AI-Powered Virtual Assistants

Agent memory has quickly become one of the most discussed topics in AI. As more teams start building real agent systems, one limitation keeps showing up: agents don’t remember much. Most agents today are essentially stateless. They can reason through tasks, call tools, and generate impressive responses, but once the session ends, the system forgets everything. And the conversation often gets simplified into two layers: 𝟏) 𝐒𝐡𝐨𝐫𝐭-𝐭𝐞𝐫𝐦 𝐦𝐞𝐦𝐨𝐫𝐲 = 𝐭𝐡𝐞 𝐜𝐨𝐧𝐭𝐞𝐱𝐭 𝐰𝐢𝐧𝐝𝐨𝐰 This is where agents keep conversation history, reasoning steps, and recent tool outputs.  But context windows are temporary. 𝟐) 𝐓𝐡𝐞𝐧 𝐭𝐡𝐞𝐫𝐞’𝐬 𝐥𝐨𝐧𝐠-𝐭𝐞𝐫𝐦 𝐦𝐞𝐦𝐨𝐫𝐲 This is where agents remember things across sessions: - user preferences - past interactions - knowledge collected over time - intermediate results from previous tasks And once you start thinking about long-term memory, the problem quickly shifts. This is really a data infrastructure problem. In many cases, long-term agent memory ends up living in the data layer. Which is why databases are starting to play a bigger role in modern AI systems. For example, 𝐌𝐨𝐧𝐠𝐨𝐃𝐁 has been building more capabilities around this idea. By integrating vector search directly into the database, application data and embeddings can live in the same system instead of being split across multiple tools. That makes it easier to store agent memory, retrieve relevant context with vector search, and keep embeddings synchronized with the underlying data. For teams building AI systems, this kind of architecture reduces a lot of the complexity around memory. If you're exploring this space, MongoDB’s Learning Hub has some useful courses and hands-on labs worth checking out https://lnkd.in/gazxytWP As agents start running longer workflows, memory quickly becomes part of the system architecture. Designing how that memory is stored, retrieved, and updated may turn out to be one of the most important pieces of agent design. #aiagents #agenticai #machinelearning #data #database

The Voice Stack is improving rapidly. Systems that interact with users via speaking and listening will drive many new applications. Over the past year, I’ve been working closely with DeepLearning.AI, AI Fund, and several collaborators on voice-based applications, and I will share best practices I’ve learned in this and future posts. Foundation models that are trained to directly input, and often also directly generate, audio have contributed to this growth, but they are only part of the story. OpenAI’s RealTime API makes it easy for developers to write prompts to develop systems that deliver voice-in, voice-out experiences. This is great for building quick-and-dirty prototypes, and it also works well for low-stakes conversations where making an occasional mistake is okay. I encourage you to try it! However, compared to text-based generation, it is still hard to control the output of voice-in voice-out models. In contrast to directly generating audio, when we use an LLM to generate text, we have many tools for building guardrails, and we can double-check the output before showing it to users. We can also use sophisticated agentic reasoning workflows to compute high-quality outputs. Before a customer-service agent shows a user the message, “Sure, I’m happy to issue a refund,” we can make sure that (i) issuing the refund is consistent with our business policy and (ii) we will call the API to issue the refund (and not just promise a refund without issuing it). In contrast, the tools to prevent a voice-in, voice-out model from making such mistakes are much less mature. In my experience, the reasoning capability of voice models also seems inferior to text-based models, and they give less sophisticated answers. (Perhaps this is because voice responses have to be more brief, leaving less room for chain-of-thought reasoning to get to a more thoughtful answer.) When building applications where I need a more control over the output, I use agentic workflows to reason at length about the user’s input. In voice applications, this means I end up using a pipeline that includes speech-to-text (STT) to transcribe the user’s words, then processes the text using one or more LLM calls, and finally returns an audio response to the user via TTS (text-to-speech). This, where the reasoning is done in text, allows for more accurate responses. However, this process introduces latency, and users of voice applications are very sensitive to latency. When DeepLearning.AI worked with RealAvatar (an AI Fund portfolio company led by Jeff Daniel) to build an avatar of me, we found that getting TTS to generate a voice that sounded like me was not very hard, but getting it to respond to questions using words similar to those I would choose was. Even after much tuning, it remains a work in progress. You can play with it at https://lnkd.in/gcZ66yGM [At length limit. Full text, including latency reduction technique: https://lnkd.in/gjzjiVwx ]

