Particle Accelerator: A Particle41 Podcast

The gap between demo and reliable ROI | Pratik Verma

49 min · I går
episode The gap between demo and reliable ROI | Pratik Verma cover

Beskrivelse

Your AI agent nailed the demo. But two weeks after going live, you're staring at support tickets you can't reproduce, cloud bills you can't explain, and behavior nobody predicted. In this episode of Particle Accelerator, our host Benjamin Johnson sits down with Pratik Verma, Founder and CEO of Okahu AI, to unpack exactly why AI agents break in production and what it takes to actually build reliable, observable agentic systems. Pratik has spent years at Microsoft and leading his own AI infrastructure companies. His open-source Monocle library and Okahu platform are purpose-built for the era of non-deterministic, LLM-powered agents, filling the gap that Datadog, New Relic, and traditional APM tools simply weren't designed for. In this conversation, Benjamin and Pratik get into: The most common root cause of agent failure (hint: it's not the model) Why the orchestration layer is where everything breaks Real-world example, how an Epson QA team cut triage from 30 hrs to 2 hrs with agents The structural difference between tracing a microservice vs. tracing an agent Evals vs. observability, what they are, why you need both, and how they connect How to instrument for token cost waste from day one MCP observability, what happens when you can't see inside the calling agent The self-healing agent loop, observability, coding agent, fix and repeat What software engineering looks like in 3 years Chapters:  00:00 - Intro & Guest Welcome 01:05 - Why agents fail: lab vs. production reality 02:28 - The orchestration layer: where flexibility becomes fragility 04:00 - The noun confusion failure mode (real-world example) 04:51 - The failure mode engineering leaders never see coming 06:29 - Why Datadog doesn't work for AI agents 11:22 - Instrumenting for cost: token budgets, session tracking & alerts 14:22 - Evals vs. observability: drawing the line 17:21 - From deterministic to non-deterministic systems 18:24 - Real-world case study: Epson QA agents 22:51 - Call center AI & observability signals in practice 30:07 - MCP observability and institutional knowledge loops 35:10 - The emerging agent economy and skill repositories 46:26 - What the next 3 years of AI looks like Pratik Verma, Founder & CEO, Okahu AI Connect with him: https://www.linkedin.com/in/pratikrverma/ Learn more at: https://www.okahu.ai/ Benjamin Johnson, Host, Particle Accelerator & CEO, Particle41 Connect with Benjamin on LinkedIn: https://www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ #AIAgents #MLOps #AIObservability #ProductionAI #EngineeringLeadership #TheParticleAccelerator #Particle41

Kommentarer

0

Vær den første til at kommentere

Tilmeld dig nu og bliv en del af Particle Accelerator: A Particle41 Podcast-fællesskabet!

Kom i gang

1 måned kun 9 kr.

Derefter 99 kr. / måned · Opsig når som helst.

