Activity
33K followers
-
Davit Baghdasaryan shared thisThe realization that we can aim to build AI tech in Armenia came to me after I had a casual lunch with Hrant Khachatrian back in 2015. That's what triggered me thinking about AI in Armenia. And that's eventually what led to starting working on Noise Cancellation in 2016 with Arto Minasyan and then officially starting Krisp in 2017. Now imagine that, multipled by 100x is what this new GPU capacity will enable in Armenia.Davit Baghdasaryan shared thisWeekend (Armenia) thought. Had the pleasure of visiting the Yerevan State University supercomputer yesterday with Hrant Khachatrian of YerevaNN. 80%+ GPU utilization. Real research happening. And this is just the beginning: $500M+ in compute infrastructure from Firebird AI and Eleveight AI is on the way - with a $4B Firebird expansion behind it. But compute alone is not the endgame. The real prize is what you build around it. Imagine: a world top-20 PhD program in physical AI and robotics. Hundreds of PhDs. Thousands of students from Eastern Europe, India, Georgia, and the wider region choosing Yerevan as their education hub. Research papers that put Armenia on the global map. The next wave of multinationals coming to Armenia will not ask how many GPUs you have - they will ask how many PhDs you have. Get that answer right, and everything else follows: the companies, deep tech startup spinoffs, the investment, the GDP lift, the prestige. It is time to THINK BIG. The hardware is secured - now be bold enough to match it.
-
Davit Baghdasaryan posted this91% of CX leaders are about to make your onshore team irrelevant. You heard it right. Onshore agents are a language tax, not a quality choice. Over 50% of voice budgets in enterprise CX go to onshore live agents. The single most expensive line item in the entire stack. And the reason is embarrassing. The language and accent barrier with offshore talent remains unsolved for most companies. So they throw money at the problem instead of fixing it. The math never made sense. They just didn't have a better option. That's changing fast. A survey of 819 CX leaders shows 91% are either using AI translation tools or actively evaluating them. What happens when any agent can serve any customer in any language in real time: → Geography stops mattering → Talent pools open up globally → Hiring becomes about skill, not accent → Costs drop by half The companies moving first will build a CX infrastructure that's extremely hard to replicate once it's running. Everyone else will be playing catch-up with a cost structure that stopped making sense years ago. Is your CX team still paying the language premium?
-
Davit Baghdasaryan shared thisToday, we shipped something very different. Not boring.Davit Baghdasaryan shared thisEvery company has a Todd. Today we finally did something about it. Krisp reduces noise. Improves clarity. It even changes accents. But sometimes the biggest meeting problem isn't background noise. It's Todd. Todd from Finance. Todd, who turns a 30-second update into a 12-minute spoken-word essay about spreadsheets. Todd, who says "just to piggyback off that" and then doesn't piggyback — he builds an entire second pig. So we built AI Deboringifier A Voice AI feature that detects boring speech patterns and automatically makes them less boring. What beta testers are saying: 💬 "I used to mass-mute everyone. Now I only mute Todd." — Sarah, Product Manager 💬 "Deboringifier saved my marriage. I stopped bringing meeting energy home." — Anonymous 💬 "I've been told I'm the reason this was built. I disagree. Also, let me give you some context on why I disagree. So basically back in Q2..." — Todd Deboringifier — Because life's too short for boring meetings. . . . Coming never. Happy April 1st. But seriously, Todd, wrap it up.
