious Goals to Tangible Impact in IT VMO A couple of years ago, our CIO laid down a challenge that truly ignited my AI journey: "50% of all IT work is AI-powered" and "Reduce employee task friction by 50%." Bold goals, right? But as Leader of IT VMO, I saw an immediate opportunity to tackle a persistent pain point that many of us in operations face. Our IT VMO team was constantly fielding the same questions from stakeholders. While we had meticulously documented answers in SharePoint, training sessions, and various forums, the sheer volume of repetitive queries was a significant manual burden. This wasn't just friction; it was a drain on our capacity to focus on strategic VMO initiatives. That's when we decided to build our own solution. Inspired by tools like Cisco IT's BridgeIT (which leveraged GPT 3.5 at the time), we developed a specialized AI chatbot for our stakeholders - VIVA (VMO Integrated Virtual Assistant). The premise was simple: stakeholders could ask questions in natural language, and our Generative AI would respond with clear, concise, and easy-to-understand answers, pulling directly from our existing knowledge base. The impact? Revolutionary. This simple chatbot has given my team back invaluable time. We've shifted from being reactive answer-providers to proactive strategic partners, focusing our expertise only on those complex matters that truly require human guidance. The numbers speak for themselves: a remarkable 60% of stakeholder questions are now answered autonomously by our AI chatbot. The remaining 40% are handled by our always-on, always-available team, who can now dedicate their energy to higher-value tasks. This isn't just a story about a chatbot; it's a living testament to how I eliminated significant manual overhead, accelerated access to information, and freed our talent to innovate. For those who fear GenAI will take away jobs, or for those who hear industry leaders say AI will enable us to do more with limited time – this is what that reality looks like. It's about augmenting human potential, not replacing it. It's about empowering teams to achieve more impactful work. This is just the first chapter in my AI journey, and I'll be sharing more insights, challenges, and successes in upcoming posts about my usage of GenAI and Agentic AI in the VMO space. What repetitive tasks are currently burdening your teams? How are you leveraging AI to transform operations and truly empower your workforce? I'd love to hear your thoughts and experiences. Let's learn from each other how we can collectively drive this AI-powered future forward. #AI #GenerativeAI #AgenticAI #ITOperations #VMO #DigitalTransformation #Efficiency #Innovation #FutureOfWork #CiscoIT #AITransformation

For decades, businesses have built call centers, service teams, and help desks to fix issues faster. Yet speed alone never created loyalty. The real measure of service has always been how it makes people feel: heard, understood, and valued.   Now, with AI transforming how we engage with customers, that emotional foundation is being redefined. 62% of customers now say they prefer chatting with a bot over waiting for a human, as long as it provides faster, more accurate service, according to Salesforce.   This statistic shows that people still seek empathy and understanding, but they also want quick, smart responses. That’s where AI chatbots and virtual assistants come in.    So, what is the role of AI chatbots and virtual assistants in improving customer support? Here are a few key roles they play: ▪Immediate Understanding: 🔅 AI can analyze tone, sentiment, and keywords to understand the customer's state of mind instantly. This allows responses to feel timely and considerate, not robotic. ▪Faster Resolutions with Context: 🔅 Virtual assistants can resolve repetitive tasks instantly while passing complex cases to human agents with full context, so customers never need to repeat themselves. ▪Consistency Without Fatigue: 🔅 Unlike human agents, AI doesn’t get tired or lose patience. It brings calm, consistent support anytime, in any language, across any channel. ▪Empathetic Language Modeling: 🔅 The latest AI models are trained to respond with warmth and tact, saying things like “I understand how frustrating this must be” or “Let me take care of that for you,” just like a well-trained agent would. ▪ Boosting Human Support: 🔅 By handling the routine, AI allows human agents to focus on high-emotion, high-stakes moments where real connection is needed, creating a more powerful hybrid model. Are chatbots naturally empathetic? Not yet. But they can be designed to behave empathetically, and that’s a game-changer for CX. Support today focuses on meeting people where they are, not just directing them where the system wants. In regions like Saudi Arabia, where expectations for digital transformation and real-time service are rapidly growing, support becomes a strategic necessity. When technology understands people and people trust technology, customer support becomes more effective. #Customerexperience #CX #AI #Chatbots #Virtualassistants