  • Podcasts kun på Podimo
  • 20 lydbogstimer pr. måned
  • Gratis podcasts

Alle episoder

111 episoder

episode The gap between demo and reliable ROI | Pratik Verma cover

The gap between demo and reliable ROI | Pratik Verma

Your AI agent nailed the demo. But two weeks after going live, you're staring at support tickets you can't reproduce, cloud bills you can't explain, and behavior nobody predicted. In this episode of Particle Accelerator, our host Benjamin Johnson sits down with Pratik Verma, Founder and CEO of Okahu AI, to unpack exactly why AI agents break in production and what it takes to actually build reliable, observable agentic systems. Pratik has spent years at Microsoft and leading his own AI infrastructure companies. His open-source Monocle library and Okahu platform are purpose-built for the era of non-deterministic, LLM-powered agents, filling the gap that Datadog, New Relic, and traditional APM tools simply weren't designed for. In this conversation, Benjamin and Pratik get into: The most common root cause of agent failure (hint: it's not the model) Why the orchestration layer is where everything breaks Real-world example, how an Epson QA team cut triage from 30 hrs to 2 hrs with agents The structural difference between tracing a microservice vs. tracing an agent Evals vs. observability, what they are, why you need both, and how they connect How to instrument for token cost waste from day one MCP observability, what happens when you can't see inside the calling agent The self-healing agent loop, observability, coding agent, fix and repeat What software engineering looks like in 3 years Chapters:  00:00 - Intro & Guest Welcome 01:05 - Why agents fail: lab vs. production reality 02:28 - The orchestration layer: where flexibility becomes fragility 04:00 - The noun confusion failure mode (real-world example) 04:51 - The failure mode engineering leaders never see coming 06:29 - Why Datadog doesn't work for AI agents 11:22 - Instrumenting for cost: token budgets, session tracking & alerts 14:22 - Evals vs. observability: drawing the line 17:21 - From deterministic to non-deterministic systems 18:24 - Real-world case study: Epson QA agents 22:51 - Call center AI & observability signals in practice 30:07 - MCP observability and institutional knowledge loops 35:10 - The emerging agent economy and skill repositories 46:26 - What the next 3 years of AI looks like Pratik Verma, Founder & CEO, Okahu AI Connect with him: https://www.linkedin.com/in/pratikrverma/ Learn more at: https://www.okahu.ai/ Benjamin Johnson, Host, Particle Accelerator & CEO, Particle41 Connect with Benjamin on LinkedIn: https://www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ #AIAgents #MLOps #AIObservability #ProductionAI #EngineeringLeadership #TheParticleAccelerator #Particle41

I går49 min
episode Leadership after rupture | Dr. Kimberly Bonniksen cover

Leadership after rupture | Dr. Kimberly Bonniksen

In this episode of the Particle Accelerator, host Benjamin Johnson sits down with Dr. Kimberly Bonniksen, founder of Legends Research and creator of Kairos AI. A researcher, Dr. Kimberly brings three decades of experience studying how leaders carry responsibility, sustain trust, and create meaning that endures.   From her early work scaling the University of Phoenix online division to her groundbreaking research with Tony Robbins and Chet Holmes, Kimberly has built a unique framework for leadership rooted in ancient narrative structures and modern data science.   In this conversation, they explore:   The critical difference between persuasion and manipulation, and why leaders cross the line without realizing Why servant leadership has created an "Atlas Complex" that's unsustainable How to escape the "box" of self-deception and lead from a place of genuine service The power of third-party validation in organizational change The three life-defining moments (her stroke, her horse fire, and a wildfire) that reshaped her research How Kairos AI uses ancient leadership principles to help organizations make better decisions The future of personalized learning in education, powered by persuasion technology   Kimberly's story of resilience, including the miraculous recovery of her Arabian horse Fire after being struck by lightning in the Nevada desert, illustrates how rupture, not strategy, is where real transformation begins. Dr. Kimberly Bonniksen, Founder of LEGENDS Research Connect with her: https://www.linkedin.com/in/kimberlybonniksen/ Learn more at: https://legendsresearch.com/ Benjamin Johnson, Host, Particle Accelerator & CEO, Particle41 Connect with Benjamin on LinkedIn: https://www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ #ParticleAccelerator #Particle41 #LeadershipPodcast #PersuasionTechnology #BusinessLeadership #AI #Education #Resilience

