"I Was Wrong About In-House Software"

What happens when the person who told everyone to build their own software publicly admits it may have been the wrong call?In this episode of KP Unpacked, KP Reddy opens with a public apology (with no regrets). A year ago he encouraged clients and LPs to vibe code, hire in-house developers, and skip the startups. This week, several of them told him the same thing: after a year of building, the startup they passed on is now three years ahead. Nick explains why. Startups put the world’s best engineers on one problem, and AI makes the slope of their improvement vertical while an internal team stays flat.KP then introduces a new framework: vitamins, painkillers, and cures. Build the vitamins and painkillers that are unique to your organization. If it’s a cure, stop building and partner. The conversation covers why KP refuses to talk to CTOs about AI (it’s a business decision), why design partnerships with exclusivity could be the unlock for both sides, what Oracle and Amazon can teach us about it, and why he thinks Q1 is when “things happen fast” arrives for AEC.Key questions answered:Why did KP publicly say he was wrong about vibe coding and in-house software?Why is a startup three years ahead of an in-house team after just one year?What’s the vitamin, painkiller, and cure framework for build versus buy?Why won’t KP talk to CTOs about AI?What does a good design partnership with exclusivity look like?What can Oracle and Amazon teach us about design partners?How should a customer pick which vertical AI startup to back?Why are record backlogs not necessarily a good sign?Why does KP think Q1 is when AEC firms start losing deals to new players?What separates world-class AEC CEOs from firms that are just showing up?If you’re an AEC leader deciding whether to build or buy, a founder wondering how to structure a design partnership, or a CEO who suspects your in-house project is falling behind, this episode will help you make the call.Listen now.

What if the reason your AI rollout isn’t showing ROI is that the business people never made it a business decision?After two months away, KP Reddy and Nick are back with a full catch-up. KP’s advice to a CEO of a 100-year-old company: stop treating AI as an IT project, take a day off, and imagine what your business looks like in 50 years if AI is magic. Then have the conviction to act on it, even if that means cutting 20% of the workforce. Nick explains why his investing lens has shifted toward hardware and deep tech, and why incumbents moving faster than startups on AI is the strongest signal yet that software defensibility is gone.The conversation also covers why KP thinks we’re all in a 1% bubble, how personal AI assistants like Muse are starting to prompt us instead of the other way around, why Formula One is the best analogy for AI and teams, and why data centers are going to be a massive maintenance business in five years. Plus a look at choppy capital markets and the Anthropic IPO.Key questions answered:Why is AI a business decision and not a technology decision?Why are incumbents building AI features faster than startups, and what does that mean for defensibility?Will companies stop buying software because token costs are unpredictable?What does it mean that we’re in a 1% AI bubble?How will personal AI assistants start prompting us instead of the other way around?What does Formula One teach us about teams, machines, and AI?Why will maintaining data centers be a growth business?What happens to capital flows over the next year?If you’re a leader still treating AI as an IT decision, an investor rethinking where defensibility comes from, or just wondering what you missed over the last two months, this episode will get you caught up.Listen now.

