What happens when cloud deployment outpaces the humans managing it? At PayPal, Suresh Mathew's team built the answer — an autonomous system that ran 2 million production remediations before he spun it out as Sedai.

In this episode, I'm joined by Suresh Mathew, founder and CEO of Sedai. We talk about the problem DevOps created: developers deploying directly to production faster than any SRE team could keep up. The instinct was to hire more engineers. Suresh's team built smarter — an autonomous system that handled 80% of production operations without human intervention. No 2 AM pages. Just 2 million remediations quietly running in the background. PayPal eventually mandated it: if your app wasn't compatible with the tool, you needed an approval to deploy. Then Suresh took the idea and made it work for the rest of the world.

We get into the hard stuff: raising a seed in May 2020, growing to ~100 employees, closing a $20M Series B, and the Lambda market mistake that forced an early ICP pivot. Suresh shares his formula for hiring — strength over perfect fit — and the one signal he uses to know you have product-market fit: when your team stops dreading customer meetings and starts racing to get into them.

He also explains why autonomous systems must run at zero tolerance for production failure. Not five nines. Zero.

Guest & Resources

Suresh Mathew, Founder & CEO, Sedai

LinkedIn: https://www.linkedin.com/in/sureshmathew/

Sedai: https://www.sedai.io

We Built It Because We Had To is the founder-backstory podcast from The Artesian Network. New episodes every Tuesday and Thursday. Subscribe so you never miss a conversation.

Visit us at: https://www.WeBuiltItBecauseWeHadTo.com

AI data centers are hitting the U.S. power grid like five-gigawatt toasters — and unlike Bitcoin, that demand can never be switched off.In this episode, I sit down with Michael Parrella, CEO of ennrgy.com, about an existential moment most founders recognize but few talk about honestly: learning your largest customer is walking out the door. Michael returned as CEO of ennrgy in late 2024, and within weeks he learned they were about to lose their biggest account to a cheaper, more modern competitor. He didn't discount his way out. His team stopped defending the old product and asked one question: what do energy operators actually need? The answer — give them the decision first, the supporting data second — became the foundation of a rebuilt company. By December they had reacquired his old firm SoftSmiths and folded it in. What started as a near-death experience became the product strategy.Michael also breaks down what he calls the five-gigawatt toaster problem: AI data centers are plugging enormous, uncurtailable demand into a grid that was never built for it. He explains the fragmented reality of U.S. energy markets, why some jurisdictions are already requiring data centers to build their own power supply, why AI demand is fundamentally different from Bitcoin demand, and what it will take to prevent AI growth from crushing the energy infrastructure it runs on.Timestamps:00:01 — Welcome and intro01:01 — Career arc: from energy CEO to Chief Strategy Officer and back02:59 — The moment they learned they were losing their largest customer03:55 — How the team rebuilt the product around decision-first thinking05:11 — What ennrgy AI does: forecasting, pricing, settlements06:36 — Inverting the pyramid: decision at the top, data underneath10:05 — AI data centers and the electricity demand surge12:07 — The five-gigawatt toaster problem explained13:06 — Why AI demand cannot be curtailed the way Bitcoin demand can14:06 — Can we physically build enough power supply fast enough?15:28 — Being an ostrich doesn't work: what must be done16:19 — Parting words: existential moments create your best momentsConnect with Michael Parrella:LinkedIn: https://www.linkedin.com/in/mikeparrella/Company: https://www.ennrgy.aiSubscribe to We Built It Because We Had To on Spotify, Apple Podcasts, and YouTube. Follow Artesian Network for more conversations with the founders and operators who built their companies the hard way.

https://www.webuiltitbecausewehadto.com

Andres Andreu sits on one of the richest datasets in cybersecurity — nearly 64 billion breached identities — and for years growth still stalled. The problem was never the data. It was who Constella was selling it to.