-
Davit Baghdasaryan reposted thisDavit Baghdasaryan reposted thisIt's official. Krisp is nominated for 3 The Webby Awards 🏆 Out of 13,000+ submissions, only 10% are nominated. The People's Voice Award is decided entirely by public vote. That's where you come in. You can vote once in each category below: ✅ Technical Achievement → https://wbby.co/58870N ✅ Developer Tools & APIs → https://wbby.co/58872N ✅ AI Voice & Conversational Interface → https://wbby.co/58853N Thank you for your support. Every vote counts! Davit Baghdasaryan Arto Minasyan Asti Pili Harry F. Anthony Canoso Daniel Haver Vimal Nair Karapet Gyumjibashyan 🚀Karen Movsisyan Shara M.Robert Maute Robert Schoenfield
-
Davit Baghdasaryan shared thisThrilled to roll out 2 powerful new capabilities in the Krisp enterprise admin portal, built for visibility and control at scale: 🔍 Live Monitoring for Admins Get a real-time view of every device, feature, and issue across your deployment. Know exactly how Krisp is performing across your entire fleet—at any moment. 🧾 Audit Logs for Admins Track every key admin action automatically—who did what, when, and what was affected. Full transparency and traceability, without manual effort. More control. More visibility. Built for enterprise CX.
-
Davit Baghdasaryan shared this3 open voice models in one week 🔥 Momentum is building, the stack is opening up, and agentic AI is quickly moving to production. Here’s what stood out 👇 - Mistral AI launches Voxtral TTS: Open-weight 4B TTS model. 9 languages, 90ms TTFA, 6x RTF. Mistral claims it beats ElevenLabs on quality benchmarks via Ivan Mehta for TechCrunch - Cohere releases Transcribe: Open-source 2B ASR model built for edge. 14 languages, 5.42 avg WER on HF Open ASR leaderboard, beating Zoom Scribe v1, IBM Granite 4.0, ElevenLabs Scribe v2, and Qwen3-ASR. - Google ships Gemini 3.1 Flash Live + Search Live goes global: Real-time voice/video model with native function calling. 90.8% on ComplexFuncBench Audio (~20% jump over prev gen). Now powers Search Live in 200+ countries with voice and camera input via Aisha Malik for TechCrunch - smallest.ai launches Lightning V3: 3.89 MOS in conversational evals, claims to beat OpenAI, Cartesia, and ElevenLabs. 15 languages with auto-detection and mid-sentence switching. Voice cloning from 5-15s of audio. - Amazon Polly adds Bidirectional Streaming: Stream text to Polly token-by-token as your LLM generates it, get audio back in real time over HTTP/2. 39% faster than batch approach, collapses 27 API calls to 1 on a 970-word passage. - Amazon Web Services (AWS) adds WebRTC to Bedrock AgentCore: Pipecat voice agents now run on AgentCore Runtime with bidirectional WebSocket and WebRTC. Ready-to-deploy examples with Pipecat, Nova Sonic, LiveKit, and Strands SDK. - Genesys reports record Q4: Genesys Cloud at ~$2.6B ARR, 35%+ YoY growth. 70%+ of customers now on AI. AI-powered conversations up 120% YoY. - Artificial Analysis updates voice benchmarks: ElevenLabs Scribe v2 leads at 2.3% WER. Best value: Mistral Voxtral Small at 3.0% WER / $4 per 1K min. TTS Arena: Inworld TTS-1.5-Max at #1, ELO 1,160. - AI chatbots handle 60%+ of banking support: Bank of America's Erica: 1.5B+ interactions, 98% resolved without human. Klarna: 66% of inquiries, saving $40M/yr. Gartner projects $80B in contact center labor cost cuts in 2026. - The economics of AI vs human agents: Voice AI now costs ~$0.40/call vs $7-12 for a human agent: 90-95% cost reduction per interaction. Analysis of how this is reshaping contact center staffing via Medium - Agentic Voice AI goes mainstream: 1 in 10 customer service interactions projected to be fully automated by agentic voice AI in 2026. 80% of businesses plan to deploy. RingCentral shipped AIR Pro, an agentic voice platform embedded in its comms stack. - Salesforce Agentforce Contact Center: Native CCaaS unifying voice, digital channels, CRM, and AI agents in one stack. - Otter.ai hits 35M users, $100M ARR - Sam Liang interview: $100M ARR with <200 employees ($500K+ rev/employee). Liang says 2026 is “the year of the voice.” The stack is getting faster, cheaper, and more open at the same time. The shift to voice-first systems is already underway.