What if the firms that win the next decade aren't the ones with the best AI tools, but the ones that invested in people who know how to keep learning?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Nora Swanson, Principal at Trinity JBNB and one of Crain's Notable Women in Tech, ahead of her appearance at AEC Summit on October 19th at The Weylin in Brooklyn. Nora leads her firm's innovation, knowledge, and learning platforms, including AI tools built in-house. Her take: technology is increasingly accessible, anyone can build right now, and that means the real differentiator is no longer which tools you use. It's how quickly your organization can learn, adapt, and apply what it knows.The conversation covers why "yet" and "imagine if" are the two most important words in a growth mindset culture, how Trinity JBNB is embedding AI as infrastructure (not just a resource or a coworker), what it means to redefine expertise when AI handles production, and why this moment gives engineers and builders the chance to focus on the problems that actually matter. The energy crisis. The housing crisis. Carbon reduction. These are the big issues that get unlocked when the small stuff gets automated.Key questions answered:What is the most important thing firms should be investing in that isn't technology?Why are curiosity, adaptability and learning becoming more valuable than software skills?What are the three phases of AI evolution: resource, coworker, and embedded infrastructure?How is Trinity JBNB using AI to automatically update Revit models and notify engineers?What does "yet" have to do with building a growth mindset inside an engineering firm?Who is Carol Dweck and what is the Power of Yet?Why does context matter as much as technical expertise right now?How is the definition of team expanding in AEC?What does it mean to define a profession by purpose rather than production?How do you keep humans in the loop when AI is embedded in workflows?Why are the big engineering challenges more accessible now than ever before?What does reteaming with agents actually look like in practice?If you're leading innovation at an AEC firm wondering where to focus beyond the technology, a young engineer trying to figure out what skills will matter most in five years, or just curious what AI as embedded infrastructure actually looks like in a real engineering workflow, this episode will show you why the firms that invest in people right now will be the ones that win.Listen now.Catch Nora live at AEC Summit on October 19th at The Weylin in Brooklyn.

What happens when a 60-year-old stadium tries to compete with the airport you flew through last week?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Chris Nations, President of Venue Development at Legends, ahead of his appearance at AEC Summit on October 19th at The Weylin in Brooklyn. Chris founded Nations Group in 2008, spent his career managing athlete venues and capital projects from Oregon State to Penn State's Beaver Stadium to the new Dallas Mavericks arena, before Legends acquired Nations Group in 2025. His read on the market: sports venue investment has never been hotter, and it's coming from two directions at once. Capital chasing sports. And fans showing up and spending money.The conversation covers the three-headed monster of every venue upgrade (infrastructure and deferred maintenance, premium diversification, and general fan experience), why premium is always the driver even if nobody wants to admit it, how Alaska Airlines and Inspirato vacation homes explain the explosion of premium tiers in stadiums, and why Penn State's Beaver Stadium renovation is one of the best examples of combining the new with the old without killing the atmosphere. Then Chris lands the line that ties it all together: we could see more change in design and construction in the next seven to eight years than in the last fifty.Key questions answered:Why is sports venue investment at an all-time high right now?What is the three-headed monster of every venue upgrade project?Why does premium always get prioritized even when infrastructure is falling apart?How did club and suite become 12 to 20 different levels of premium?What does Alaska Airlines finally going premium have to do with stadium design?How does Legends use fan survey data to design revenue into a project before it's built?What made Penn State's Beaver Stadium renovation get it right?What is the Dallas Mavericks' vision for the best fan experience in the NBA?Why is the airport you flew through last week the right benchmark for stadium experience?How do you get 100,000 people a hot dog and a Coke and back to their seat in time?Why could AEC see more change in the next seven years than in the last fifty?What does owning the fan experience from the moment they leave their house actually mean?If you're in AEC wondering where the biggest capital projects are flowing, a municipality trying to understand what drives stadium investment decisions, or just curious what the Dallas Mavericks arena is going to look like, this episode will show you why the fan experience problem is really a design and construction problem in disguise.Listen now.Catch Chris live at AEC Summit on October 19th at The Weylin in Brooklyn.