In this episode of We Built It Because We Had To, Andres Andreu, CEO of Constella Intelligence, walks through the breach-data lake his company has amassed over 16-plus years and how it now feeds OEM cybersecurity vendors like Norton LifeLock and intelligence agencies across Europe, including Europol. You'll hear how a single ICP mistake — pitching CISOs who had no team to operationalize the data — held the company back, and how the shift to an OEM model drove 17% revenue growth in a single year. Andres also traces his arc from building federal law enforcement wiretap systems in the 1990s, to taking Bayshore Networks from employee number three to exit in 2021, to stepping into the CEO seat at Constella just six months after joining as COO. Along the way he shares the "fail fast" moment that forced his team to scrap part of their product and build a machine learning engine that earned an internationally granted patent — and saved the company.

If you're a founder or operator wrestling with product-market fit, ICP, or when to kill what you've built, this one is for you.

Chapters

00:00 What Constella Intelligence actually does

03:09 Why passwords barely change over 15 years

04:34 Inside the 24/7 breach-hunting program

05:18 Two ICPs: OEM vendors and intelligence agencies

05:47 From 90s federal wiretap systems to founding Bayshore Networks

09:20 The ICP mistake: selling to CISOs who couldn't use the data

10:25 OEM partners: Norton LifeLock, Europol, and beyond

13:07 "My vendor has 17 billion identities." "I have 64."

15:38 The fail-fast moment that led to a patent — and saved the company

19:13 What's driving growth now: customer-driven APIs and ICP focus

About the guest

Andres Andreu is the CEO of Constella Intelligence, a cyber-intelligence company operating one of the world's largest breach-data lakes. He brings 33 years in cybersecurity, from federal law enforcement to a 2021 startup exit with Bayshore Networks, and is the author of an internationally granted machine learning patent.

Connect with Andres: https://www.linkedin.com/in/andresandreu/

Learn more about Constella Intelligence: https://constella.ai/

Subscribe to We Built It Because We Had To for new founder stories every Tuesday and Thursday. Hosted by Jonathan Buckley of The Artesian Network — fractional CMO guidance for early-stage tech founders. Learn more at https://www.artesiannetwork.com

Most companies build their front door out of glass — and the guard dog is asleep all day.That's how Karen Nguyen describes the security gap her company exists to close. In this episode, Karen — co-founder and CEO of OFFENSAI — joins Jonathan Buckley on the morning she launched v2 of her autonomous cybersecurity platform. We get into what "exploitation validation" actually means and why it's not just another vulnerability scanner, how she's funding the company with paying customers and pre-seed angels instead of chasing a bloated Series A, and the over-hiring failure that taught her the most honest lesson we've had on the show: undershoot or overshoot, and nobody wins — land exactly where you said you would. It's a candid conversation about building lean, selling to skeptical CISOs, and the discipline of capacity planning in a market drunk on nine-figure rounds.What you'll learn:- Why "blinking red dots everywhere" is a noise problem, and how attack-chain validation cuts through it- How to run product-led growth in a category where CISOs say no to PLG by default- The case for paying customers over a giant Series A, and when institutional money actually helps- How an over-hire cycle right before the bubble burst reshaped how Karen plans capacity- What 15 years in startup go-to-market taught her about handling fundraising rejectionTimestamps:(00:00) Meet Karen Nguyen and OFFENSAI's v2 launch day(01:55) From immigrant kid to 15 years in startup go-to-market(03:29) Female founders, fundraising, and not taking "no" personally(06:47) Pre-seed angels and the trust behind early funding(07:51) Bootstrapping vs. a Series A market gone inflationary(09:51) Building a lean go-to-market system(12:00) PLG for CISOs: getting to value in the first 15 minutes(14:35) Exploitation validation and the "front door of glass"(16:38) Defying the odds: the first-generation immigrant arc(20:06) The over-hire failure right before the bubble burst(24:11) Capacity planning and "Build It. Test It. Prove It."(26:41) Why a nine-figure Series A sets a trapAbout the guest:Karen Nguyen is co-founder and CEO of OFFENSAI, an AI-powered cloud security testing platform built around autonomous red teaming and continuous exploitation validation. She spent 15 years in startup go-to-market and eight years selling cybersecurity to CISOs before co-founding the company.Connect with Karen Nguyen: https://www.linkedin.com/in/knguyen4/OFFENSAI: https://www.offensai.comWe Built It Because We Had To is hosted by Jonathan Buckley, fractional CMO at The Artesian Network. New founder stories every Tuesday and Thursday — subscribe so you never miss one. Learn more at https://www.artesiannetwork.com