-
Davit Baghdasaryan reposted thisDavit Baghdasaryan reposted this💥 Speed wins deals. The first rep to respond usually gets the meeting. 8loop builds AI sales and support agents designed to follow up instantly, qualify leads, and keep conversations moving. But real calls are messy. Background voices and noise can derail even the best AI. To keep agents focused, 8loop uses Krisp’s Voice AI for Developers to remove background noise and secondary voices before the audio reaches the AI. The result is cleaner input, faster responses, and conversations that actually convert.
-
Davit Baghdasaryan shared thisScaling STT systems isn't just a model problem. It's a scale, cost, and latency problem. In this episode with Maxime Gaudin, CTO at Gladia, we get into what breaks in production. Not just models, but infrastructure, GPUs, and economics. Here's what stood out 👇 - Winning isn’t just about model quality, it is surviving brutal tradeoffs between latency, cost, and scale. - The real challenge is not training one great model, it is running it cheap enough to meet market pricing without breaking performance. - STT is getting commoditized so fast that providers have to chase better accuracy while selling at margins that keep shrinking. - Big models don’t matter if they are too expensive to run at scale. - Real-time voice AI lives or dies under a hard latency budget, and staying under 300 milliseconds leaves little room for mistakes. - The industry obsession with one model that does everything may be the wrong path if smaller specialist models can outperform it in the moments that matter. - Every model upgrade is risky because improving one language or task can make another one worse. - Testing speech systems is harder than people admit because teams know something broke, but don’t know what. - General transcription errors can be patched by an LLM, but once a name, phone number, email, or address is lost, it is gone. - The next edge in voice AI may come from tiny models trained for high-value details like PII, not from one giant model trying to handle everything. - Email addresses sound simple until real accents, pauses, corrections, and spelling cues expose how messy spoken language really is. - The companies that win enterprise voice AI will be the ones that orchestrate many narrow models well, not the ones chasing a single universal model. - Infrastructure strategy is becoming a product decision because legal rules, traffic spikes, and customer use cases all change what “best” deployment looks like. - Cloud scaling breaks in real-time spikes, like emergency calls. - Using managed infra and large DevOps teams at once wastes money. - Customers want one vendor for everything, even if quality drops. - The market will reward depth over breadth if a vendor can become truly exceptional in one painful, business-critical part of the voice stack. If STT is becoming commoditized, does the real advantage shift to specialized models that win on PII?
-
Davit Baghdasaryan shared thisProud to be partnering with EXL. At Krisp, we’ve always believed that voice is the most human interface and yet, for too long, it’s been limited by noise, accents, and geography. Together with EXL, we’re helping teams deliver more confident, more human interactions, where every conversation is clear, and every customer feels understood.Davit Baghdasaryan shared thisAt EXL, we believe that every customer interaction is an opportunity to build trust. We’re teaming with Krisp, integrating industry-leading Accent Conversion and AI-driven Noise Cancellation, to set a new gold standard in customer engagement. In an increasingly globalized world, seamless communication is the cornerstone of superior Customer Experience (CX). By leveraging Krisp’s advanced Neural Networks and Deep Learning models, we are removing the friction of background noise and linguistic barriers in real-time. This collaboration goes beyond just technology. It’s Human-Centric AI in action. We are empowering our global workforce with that amplify clarity, professionalism and impact so every customer feels heard, understood and valued. Together, EXL and Krisp are shaping the future of enterprise communication — where every voice is crystal clear. Learn more via the link in the comments.
-
Davit Baghdasaryan reacted on thisDavit Baghdasaryan reacted on thisWeekend (Armenia) thought. Had the pleasure of visiting the Yerevan State University supercomputer yesterday with Hrant Khachatrian of YerevaNN. 80%+ GPU utilization. Real research happening. And this is just the beginning: $500M+ in compute infrastructure from Firebird AI and Eleveight AI is on the way - with a $4B Firebird expansion behind it. But compute alone is not the endgame. The real prize is what you build around it. Imagine: a world top-20 PhD program in physical AI and robotics. Hundreds of PhDs. Thousands of students from Eastern Europe, India, Georgia, and the wider region choosing Yerevan as their education hub. Research papers that put Armenia on the global map. The next wave of multinationals coming to Armenia will not ask how many GPUs you have - they will ask how many PhDs you have. Get that answer right, and everything else follows: the companies, deep tech startup spinoffs, the investment, the GDP lift, the prestige. It is time to THINK BIG. The hardware is secured - now be bold enough to match it.