What if the best innovation in construction has never been the technology, but the people willing to listen?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Ravi Bhatia, Business Development Manager at Skanska USA Civil in New York, ahead of his appearance at AEC Summit on October 19th at The Weylin in Brooklyn. Ravi brings 35 years of commercial high-rise, transportation, critical systems, and design-build infrastructure experience, and teaches pre-construction and planning at Fordham. His through line across all of it: innovation is never really about technology. It's about people with a growth mindset who care about what they do.The conversation covers the full arc from CAD to BIM to 3D, 4D, and 5D modeling, why a change request that took a week in 2005 takes 20 minutes today, and why the next unlock isn't capital cost efficiency but lifecycle cost. Buildings, bridges, tunnels, airports: the question that should be on every bid evaluation is not who's cheapest, but who's building something future-proof. And the best ideas? They come from the factory floor, not the corner office.Key questions answered:Why does innovation in AEC always start with safety first?What is the through line from CAD to BIM to 3D, 4D, and 5D modeling?How did a change request that took a week in 2005 shrink to 20 minutes today?Why is lifecycle cost the next frontier for AEC innovation?What does future-proof actually mean and how do you quantify it for an owner?What's the difference between DBIA and DBOIA and why does it matter?Why do the best practical ideas in construction come from the field, not the office?What did a GE factory foreman tell a CEO that stopped everyone in their tracks?What would Ravi say to a young professional considering a career in AEC?Why is the project leader actually a project servant?What does servant leadership look like when your job is to remove hurdles, not give orders?How does curiosity connect the dots from CAD to whatever comes next?If you're a young professional wondering whether AEC is the right industry, a project leader trying to figure out what servant leadership actually looks like in practice, or an owner trying to understand why lifecycle cost should be a bid criteria, this episode will remind you why the people who build things that outlast their own lifetimes never lose their curiosity.Listen now.Catch Ravi live at AEC Summit on October 19th at The Weylin in Brooklyn.

What if the biggest impact of AI in construction isn't cutting costs or headcount, but raising the standard of what clients expect?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Brad Hargreaves, co-founder of General Assembly and Common, and founder of Thesis Driven, a newsletter reaching the investors, developers, and operators building American cities. Brad's take on AI in AEC cuts through the noise: the tools getting real enterprise adoption aren't the ones generating plans. They're the ones checking them. Validating them against code. Catching conflicts before they become change orders. And once clients know that capability exists, the question stops being "did you use AI?" and starts being "why didn't AI catch this?"The conversation covers where AI savings are actually showing up in multifamily operations (hint: it's collections and renewal optimization, not headcount cuts), why the unified data layer is still largely unsolved, why feedback loops at project close are the most underutilized AI opportunity in construction, and why no amount of automated conflict checking can replace the human layer of project alignment. Technology is the augmentation. Teamwork is still the unlock.Key questions answered:Why are plan checking tools getting enterprise adoption before plan generation tools?What does "why didn't AI catch this?" mean for client expectations going forward?Where are AI savings actually showing up in multifamily operations?Why was everyone's prediction about AI headcount cuts mostly wrong?What is a positive error rate and why do different firms come to different conclusions?How are tools like Trunk Tools, OpenSpace, and Alice Technologies connecting the dots across trades?Why is the unified data layer still largely unsolved across design and construction teams?Why are feedback loops at project close the most underutilized AI opportunity?What does Brad mean when he says AI can't replace the human layer?How does optimizing the worker's path generate savings without eliminating the worker?What's the difference between AI as efficiency gain versus AI as quality bar raiser?How does Thesis Driven think about second and third order effects of AI in AEC?If you're in AEC wondering where AI is actually generating value versus where it was overhyped, a developer trying to understand what's coming next in multifamily operations, or building tools for the construction industry and trying to figure out what gets enterprise adoption, this episode will show you why the savings showed up exactly where nobody predicted.Listen now.Catch Brad live at AEC Summit on October 19th at The Weylin in Brooklyn.