The B2B sales cycle just hit 272 days. That is not friction. That is a buying committee with a new veto holder back at the table.In this episode of We Built It Because We Had To, I sit down with Nick Turner, CEO of Dreamdata, the Copenhagen-founded B2B revenue attribution platform that closed a $55M Series B in October. Nick spent twenty years in MarTech sales before joining Dreamdata as CRO to open the US market, and recently stepped into the CEO seat. We get into the counter-intuitive move that preceded the round — narrowing the ICP and the user persona instead of broadening it — and what Dreamdata's latest LinkedIn benchmark data is telling us about how B2B buying actually works in 2026.You will learn why the average B2B sales cycle stretched from roughly 205 days a year ago to 272 days today, why finance teams are reasserting veto power inside the buyer, why roughly eight different people are active in every account every week, and why Nick cut the Dreamdata product roadmap to six months in an Anthropic-funded world where assumptions expire fast. Nick explains why his North Star metric is the tenure of the CMO who uses the platform, and why you should never ask a marketer to manufacture demand for a deal that has to close this quarter. He closes with three lessons that shaped his path: you are a product of your failures more than your successes, all you need is one yes (he pitched 74 investors to close the round), and know yourself — he once joined a pre-revenue company and fell flat because scaling, not zero-to-one, was his game.If you run revenue, sit in the CMO chair, or own a number that depends on a buying committee saying yes, this one is for you.GuestNick Turner — Chief Executive Officer, DreamdataLinkedIn: https://www.linkedin.com/in/cnickturner/Company: https://dreamdata.ioAbout the showWe Built It Because We Had To is a podcast about early-stage tech founders, the spark behind what they built, and the journey to make it exist. Hosted by Jonathan Buckley of The Artesian Network, where we have helped founders scale companies and have seen more than half of our clients reach an IPO or acquisition.Subscribe wherever you listen, follow the show on YouTube for the full video, and learn more at https://artesiannetwork.com.— Jonathan Buckley, The Artesian Network

On his fourth B2B SaaS company, Michael Riley walked into the most crowded corner of healthcare tech — claim denials — and walked right back out. He pivoted Gabeo.ai into a niche almost no one was working: claim write-offs, the revenue hospitals and provider groups have already given up on. In the first 90 days at a large Midwest hospital group, his agentic AI layer pulled back $1.2 million.In this episode of We Built It Because We Had To, Michael and host Jonathan W. Buckley get into the mechanics of finding a niche inside a saturated market, the two-product strategy at Gabeo (Zoey for fee-for-service, Aria for value-based care), and why value-based care flips the economic incentives of US healthcare. Michael walks through how their AI agents cut through hundreds of pages of contract language that "you'd need a team of lawyers and months to decode."Then they rewind to his arc — learning to program at seven, co-founding Simple Post, surviving a year of failure-after-failure at Funnl before landing on an analytics product that got acquired. Michael gets candid about a 45x-ROI pilot he renegotiated DOWN to a fairer 5-10x, why his entire GTM at Gabeo is warm intros (no paid ads, no sales team), and the single biggest tell that you've actually hit product-market fit: the body language in the room.What you'll hear:- Why Gabeo abandoned denials and chased write-offs instead- How agentic AI cuts through 100s of pages of contract language in minutes- The pricing lesson behind the 45x ROI pilot they renegotiated- 12 product pivots, one acquisition — and what Michael learned about MVPs- Why "selling is asking questions" beats pitching, every time- What he wishes he'd known 15 years agoConnect with Michael Riley: https://www.linkedin.com/in/themichaelriley/Learn more about Gabeo.ai: https://gabeo.aiSubscribe to We Built It Because We Had To for new episodes every Tuesday and Thursday. More from Jonathan W. Buckley and The Artesian Network: https://artesiannetwork.comIf this episode hit, share it with a founder who's still trying to brute-force product-market fit with a bigger sales team.#FounderStory #HealthcareAI #B2BSaaS #ProductMarketFit #AgenticAI