People are building their own AI executive assistants that run 24/7 on a $6/month server. It's called ClawdBot and it's changing how I think about AI productivity. Here's the concept: Instead of opening ChatGPT or Claude every time you need something, ClawdBot connects AI directly to your existing apps (WhatsApp, Telegram, Slack, Discord, even iMessage.) You text it like you'd text an assistant. It texts back. But here's where it gets interesting: It runs on YOUR server. Your data stays yours. And it can do things ChatGPT can't: → Monitor your inbox and surface what actually matters → Send you a morning briefing before you wake up → Research people before your meetings → Track your bills and remind you before they're due → Control your smart home → Execute code, browse the web, manage files Real use cases I've seen people run: • A founder has 3 AI agents that SSH into each other's machines and debug each other • A developer queries 7,800 financial transactions via text message • A creator gets daily AI-generated "scene" images with weather, tasks, and quotes • Someone cleared 10,000 emails from their inbox using automated triage The setup: - Hetzner VPS: $6/month - ClawdBot: Free (open source, 18K+ GitHub stars) - Claude API: $30-100/month depending on usage - Telegram bot: Free Total: ~$40-110/month for a personal AI that never sleeps. The learning curve is real — you need basic terminal comfort. But once it's running, you interact through apps you already use. This is what "AI assistant" should have meant all along. Not another chat window. An actual assistant that lives in your pocket and handles things before you ask.

What if your AI assistant didn’t just reply but remembered your life, understood your habits, and acted before you asked? Most AI today is reactive. You prompt → it responds → it forgets. That’s helpful. But it’s not how a real assistant works. I’ve been exploring memU bot, and it feels like a glimpse into the next phase of personal AI: a 24/7 proactive assistant that runs on your own machine. Instead of living inside a chat window, it works in the background learning from your context, memory, and long-term usage. Here’s what makes it interesting: ✅ Proactive by design It takes action based on your behavior and context, not just prompts. ✅ Highly personal It builds long-term memory and adapts to your workflow and preferences. Over time, it becomes your assistant, not a generic AI. ✅ Very easy to use Download and run. No complicated setup. Even non-technical users can get started quickly. ✅ Local-first & secure Runs locally on your device. Your data doesn’t need to be uploaded to third-party servers. ✅ Lower token cost Designed to reduce LLM calls and token usage while staying always-on more efficient than Openclaw. A simple real-world example: Imagine applying for a visa. A proactive assistant like this could: • remember required documents • track missing files • organize folders • remind you of deadlines Without you constantly prompting it. That’s the shift happening: Reactive AI → Proactive AI Generic tools → Personal assistants Memory is what makes proactivity possible. And proactivity is what makes AI actually useful long-term. If you’re curious about where AI assistants are heading, this is worth exploring: 👉 memU bot: https://memu.bot/ ⭐ GitHub (add memory to any AI agent): https://lnkd.in/dv8iM8TU 🌐 Official website: https://memu.pro/ 💬 Discord community: https://discord.gg/memu Would you trust a 24/7 AI that learns from your work and life to assist you better? Curious to hear your thoughts 👇