The anatomy of a sales call has changed dramatically. Last week, I shadowed some of HubSpot’s top reps and what struck me was how differently the best sellers work today. They’re using AI at every stage: before, during, and after the call. And the results are real. The brain: before the call. AI does the heavy research — scanning 10Ks, news, emails, and past calls to surface the insights that matter most. Tools like Breeze Assistant can prep a full company overview in seconds. According to our State of Sales Report, 74% of sellers say buyers are showing up to calls more informed than ever before. Salespeople need to be just as ready. The heart: during the call. AI notetakers capture everything: next steps, budget mentions, open questions, so reps can focus on listening, not typing or scribbling notes on the side.  Also, AI assistants surface the right case study or testimonial in real time, making every answer sharper and every example more relevant. That means as a sales rep you are more engaged and relevant. The muscle: after the call. AI follows through fast. It drafts personalized follow-up emails in your own voice, outlines next steps, and flags what needs attention. More time with customers and less time writing emails. The result: sellers who prepare better, connect deeper, and close faster. The anatomy of a great sales call used to be manual effort and hustle. Now, it’s human connection powered by intelligence.

OpenAI just launched an Agent Builder. Say goodbye to thousands of startups AI is progressing so fast, most founders don't know how to build defensibility. Here’s what the smartest founders do differently: 1. They own the data APIs are temporary. What compounds is your proprietary dataset. → throxy (yc x25) scrapes its own data instead of relying on LinkedIn or Apollo. Every interaction improves the product and makes it harder to replicate. 2. They build a UX moat When tech is commoditized, experience becomes the differentiator. → Granola is a meeting recorder, but what sets it apart is the interface. It feels delightful, fast, frictionless—like it was made for you, not for enterprise IT. In a world of clones, great UX builds loyalty. 3. They unlock power for non-technical users Defensibility comes from enabling outsiders to do what only experts could before. → Lovable turns brand design into something anyone can do. In minutes, non-designers generate entire visual identities. 4. They move up the stack Commoditized infra dies. Winners package it into workflows users already need. → Perplexity started as a research tool. Now it’s becoming an AI-native browser. Owning the user’s daily search habits is how you defend long-term. 5. They go deep, not wide Horizontal tools get replaced fast. But narrow use cases can dominate markets. → ElevenLabs focused on voice. It didn’t try to do everything—just build the best voice AI in the world. Now it powers thousands of creators and companies. What am I missing? Full breakdown in my last article: https://lnkd.in/dSbpWnBz

Sam Altman has been on a podcast blitz this week. 3 appearances in 5 days, each one a post-Dev Day sermon about the future of intelligence. I went through them all (fine, I read the transcripts) partly out of curiosity, partly out of professional obligation. When the person architecting the next platform shift narrates his thought process in public, you pay attention. Takeaways: ▪️The Verticalization of Intelligence → “I was always against vertical integration, and now I think I was wrong about that.” OpenAI’s biggest pivot since its founding: the lab is now an empire - building chips, models, and end-user interfaces in one continuous loop. In the intelligence economy, whoever controls compute and energy controls cognition. ▪️ Strategy as Evolution →“Let tactics become a strategy.” OpenAI’s R&D is Darwinian. Ship chaos, observe order, scale the mutation. Memory wasn’t conceived as a moat - users made it one. Altman’s genius isn’t foresight; it’s feedback. ▪️AI Scientists →“For the first time with GPT-5, we’re seeing little examples where models are doing science, making discoveries.” Altman’s AGI test is novel scientific discovery. Within two years, he predicts AIs will generate publishable research - and soon after, it’ll feel routine. Civilization’s next compounding force: automated invention. ▪️ Customization Is the New UX →“It would be unusual to think you can make something that would talk to billions of people and everybody wants to talk to the same person.” ChatGPT’s uniformity was naïve. The future: AIs that adapt tone, personality, and worldview to each user - an identity layer that mirrors your cognitive and emotional style. ▪️Post-Interface Computing →“You talk to your device and it does exactly what you want - then gets out of your way.” Voice is the natural endpoint of human-AI interaction - ambient, context-aware, invisible. The rumored io device is his post-screen bet: a computer that listens, reasons, acts. He is betting on the disappearance of interfaces. ▪️ Distribution Moves Inside the Assistant →“There will be a new distribution mechanic developers figure out… we’ll learn together.” Future startups will live or die by whether ChatGPT mentions them. It’s not SEO anymore; it’s AIO - Assistant Optimization. ▪️ The Democratization of Creation →“In the first few days, ~30% of users were active creators...” Altman sees creativity as universal, just bottlenecked by friction. Sora removes it, turning everyone into a micro-studio. The economics will follow: per-generation pricing for heavy users, rev-share for cameos, maybe ads if it tilts social. Compute is the new canvas: 1M downloads in <5 days, faster than ChatGPT. Altman’s worldview in one loop: Build → Release → Observe → Scale → Moralize Later. He’s a capitalist empiricist, not a philosopher. He summarizes: “AGI will come; it will go whooshing by… the world will not change as much as you’d think in a big-bang sense.”

