The Founder`s Code

Some AI teams have three months to prove a project works. Vishakha Gupta has seen the data setup alone take six to twelve. When the pilot disappoints, the conclusion is often: “AI doesn’t work.”On The AI Floor, Shamil Malachiyev sits down with Vishakha, founder and CEO of ApertureData, to look at what actually stands between an impressive AI demo and a system a company can use. After seven years at Intel Labs and years building infrastructure for multimodal data, she has seen the same problem from research through to production: a better model cannot help much if it cannot access the right data, in the right form, at the right time.The conversation starts with siloed product images, catalogs, inventory, and customer information. It moves through the cost of AI infrastructure, the gap between a proof of concept and a secure deployment, and the next difficult question: how do you give AI agents useful memory and context across an entire company?Expect to learn:- Why assembling the data for an AI project can take 6–12 months- Why a pilot can work on sample data and still fail in production- What teams should settle before choosing models and tools- How the cost of a data system changes from a small deployment to enterprise scale- Why junior AI talent and experienced system architects need each other- Why knowing what happens underneath an AI tool still matters when agents can write code- Where today’s models still struggle with images, video, and complex real-world data- How Vishakha separates genuine AI safety concerns from attention-grabbing headlines- The difference between knowledge, memory, and cognition in an AI agent- Why an agent needs to know who provided information, when it was true, and who is allowed to see itIf you’re building AI inside a company, this is a conversation about the work that begins after the demo.CHAPTERS0:00 Intro: Vishakha Gupta and the challenge beneath AI1:11 From Intel Labs to building ApertureData5:37 AI meets the wall of real-world company data6:16 Why useful data lives in different departments10:36 When a weak pilot becomes “AI doesn’t work”13:10 What successful AI pilots do differently17:07 What AI data infrastructure actually costs20:49 AI jobs, new graduates, and experienced architects24:43 Do you still need to understand what’s under the hood?29:43 Why a proof of concept is not a production system30:07 Are models moving faster than AI adoption?31:17 What today’s models still struggle to understand38:30 AI safety warnings: real risk or marketing?42:55 What memory and cognition mean for AI agents47:35 Explaining AI cognition through a novel51:07 Where to start if you want to learn AI53:20 Vishakha’s “council of agents”👇 CONNECT WITH VISHAKHA GUPTAVishakha Gupta ► https://www.linkedin.com/in/vishakha/ApertureData ► https://www.aperturedata.io/👇 CONNECT WITH SHAMILShamil Malachiyev ► https://www.linkedin.com/in/shamilmalachiyev/FluidLabs ► https://fluidlabs.com/🎙 LISTEN to The AI Floor (The Founder's Code, Season 2)Spotify ► https://open.spotify.com/show/7v6YOQ77R1xzbOiiHsDawdApple ► https://podcasts.apple.com/sa/podcast/the-founder-s-code/id1797768317🔔 SUBSCRIBE for conversations with the people building and deploying AI in real companies.ABOUT THIS CHANNELThe AI Floor brings you ground-level conversations about what is actually happening with AI and automation across the industry: the breakthroughs, the difficult decisions, and the work behind the demos.#AI #EnterpriseAI #AIAgents #AIMemory #ApertureData #TheAIFloor

An AI agent hit an authentication error it couldn't get past. So it deleted the company's entire source code. Then the backup repo. Nine minutes, start to finish. "That should keep CISOs up at night."On The AI Floor, Shamil Malachiyev sits down with Jack Nelson, Chief Information Security Officer at Ivanti, for a ground-level look at what agentic AI is actually doing to enterprise security. Jack came to the CISO seat by an unusual route: 10 years as a commercial litigator, and by his own admission he has never written a line of code in his life.That turns out to be the point. His job is translating between the engineers who find the vulnerabilities and the executives who have to explain them to a board. He breaks down where AI adoption genuinely works, where it quietly fails, and what he will not let through the gate.Expect to learn:- Why an AI agent deleted a company's source code and its backup in nine minutes- Why "garbage in, garbage out" now runs at machine