7. juli 202652 min
episode The Question Jay Abraham Asks Every Founder cover

The Question Jay Abraham Asks Every Founder

If you've ever used the phrase "unique selling proposition" in a meeting - you've used Jay Abraham's work. Same with strategy of preeminence, risk reversal, power partnering, and relational capital. These are frameworks the business world now runs on, often without knowing where they came from. In this episode, Particle41 CEO Benjamin Johnson sits down with Jay Abraham  - founder of The Abraham Group, Forbes Top 5 Executive Consultant, 10,000+ clients across 400+ industries - for a wide-ranging conversation on what's working, what's broken, and what the next three years in business actually look like. What we cover: Why AI is following the same pattern as every "best practice" wave before it, and what the real competitive edge will be The over-automation trap: how replacing human touchpoints costs you your best, most loyal customers "Successfully stuck", the most dangerous place a business can be when things appear to be going well OPIs (Overlooked Performance Indicators): the 40–60 drivers sitting behind your KPIs that almost no one manages The difference between preeminence and influence, and why confusing them is quietly eroding market trust Power partnering and joint ventures: why companies like Microsoft and Zoom generate billions through partnerships most founders never consider Jay's vault of 24 unpublished books, 97 methodologies, and why he's spent decades frustrated by the gap between learning and application What the entrepreneurial world looks like in three years, and what he's most worried about About Jay Abraham: Jay Abraham is the founder of The Abraham Group and one of the most recognized business growth strategists in the world. Tony Robbins devotes a full day of his premier five-day business event to teaching Jay's three-ways-to-grow framework. Forbes named him one of the top five executive consultants in the United States. He has generated an estimated $9 billion in results for his clients across more than four decades of work. Chapters:  0:00 - Introduction & Jay's background  1:28 - Indianapolis roots & going back to your origins  8:40 - Starting at 28: the mail order world and the beginning  13:55 - What founders get wrong about AI  17:17 - When automation kills customer relationships  20:00 - Fast gears, slow gears & the hybrid model  22:06 - How to spot strategy vs. execution gaps in the first meeting  27:24 - What's missing when great strategy fails to execute  33:00 - Cultural connection to mission & global teams  35:32 - Power partnering, joint ventures & distribution channels  37:00 - Being "successfully stuck"  40:00 - OPIs: Overlooked Performance Indicators  43:17 - Agentic AI and call transcript analysis  45:30 - Jay's vault: 24 books, 97 methodologies, and legacy  52:13 - The AI clone experiment & why people don't apply knowledge  55:10 - What the inspired Jay wants to build  1:00:45 - Where Jay started - and what still drives him  1:04:08 - Preeminence vs. influence  1:06:09 - What the world looks like in 3 years  1:09:09 - Closing: the obligation to add value Jay Abraham, Founder, The Abraham Group | Forbes Top 5 Executive Consultant  Connect with Jay Abraham on LinkedIn  www.linkedin.com/company/the-abraham-group/  Learn more here: www.abraham.com/ Benjamin Johnson, Host, Particle Accelerator & CEO, Particle41 Connect with Benjamin on LinkedIn www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ #FounderMindset #BusinessPhilosophy #Preeminence #JayAbraham #Entrepreneurship #Particle41 #ParticleAccelerator

7. juli 20261 h 10 min
episode How One Founder Built the "Google Street View" for HOAs | Wesley Hunt cover

How One Founder Built the "Google Street View" for HOAs | Wesley Hunt

Wesley Hunt co-founded Optic Systems after nearly a decade inside the property management industry - and the company he built looks nothing like what you'd expect. It started with a simple observation that property managers are expected to respond to homeowners instantly while also physically driving to every community they manage. You can't do both. Something always breaks down. The solution Wes and his co-founder landed on? "Google Street View, but on demand and for the private sector." 360-degree video inspections, GPS-synced, quality-checked, and delivered to a platform that property managers, landscapers, paving companies, and city code enforcement offices can all use. In this episode, we trace the full arc - from Wesley's first job at 17 as a land survey technician in Pennsylvania, through law school, HOA management, a patent filing, a failed SaaS model, a successful pivot to a driver service model, a full platform replatforming, and now a new AI-ready hardware system launching at CAI. We also get into the hard stuff, what it actually takes to build real-time integrations with legacy property management platforms, what broke when their MVP hit unexpected growth, and what the aging HOA manager workforce means for the whole industry. If you work in property management, proptech, public works, landscaping, land development, or you're building a B2B SaaS platform with a physical operations component - this one is for you. What we cover: 00:08 - Intro & lightning round: bills fan, drone pilot, Adam Grant 04:18 - Age 17, GPS computation, and land surveying in Pennsylvania 06:10 - Why law school, and what transactional law taught him 08:22 - Falling into property management and running 19 communities 11:18 - The worst weeks on the job and what actually breaks in HOA world 14:16 - The 2018 conversation that became Optic Systems 17:38 - The patent process and what it did for investor conversations 22:24 - Vertical strategy: HOAs, landscaping, construction, public works 26:33 - Opening doors with municipalities 30:35 - What happens behind the scenes when a manager opens their dashboard 34:44 - What Wes is most proud of 36:35 - The integration problem nobody warned them about 43:33 - The platform replatforming decision - and how to do it without losing customers 46:29 - What's the difference between a vendor and a partner 52:18 - What they're launching at CAI: AI-ready triple-camera system 57:30 - Autonomous vehicles, driverless fleets, and the long-term roadmap 58:58 - What Wesley would tell any property management exec right now Wesley Hunt, President & Co-Founder of Optic Systems Connect with Wesley Hunt on LinkedIn - https://www.linkedin.com/in/wesley-f-hunt-b955b340/  Benjamin Johnson, Host, Particle Accelerator Podcast & CEO, Particle41  Connect with Benjamin on LinkedIn https://www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ #PropertyManagement #PropTech #HOA #StartupFounder #B2BSaaS #Particle41 #ParticleAccelerator