Uber used RAG and AI agents to build its in-house Text-to-SQL, saving 140,000 hours annually in query writing time. 📈 Here’s how they built the system end-to-end: The system is called QueryGPT and is built on top of multiple agents each handling a part of the pipeline. 1. First, the Intent Agent interprets user intent and figures out the domain workspace which is relevant to answer the question (e.g., Mobility, Billing, etc). 2. The Table Agent then selects suitable tables using an LLM, which users can also review and adjust. 3. Next, the Column Prune Agent filters out any unnecessary columns from large tables using RAG. This helps the schema fit within token limits. 4. Finally, QueryGPT uses Few-Shot Prompting with selected SQL samples and schemas to generate the query. QueryGPT reduced query authoring time from 10 minutes to 3, saving over 140,000 hours annually! Link to the full article from Uber: https://lnkd.in/gy8yhAZY #AI #LLMs #RAG #AIAgents

If 2025 is the year of agents, 2026 will be the year of abundant software. I want to give you the no-BS summary of OpenAI Dev Day yesterday and why I hope every business professional who follows me is paying attention. Top takeaways: ⏩ Apps in ChatGPT— Apps from Canva, Coursera, Expedia, Booking.com, Figma, Spotify, Zillow, and more can now run inside ChatGPT. Devs can also get the Apps SDK and bring their own interactive tools into ChatGPT, complete with UI, context, and instant checkout. Commerce is getting prioritized. Crowd didn’t go nuts for this but brands might. If you have not yet realized the power of context engineering and connecting your AI systems with apps, I’m dropping a link to a guide I built below. ⏩ AgentKit — The closest we’ve seen to a real agent production workflow from OpenAI: agent builder, ChatKit, evals, connectors, guardrails checklist, all inside OpenAI’s dev platform. AgentKit takes on Make, Zapier, and n8n. In my opinion, enterprises need a bullpen (yes, like investment banking) of 30 Swiss Army knife employees that act as AI operators to build these out. I’d be looking for sharp operations or analyst minds and make them take an intro to python intensive, then have two energetic-for-AI engineers learn agent builder and upskill the 30. ⏩ Codex — Prepare for self-fixing software. Kind of. Upgraded with a Slack integration, SDK, and tools for code review and refactoring. Cisco cut code review time by 50% using Codex. The demos weren’t my favorite (they used it to control the lights on stage), but the much bigger hint was integrating Codex inside of apps with the SDK. Send this sentence to your engineers “Allie told me to tell you to review the Codex SDK announcement and consider embedding it into some test apps - it would allow our software to build more software and turn it into a self-evolving app.” And if you use slack “maybe we also add it into slack as a coworker so we can ping it requests like adding dark mode or voice input?” ⏩ New models — GPT-5 Pro is now in the API (big for legal, finance, and healthcare). GPT-realtime-mini is 70% cheaper (expect more voice-first interfaces). Sora-2 API is in preview for semi-controllable video + audio (big for design and product development). ⏩ Overall signal from the room — agents took the cake. Hints at a universal OpenAI login. Software will become more ephemeral. Strong signals the models will keep getting better. More interactive, adaptive, personalized AI. More interface changes to come. More user-to-user collaboration opportunities. More everything, probably. What questions do you have about the releases? 🎥 I also had 1-on-1 chats with Greg Brockman and Sam Altman - sharing more soon. Get the free guide here: https://lnkd.in/e2h_9gsr