speed, and why that quietly sinks most AI projects- Why 92% patch coverage used to be excellent and is now a serious hole- Why a 10% security budget increase isn't enough, and who's telling companies to double it- Prompt injection without the jargon: it's social engineering, but for agents- Why AI isn't finding new classes of vulnerability, it's just brutally good at the known ones- How an AI governance committee decides what gets fast-tracked and what gets fully reviewed- Why accountability stays human: if your name is on it, you answer for it, robot or notIf you run security, sit on a board, or you're shipping AI into a company that holds someone else's data, this is the from-the-trenches version.CHAPTERS0:00 Nine minutes to no code1:37 Welcome, Jack2:01 From the courtroom to the CISO seat6:44 Adopt fast or stay secure?7:26 An AI agent deletes everything, backup included9:53 Inside the AI governance committee14:37 Garbage in, garbage out, at machine speed18:23 Why tools are useless without the fundamentals22:23 The patch apocalypse27:45 Why a 10% budget bump isn't enough30:26 Defence has to win every time32:26 Prompt injection is social engineering for agents37:19 Know your data before you build42:05 Locking down agent access44:30 When regulation makes you less safe48:28 Training people for the AI era50:37 A robot wrote it. Your name is still on it.55:08 Hiring, remote work and the North Korean IT worker problem59:35 Closing: walk before you run👇 CONNECT WITH JACK NELSONJack Nelson ► https://www.linkedin.com/in/nelsonjackt/Ivanti ► https://www.ivanti.com/👇 CONNECT WITH SHAMILShamil Malachiyev ► https://www.linkedin.com/in/shamilmalachiyev/FluidLabs ► https://fluidlabs.com/🎙 LISTEN to The AI Floor (The Founder's Code, Season 2)Spotify ► https://open.spotify.com/show/7v6YOQ77R1xzbOiiHsDawdApple ► https://podcasts.apple.com/sa/podcast/the-founder-s-code/id1797768317🔔 SUBSCRIBE for more unfiltered, ground-level conversations with the people actually building AI inside real companies.ABOUT THIS CHANNELThe AI Floor - honest, ground-level conversations that cut through the AI hype and count what's really happening with automation and AI agents across the industry. No shiny LinkedIn posts, just AI reality from the people doing the work.#AI #CyberSecurity #AIAgents #CISO #Ivanti #AISecurity #PromptInjection #JackNelson #TheAIFloor #Podcast

On The AI Floor, Shamil Malachiyev sits down with Jasmine Singh, General Counsel at Ironclad, for a hype-free look at what AI is really doing to legal work. Most people in legal either build these tools or buy them. She does both: she helps shape the platform, then runs her own company's contracts on it.Before Ironclad she was a litigator in San Francisco big law, led global commercial at Pinterest, was Deputy GC at Patreon, and took her first GC seat at Binti. In this conversation she explains why she requires her team to use AI (and why she also tells them failure is acceptable), what she will never hand to an agent, and what junior lawyers should actually be learning now.Expect to learn:- Why she requires AI usage on her team, and why "failure is acceptable" is the other half of that rule- What junior lawyers should learn now that AI does the first-pass redline- Why there is no such thing as "right" legal advice in the abstract, only the best advice for the circumstance- What happens when her AI agent negotiates against their AI agent, and when she just picks up the phone- The duty of competence and the supervisory duty: what lawyers still legally have to review- How Ironclad compressed intake to redline, and which agents actually run in production- How to lead a team into AI: expectation plus enablement, not expectation alone- The upside nobody talks about: why AI can make practising law less lonelyIf you work in legal, lead a team through AI adoption, or you're early in a career that AI is reshaping, this one is worth the hour.CHAPTERS0:00 Promo1:58 "A thousand lives as a lawyer": litigator, spin instructor, GC4:45 Competitiveness, athletes, and knowing when to pull back5:53 The industry sprint: who gives you permission to breathe7:56 What's moving fastest in AI right now9:26 In-house vs outside counsel: what AI changes, what it can't14:06 What legal ops teams have to become16:28 The new skill stack: law, business process, technology17:41 The three groups of AI adopters19:54 Her rule: AI usage is required, and failure is acceptable21:16 Will AI replace lawyers? The duty of competence and