6. juli 20261 h 4 min
episode Building in the Dark | Robbie Abed cover

Building in the Dark | Robbie Abed

Robbie Abed is not a developer. He's a writer who once penned a resignation letter so good that the CEO of Deloitte Consulting forwarded it to the entire firm. He's an Inc. columnist with millions of views. He took 250 coffee meetings in 400 days. He wrote a book - Fire Me I Beg You - and James Altucher wrote the foreword. Then he disappeared into his basement for six months and built a SaaS product. This is that conversation. In this episode of the Particle Accelerator podcast, Benjamin Johnson (CEO of Particle41) sits down with Robbie in a peer-to-peer conversation - no interview format, no script - about what it really takes to go from services to software, from idea to production, and from builder to founder. What we get into: - Why Robbie spent 6 months alone building before talking to a single customer - The George Test - what happens when a non-technical user tries to use your software for the first time - Vibe coding: the gift and the curse - and why "done" from an AI means there's a 5% chance it's actually done  - What enterprise due diligence really reveals about your product  - Where the real money in enterprise AI will move over the next 18 months - The slow gear vs. fast gear framework - what AI has sped up and what it absolutely has not - Why most marketing agencies don't have confidence in their own playbook - Nuclear simplicity as a product strategy - Why QA and requirements must be the same document - Logging everything: why PostHog, Sentry, and GCP logs are now the system of truth Whether you're a founder in the basement, a CIO wondering what to do with your Claude credits, or a growth leader trying to turn LinkedIn into a revenue channel - this episode will give you a new framework. Find Robbie at SoManyLemons.com | robbie@soManylemons.com  Connect with him on LinkedIn https://www.linkedin.com/in/robbiejabed/ Connect with Benjamin on LinkedIn https://www.linkedin.com/in/benjaminrjohnson/  Learn more about Particle41: https://particle41.com/ Chapters:  0:00 - Intro & Robbie's background  2:34 - The worst coffee meeting (and what 250 coffee meetings taught him)  5:22 - The Deloitte resignation letter & finding out he was a good writer  10:25 - Six months in the basement: the lowest moment  14:00 - The agony and absolution cycle in software  19:30 - Vibe coding vs. pre-vibe: what actually changed  22:31 - The freelance client who sparked So Many Lemons  25:47 - When did you know it was a real product?  29:46 - Productizing a service vs. building a real product  32:40 - Why agencies get it wrong (and how confidence in a playbook changes everything)  38:00 - Pitching to enterprise: what the conversation looks like  40:03 - What enterprise due diligence actually tests for  43:00 - Slow gear vs. fast gear in the AI era  49:01 - The George Test  51:15 - Going nuclear on simplicity  58:48 - QA is the same as requirements (the shift-left moment)  1:05:00 - Observability, logging, and self-healing software  1:12:00 - Where the money moves in enterprise AI (next 18 months)  1:17:30 - The Claude credits economy  1:23:00 - The problem with Apollo and feature-bloated SaaS  1:24:22 - How Particle41 uses So Many Lemons  1:28:00 - Wrap-up & how to find Robbie #ProductBuilding #VibeCoding #StartupLife #SoftwareDevelopment #EntrepreneurMindset #AITools #LinkedInMarketing #SoManyLemons #Particle41 #ParticleAccelerator

5. juli 20261 h 28 min