I spent 3+ hours in the last 2 weeks putting together this no-nonsense curriculum so you can break into AI as a software engineer in 2025. This post (plus flowchart) gives you the latest AI trends, core skills, and tool stack you’ll need. I want to see how you use this to level up. Save it, share it, and take action. ➦ 1. LLMs (Large Language Models) This is the core of almost every AI product right now. think ChatGPT, Claude, Gemini. To be valuable here, you need to: →Design great prompts (zero-shot, CoT, role-based) →Fine-tune models (LoRA, QLoRA, PEFT, this is how you adapt LLMs for your use case) →Understand embeddings for smarter search and context →Master function calling (hooking models up to tools/APIs in your stack) →Handle hallucinations (trust me, this is a must in prod) Tools: OpenAI GPT-4o, Claude, Gemini, Hugging Face Transformers, Cohere ➦ 2. RAG (Retrieval-Augmented Generation) This is the backbone of every AI assistant/chatbot that needs to answer questions with real data (not just model memory). Key skills: -Chunking & indexing docs for vector DBs -Building smart search/retrieval pipelines -Injecting context on the fly (dynamic context) -Multi-source data retrieval (APIs, files, web scraping) -Prompt engineering for grounded, truthful responses Tools: FAISS, Pinecone, LangChain, Weaviate, ChromaDB, Haystack ➦ 3. Agentic AI & AI Agents Forget single bots. The future is teams of agents coordinating to get stuff done, think automated research, scheduling, or workflows. What to learn: -Agent design (planner/executor/researcher roles) -Long-term memory (episodic, context tracking) -Multi-agent communication & messaging -Feedback loops (self-improvement, error handling) -Tool orchestration (using APIs, CRMs, plugins) Tools: CrewAI, LangGraph, AgentOps, FlowiseAI, Superagent, ReAct Framework ➦ 4. AI Engineer You need to be able to ship, not just prototype. Get good at: -Designing & orchestrating AI workflows (combine LLMs + tools + memory) -Deploying models and managing versions -Securing API access & gateway management -CI/CD for AI (test, deploy, monitor) -Cost and latency optimization in prod -Responsible AI (privacy, explainability, fairness) Tools: Docker, FastAPI, Hugging Face Hub, Vercel, LangSmith, OpenAI API, Cloudflare Workers, GitHub Copilot ➦ 5. ML Engineer Old-school but essential. AI teams always need: -Data cleaning & feature engineering -Classical ML (XGBoost, SVM, Trees) -Deep learning (TensorFlow, PyTorch) -Model evaluation & cross-validation -Hyperparameter optimization -MLOps (tracking, deployment, experiment logging) -Scaling on cloud Tools: scikit-learn, TensorFlow, PyTorch, MLflow, Vertex AI, Apache Airflow, DVC, Kubeflow

This is an incredibly fundamental shift that we’re not talking enough about 👇 The ongoing AI platform shift is quietly dismantling SEO as we know it. As Andrej Karpathy recently pointed out on X: “It’s 2025 and most content is still written for humans instead of LLMs. 99.9% of attention is about to be LLM attention, not human attention.” Think about it: when researching software today, operational buyers and procurement teams are increasingly using AI tools like ChatGPT, Claude, Gemini, or Perplexity. Typing something like “Compare Salesforce & Hubspot for a 10-person sales team - which is better?" into an AI assistant is rapidly becoming the norm. This shift has massive implications for B2B marketers and sales teams: ✅ SEO is no longer enough ✅ The traditional “10 blue links” search model is fading ✅ The AI does the discovery work now, not the user ✅ Companies must now optimize for LLMs (not just Google search) As Tomasz Tunguz and others have noted, AIO (AI Optimization) will emerge as the new frontier in content and inbound strategy. Some early signs: ➡️ G2’s traffic is down 50% in two years post-ChatGPT (Elena Verna) ➡️ We've all read about how e.g. StackOverflow’s traffic and PMF are crumbling ➡️ As a concrete startup example - StackAI now receives more inbound from ChatGPT & Perplexity than from Google (per Antoni Rosinol) As Erik Wikander put it: 1️⃣ AI will do the heavy lifting in information discovery 2️⃣ The search results page is dying 3️⃣ Companies must optimize for AI aggregation, not human eyeballs We’re at the start of a major shift in search, discovery, and attention. E.g. HubSpot, Figma and Canva are loosing organic traffic due to content that is now answered by AI - as seen in the 3-month charts below. And yes - I know Similarweb data isn’t perfect, but the trend lines are too clear to ignore. SEO is dead. Long live AIO! #artificialintelligence