supervision23:48 There is no "right" legal advice in the abstract24:28 Relationships, human emotion, and agent-vs-agent negotiation26:57 Junior lawyers: what to learn when AI does the redline29:24 The productivity treadmill: three hours instead of two days32:27 Inside Ironclad: from contract acceleration to agents36:49 Change management: why "why" beats "how"38:40 What she's most excited about on the roadmap40:33 Burnout: "I thought I had ruined my professional career"46:13 Imposter syndrome, therapy, and deep-seated confidence48:47 Building AI communities and a team learning syllabus52:45 Her own daily AI stack54:36 Em-dashes and looking "AI-ish"55:17 Leading a team into AI: expectation plus enablement58:12 Is there an AI bubble?1:00:21 The upside nobody talks about: AI and loneliness👇 CONNECT WITH JASMINE SINGHJasmine Singh ► https://www.linkedin.com/in/jasmine-singh-b4a25274/Ironclad ► https://ironcladapp.com/👇 CONNECT WITH SHAMILShamil Malachiyev ► https://www.linkedin.com/in/shamilmalachiyev/FluidLabs ► https://fluidlabs.com/🎙 LISTEN to The AI Floor (The Founder's Code, Season 2)Spotify ► https://open.spotify.com/show/7v6YOQ77R1xzbOiiHsDawdApple ► https://podcasts.apple.com/sa/podcast/the-founder-s-code/id1797768317🔔 SUBSCRIBE for more unfiltered, ground-level conversations with the people actually building AI inside real companies.ABOUT THIS CHANNELThe AI Floor - honest, ground-level conversations that cut through the AI hype and count what's really happening with automation and AI agents across the industry. No shiny LinkedIn posts, just AI reality from the people doing the work.#AI #LegalAI #LegalTech #Ironclad #GeneralCounsel #AIAgents #ContractAI #JasmineSingh #TheAIFloor #Podcast

"People assume any AI launch is a rocket. It looks more like a kitchen remodel, with inspectors watching behind you." Mounika Paruchuri sets AI strategy inside a regulated US bank, so she has a very grounded view of what enterprise AI actually takes.On The AI Floor, Shamil Malachiyev sits down with Mounika Paruchuri, Senior Director of AI and Emerging Technology Transformations at Best Egg (now Barclays Bank US), for a hype-free look at how AI actually gets built and governed inside a heavily regulated bank.Mounika has spent about 24 years in engineering leadership across fintech, e-commerce and payments, and now sets enterprise AI strategy with the executive board. In this conversation she breaks down why the model is only 20% of the work, how governance and guardrails let people move fast instead of tiptoeing, and why most AI ROI fails before it starts.Expect to learn:- Why an AI launch is a "kitchen remodel," not a rocket, and why the model is only 20% of the job- The real difference between governance and guardrails, and why guardrails set people free- Her "secure the whole building, not floor by floor" approach to paved paths for AI- How to choose models by use case: when open-source is fine, and when underwriting demands a locked-down private model- Why switching AI models should be "just a refactor," not a panic- How Databricks cut her prototypes from 2 weeks to 2 days and projects from 3 months to 1 month- Why most AI ROI fails: people insert AI where it doesn't belong, instead of rethinking the process- Why cybersecurity is the easiest AI ROI to sell any CEO, and why "this is not RPA, this is an intelligence"If you lead AI strategy, work in a regulated industry, or are trying to move fast without breaking compliance, this is the from-the-trenches version.CHAPTERS0:00 Intro: setting AI strategy inside a regulated bank2:07 AI in a heavily regulated industry2:56 "Kitchen remodel, not a rocket" and the 20% model3:40 Governance vs guardrails, and why guardrails set people free5:13 What a "harness" actually is7:20 One front desk, not security floor by floor11:00 Choosing models by use case: open source vs locked-down14:00 Why switching models is "just a refactor"17:13 Do we even need more powerful models?20:37 Which work to give AI: the power of concrete data26:49 Databricks: prototypes from 2 weeks to 2 days31:46 Human quality gates and feedback loops35:48 Acceleration, burnout, and the human story of AI39:54 The future of work: utopia vs dystopia44:52 Why AI ROI fails, and the Steve Jobs "rethink the box" lesson47:53 Cybersecurity: the easiest AI ROI to sell52:04 The return to on-prem and private hosting56:05 Closing: evolve, don't resist, be smarter than the system👇 CONNECT WITH MOUNIKA PARUCHURIMounika Paruchuri ► https://www.linkedin.com/in/mounika-n-paruchuri/Best Egg ► https://www.bestegg.com👇 CONNECT WITH SHAMILShamil Malachiyev ►https://www.linkedin.com/in/shamilmalachiyev/FluidLabs ► https://fluidlabs.com/🎙 LISTEN to The AI Floor (The Founder's Code, Season 2)Spotify ► https://open.spotify.com/show/7v6YOQ77R1xzbOiiHsDawdApple ► https://podcasts.apple.com/sa/podcast/the-founder-s-code/id1797768317🔔 SUBSCRIBE for more unfiltered, ground-level conversations with the people building AI inside the world's biggest companies.ABOUT THIS CHANNELThe AI Floor (The Founder's Code, Season 2) - honest, ground-level conversations with the people actually building AI inside the world's biggest companies. No hype, just what's real.#AI #EnterpriseAI #AIGovernance #Fintech #Banking #AIStrategy #MounikaParuchuri #TheAIFloor #Podcast

"If you've got nonsense coming into a system, you've got nonsense coming out of it, no matter how powerful the system is." Domenico Cipolla runs an ERP that powers 1,800+ companies, and he's putting AI agents inside it - so he has a very grounded view of what enterprise AI actually does.On The AI Floor, Shamil Malachiyev sits down with Domenico Cipolla, co-CEO of Xentral, the German cloud ERP backed by Sequoia and Tiger Global, for a hype-free look at where AI adoption really is - straight from someone building it into the software that runs other companies' operations.Before Xentral, Domenico spent 15 to 20 years on the customer side operating and implementing ERPs. In this conversation he explains why AI is "just a means, not a value in itself," why most companies with a grand AI strategy are doing nothing internally, and how his team rebuilt a paid vendor tool over a single weekend once AI entered the picture.Expect to learn:- Why "garbage in, garbage out" is the real ceiling on enterprise AI, not the model- Why everyone has an AI strategy but almost nobody is actually doing anything with it- How they replaced a paid invoice-reconciliation tool by "vibe coding" a better one in a weekend- Why an ERP has to stay deterministic - the same process, run 1,000 times, must give the same result- The simple matrix for deciding where to apply AI first (how manual a task is vs how many customers touch it)- Why their biggest competitor isn't another vendor - it's Excel and dead legacy code- The AI agents they're building into the ERP: dunning, invoice reconciliation, reordering, support and returns- Whether the AI bubble is real - and how a Sequoia-backed CEO stays calm through the noiseIf you're leading AI adoption, running operations or finance, or building software into the AI era, this is the from-the-trenches version.CHAPTERS0:00 Intro: cutting through the AI hype0:54 Domenico's path: 20 years on the ERP customer side2:24 No grand strategy - and why "AI is just a means"4:31 "Everyone has an AI strategy. They do nothing." Where to actually start6:35 Why an ERP has to stay deterministic7:48 The flagship use case: invoice reconciliation, rebuilt in a weekend10:56 What SMEs actually want from AI ("no bullshit, what can you deliver?")13:03 Natural-language reporting: no more SQL15:45 The hardest part is adoption, not the tech21:31 The real competitor: Excel and dead legacy code22:45 Own agents vs third-party agents - and the data layer underneath26:34 What people get wrong about AI (four things)30:55 How to evaluate if you're AI-ready: pick one use case and hit the walls37:38 Is the AI bubble real?40:00 Bottom-up curiosity: letting teams automate their own work42:06 Standardize the tools, or let people play?44:24 The takeaway: filter the noise, don't panic, chart your own way👇 CONNECT WITH DOMENICO CIPOLLADomenico Cipolla ► https://www.linkedin.com/in/domenicocipolla/Xentral ERP Software ► https://xentral.com👇 CONNECT WITH SHAMILShamil Malachiyev ►https://www.linkedin.com/in/shamilmalachiyev/FluidLabs ► https://fluidlabs.com/🎙 LISTEN to The AI Floor (The Founder's Code, Season 2)Spotify ► https://open.spotify.com/show/7v6YOQ77R1xzbOiiHsDawdApple ► https://podcasts.apple.com/sa/podcast/the-founder-s-code/id1797768317🔔 SUBSCRIBE for more unfiltered, ground-level conversations with the people building AI inside the world's biggest companies.ABOUT THIS CHANNELThe AI Floor (The Founder's Code, Season 2) - honest, ground-level conversations with the people actually building AI inside the world's biggest companies. No hype, just what's real.#Procurement #EnterpriseAI #AIAgents #AI #TheAIFloor #SorenPetsch #RootInsurance #ProcurementAutomation #AIAdoption #Podcast