𝗖𝗹𝗶𝗰𝗵é: 2025 will be the year of the AI agent. 𝗖𝗼𝗻𝗳𝘂𝘀𝗶𝗼𝗻: What is the difference between AI Assistants, Agentic Workflows, AI Agents, etc.? I first encountered "intelligent agent" in an AI intro course in the late 1990s, in the textbook by Russell and Norvig (now in its 4th ed., still the most-used AI textbook). 🔷 𝗔𝗴𝗲𝗻𝘁𝘀 "An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors." Sensors and effectors (or actuators) are often defined as read-only and write actions, respectively, and can be performed using the available set of tools for the environment. 🔷 𝗥𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗔𝗴𝗲𝗻𝘁𝘀 = Agent + Performance Measure A rational agent evaluates the possible actions given what it knows about its environment and picks those that will make it most successful. 🔷 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝘁 𝗔𝗴𝗲𝗻𝘁𝘀 = Rational Agent + Autonomy Autonomy is the ability to learn from one’s own experiences gained by reacting to the changes in the environment and evaluating the outcomes. An intelligent agent does not respond to inputs with pre-program outputs based on just built-in knowledge. It may start with built-in reflexes, but then it adapts based on the outcome of its actions. This is what makes it possible to respond to unforeseen circumstances. Humans are the most Intelligent Agents among all known species. 🔷 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁𝘀 = Non-natural Intelligent Agents Intelligent Agent that is created using AI techniques. AI Agent ⊂ Intelligent Agent ⊂ Rational Agent ⊂ Agent 🔶 𝗔𝗜 𝗔𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁𝘀 (aka Virtual Assistants) A system with a natural language interface. You can interact with an AI Assistant by speaking or typing in a natural language. Speech Recognition, Natural Language Processing & Understanding, and Text-to-speech are key Conversational AI technologies used in such systems. Within AI Assistant, the agent is the part that performs the task or actions. That agent can be a Rational or AI agent. 🔶 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 A system of multiple agents working together, each performing a specific aspect of a larger task and then, based on the outcome, passing the control to one of the next states in the workflow. Each of the agents can be a rational or AI agent. It is possible (at least in theory) for an Agentic Workflow to be an AI Agent. i.e., an AI Agent first generates an agentic workflow for the task, and then executes that workflow. 🔶 𝗥𝗼𝗯𝗼𝘁𝗶𝗰 𝗣𝗿𝗼𝗰𝗲𝘀𝘀 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 (𝗥𝗣𝗔) Workflows that automate repetitive tasks. It is a pre-programmed sequence of actions that follow fixed rules and paths but cannot learn and adapt. So, RPAs are Rational Agents. 💎 Autonomy (or agency) — the ability to plan, try, learn, and adapt — is the key differentiating characteristic between Rational and Intelligent Agents. 👉 Share your examples of rational/AI agents in the comments. #ML4Devs #AI #AIAgents #AgenticAI #AgenticWorkflows

LLMs like the new AI assistant form Perplexity are taking over whole workflows and why that's exciting 🤖 Last week, ChatGPT released its reminder and scheduling features. Now, Perplexity AI is stepping up with their new AI Assistant—a tool that blends reasoning, web search, and app integrations to help users handle both simple and complex daily tasks. This goes beyond just answering questions. The assistant connects with apps like OpenTable to book dinner reservations or Uber to call a ride. It’s a great example of LLMs evolving to not only provide information but also take action. Similar concept to ChatGPT and the plugin connection. Why This Matters for UX Designers? 🤔 From a UX perspective, tools like this are shifting the focus from isolated tasks to end-to-end workflows, like apple intelligence e.g. It’s no longer just about delivering information but enabling users to achieve specific goals seamlessly. A bit like what the rabbit R1 promised with their large action models (some of you might remember. The device was not a hit but the idea and concept was definitely innovative) Take these examples: 💡 Job: Book a dinner for you and a friend with dietary restrictions. Workflow: Research restaurants → Filter by dietary needs → Pick one → Book the table. 💡 Job: Find out what Steve Jobs said about Japanese design. Workflow: Search for quotes → Check sources for credibility → Present the best answer → watch YouTube video 💡 Job: Get a great coffee nearby. Workflow: Search for nearby cafes → Check ratings → Get directions. I am talking a lot about this but this kind of integration pushes us to think more deeply about user goals and workflows. Instead of designing for static interactions, we’re creating systems that think and act on behalf of the user. It’s a step closer to what tools like Apple Intelligence and Rabbit R1's large action model are exploring—smart assistants that truly support users in their everyday lives. For UX designers, this is an opportunity to dive into designing for automation, context awareness, and seamless multi-step experiences. The potential? An entirely new way of thinking about how people interact with technology to get things done. What do you think—would you trust an assistant like this with your daily tasks? 😊