-
Davit Baghdasaryan reacted on thisDavit Baghdasaryan reacted on thisI just realized why some founders build billion-dollar networks while others stay stuck. It's not their product, connections, or timing. It's whether they're lifting others while they climb. Davit Baghdasaryan is one of those people. He founded Krisp, a noise-canceling AI meeting assistant business that makes voice communication actually work in chaotic environments. If you have ever been on a call with a barking dog, construction noise, or a coffee shop in the background and couldn't hear it, that was probably Krisp doing its thing. But here is what stands out beyond the company he built. Davit hosts the Future of Voice AI podcast. He is using his platform to spotlight founders, technologists, and builders who are shaping what voice AI becomes. Not just interviewing them for content, but giving them space to share predictions, challenges, and the messy truths about building in this space. Here is what I have been noticing: → The founders who give platforms to others build stronger networks than the ones who only take stages → When you spotlight people doing hard work in your space, you elevate the entire industry instead of just your corner of it → The best collaborations come from people who default to generosity before extraction Davit is doing that by not only building Krisp but also hosting conversations and creating space for others to be heard. That matters more than people realize when you are navigating a space as fast-moving and noisy as voice AI. If you are building something, ask yourself: are you only climbing or are you also lifting? Thank you, Davit, for doing both.
-
Davit Baghdasaryan liked thisDavit Baghdasaryan liked thisAnother year, another clean audit. ✅ Krisp has completed its annual HIPAA audit with zero inconsistencies and no recommendations for improvement. Behind every conversation we power is a commitment to protecting sensitive data and customer trust. 🔒
-
Davit Baghdasaryan liked thisDavit Baghdasaryan liked thisDo you know which LLMs are best in Armenian? We are releasing ArmBench-LLM 1.0, which compares popular LLMs on our Armenian benchmark. It checks both knowledge (e.g. grammar, history, literature etc.) as well as generation capabilities (e.g. translating text, summarizing emails etc.). It also gives insight on cost vs accuracy in the spend report. Quick insights: • Gemini 3 Flash is the overall leader • Qwen 3.5 27B is the only OSS model in top 10 • Grok scores 18.75 on math exam in Armenian Links in the comments. #opensource #ArmenianAI #Metric #ArmBench
-
Davit Baghdasaryan liked thisExcited to see this happening around GTC. Our portfolio company Activeloop, led by Davit Buniatyan, continues to bring together some of the most interesting builders working at the intersection of AI infrastructure, robotics, and physical AI. At AAL VC we invested in Activeloop because we believe that data infrastructure for AI will be one of the defining layers of the stack in the coming decade. The explosion of multimodal models, robotics, and simulation systems only reinforces that thesis. What I like about gatherings like this one is simple: the most interesting ideas in AI rarely emerge from slides — they emerge from rooms full of builders comparing notes. Huge credit to Davit and the Activeloop team for convening this community. Proud that AAL VC is one of the investors supporting this journey.Davit Baghdasaryan liked thisJensen just announced the start of the GPU-accelerated database era at #GTC26. AI runs on GPUs. But your data still runs on CPUs. That mismatch is breaking the AI stack. For the last two months, we’ve been busy solving this problem. Excited to announce Deeplake becoming the GPU Database. Deeplake brings your database directly onto the GPU, eliminating the CPU <-> GPU bottleneck for AI workloads. The pendulum has switched. GPU-native queries are now 10× faster and an order of magnitude cheaper to run. Last week we even put up a 101 banner in San Francisco. And this is just the beginning. We’re planning a huge set of announcements starting this week. Stay tuned.
Patents
-
Secure User Authentication Using Biometric Information
Issued US US8799666 B2
Recommendations received
5 people have recommended Davit
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content