🎙 Season 1, Episode 13Rana Gujral thinks the AGI conversation is missing a fourth axis: Artificial General Experience (AGE). On this episode of We Built It Because We Had To, the founder of Behavioral Signals and author of the upcoming book The AI Instinct: The Future of AI and Human Decision Making explains why intelligence without experience, memory, valence, or a persistent self-model is hollow — and why superintelligence won't arrive as an external god-like machine. It will emerge from hybrid cognition: the tight loop between a human's priors and a machine's scale.We also get into the through line from founding TIZE (acquired), leading Smart Home at Logitech, and helping rebuild Cricut during a near-bankruptcy stretch that turned into a successful IPO. Rana explains why Behavioral Signals doesn't use third-party LLMs and how its paralinguistic models read tone, pitch, prosody, and micro-pauses to extract the signal under the words. He breaks down the company's two business lines (CCaaS and national security, where they're backed by In-Q-Tel), and how their new behavioral mapping approach to deepfake audio detection works at half-second granularity and 98–99% accuracy when classical artifact analysis has stopped working.🎙 In this episode:→ AGE: Artificial General Experience as the missing AGI axis→ Why Behavioral Signals refuses to use third-party LLMs→ Paralinguistic models that read tone, prosody, and micro-pauses→ Deepfake audio detection by behavioral mapping (98–99% accuracy)→ In-Q-Tel backing and the national security business line→ From Logitech Smart Home to Cricut's IPO turnaround→ "Measure what the system does to the person, not just for them"📌 Connect with Rana Gujral:LinkedIn: https://www.linkedin.com/in/ranagujral/🎧 About "We Built It Because We Had To"The Artesian Network's podcast — hosted by CEO Jonathan Buckley. Real founder stories about the spark, the struggle, and the strategy behind building a tech company. More than half of our clients reach IPO or acquisition.→ Subscribe for new episodes every Tuesday and Thursday at 11 AM ET→ More from Jonathan: https://artesiannetwork.com→ Connect with Jonathan: https://www.linkedin.com/in/jonathanwbuckley/⏱ Chapters:00:00 Intro01:08 From Cricut's near bankruptcy to a successful IPO02:30 Behavioral Signals as research roots, not chatbot05:19 Two equal legs: call centers and national security06:56 Why old deepfake detection broke07:43 The cybersecurity manipulation problem10:12 Series A in 2019, before Transformers12:33 Inventing the category nobody had a bucket for16:09 Artificial general experience explained18:50 The chess engine that lives through nothing20:02 Hybrid cognition and the new unit24:23 Measure what the system does to the person#FounderStories #AI #AGI #VoiceAI #DeepfakeDetection

Most cybersecurity vendors win by telling the buyer their existing tools are broken. Stephen Franklin filed a U.S. patent on doing the opposite.In this episode, Stephen — Founder and CEO of Netwatch.ai — walks through how he went from a teenage job at Pitney Bowes Software to building the patented AI integration layer that now sits on top of AWS, Azure, Carbon Black, Meraki, Cisco Cybervision, SolarWinds, and Nutanix. The thesis is simple: stop forcing rip-and-replace. Plug in. Reduce the customer's tool sprawl. Use AI to collapse mean time to contain from hours to seconds.What you'll hear:How a decade of technical sales at Argent taught Stephen exactly which integrations every cyber buyer wishes they had.Why bootstrapping Netwatch from his own Argent commissions kept the product roadmap honest.How utility co-ops — a vertical almost nobody else is actively selling cyber to — became one of Netwatch's sharpest early segments.The two AI "personas" Netwatch ships: a sysadmin and a CISO, with guardrails for what each one is allowed to do.How a Florida statewide cybersecurity contract opened the door to North Carolina, South Carolina, and Georgia.Why Stephen has turned down VC offers at 50 customers and a 100% renewal rate.50 customers in. 100% renewal. A real US patent. No rip-and-replace pitch in sight.Guest: Stephen Franklin, Founder and CEO, Netwatch.aiConnect with Stephen on LinkedIn: https://www.linkedin.com/in/stephen-franklin-ii/Netwatch: https://netwatch.aiHosted by Jonathan Buckley, CEO of Artesian Network. Subscribe to "We Built It Because We Had To, stories about founders and their journeys" wherever you get your podcasts. More at https://artesiannetwork.com.