What if the biggest factor in America's reindustrialization isn't capital, technology, or policy, but whether your state ever learned to play well with others?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Tom Stringer, founder of Stringer Site Selection, ahead of his appearance at AEC Summit on October 19th at The Weylin in Brooklyn. Tom has spent two decades deciding where things get built. Aerospace plants, EV lines, defense facilities, Fortune 500 headquarters. He's secured billions in incentives and he's a licensed attorney and real estate broker in New York. When a company picks a state, a county, a site, Tom is often the one in the room.The conversation covers why the CHIPS Act was the first real national industrial policy since World War II, why COVID exposed the fatal flaw in just-in-time supply chains, and why SpaceX is responsible for the volcanic eruption of defense tech startups now building in Southern California. Then Tom gets surgical about what new manufacturers actually need from municipalities, and why the states winning the reindustrialization race aren't the ones with the most incentives. They're the ones with the most integrated teams. Texas. Arizona. The Carolinas. Ohio. Florida. The same names keep coming up, and it's not an accident.Key questions answered:Why is this the first real national industrial policy since World War II?What did COVID expose about just-in-time supply chains that economists missed for 20 years?Why is SpaceX the origin story for the defense tech boom happening right now?What do new manufacturers actually need from municipalities and site selectors?Why does the workforce problem create a level playing field for unexpected states?Which states are winning reindustrialization projects and why are the same names always on the list?Why do incentives alone not win projects anymore?What does bringing back shop class have to do with national security?Why is New York losing despite offering more incentives than almost anyone?What does American Idol leaving California for Atlanta tell us about value propositions?How do utilities, community colleges, permitting offices and economic development agencies have to work together?What does project success as a team sport mean at the scale of a federal republic?If you're in AEC wondering where the next wave of industrial projects will land, a municipality trying to figure out how to compete for reindustrialization investment, or just curious why the same five states keep showing up in every major site selection announcement, this episode will show you exactly what separates the winners from the ones that need to go back to the gym.Listen now.Catch Tom live at AEC Summit on October 19th at The Weylin in Brooklyn.

What if the most human-centered innovation in construction isn't software, a robot, or a platform, but what workers are wearing on the job?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Tiffany Yeh, CEO of Eztia and Forbes 30 Under 30, ahead of her appearance at AEC Summit on October 19th in Brooklyn. Tiffany walked away from medical residency to build HydraVolt, a cooling gel worn as apparel that drops skin temperature roughly 10 degrees Celsius for up to eight hours, with no batteries, no maintenance, and no infrastructure required. It is already in the field with construction crews, logistics teams, and public safety officers, backed by a $4M seed round led by Shadow Ventures.The conversation covers why Tiffany chose to cool people instead of spaces, how her mission was shaped by living in global health settings where air conditioning was never an option, and why the hydrogel technology behind HydraVolt was originally developed to passively cool buildings and store vaccines during COVID. The thread running through all of it: technology should serve all of society, not just the people who can afford the infrastructure. And the orchestra doesn't sound good unless every instrument is playing in tune.Key questions answered:Why did Tiffany walk away from medical residency to build a cooling apparel company?What is HydraVolt and how does it drop skin temperature 10 degrees with no batteries?Why did Eztia choose to cool people instead of buildings?How does mission shape product trade-offs in resource-constrained environments?What does human-centered innovation actually look like on a construction jobsite?How did vaccine cold chain research during COVID lead to wearable cooling technology?Why does Tiffany believe technology should be accessible to people without infrastructure?What does a soccer team have to do with how startups should think about jobsite adoption?How do you sell to the safety team, the sustainability team, and the individual worker all at once?What does stadium design have to do with personal cooling strategy?What would Tiffany build if resources were unlimited?What does project success as a team sport mean when every player has different goals?If you're in construction wondering how to keep crews safe and productive in extreme heat, a founder thinking about how mission shapes product decisions, or just curious what happens when a materials engineer and MD walks away from medicine to solve a global problem, this episode will make you rethink what human-centered innovation actually looks like from the ground up.Listen now.Catch Tiffany live at AEC Summit on October 19th at the Weylin in Brooklyn.

What if the reason we can't electrify the world isn't that we don't have good enough batteries, but that we can't make enough of them?In this September takeover episode of KP Unpacked, guest host Tyler Sellhorn sits down with Alexander Girau, CEO and co-founder of Group1, ahead of his appearance at AEC Summit on October 19th in Brooklyn. Alexander has worked on two Nobel Prize-winning technologies out of UT Austin, including technology developed by John Goodenough, the inventor of the lithium-ion battery. Group1 is his latest venture, commercializing potassium-ion battery chemistry that moves faster than lithium and sodium, and doing it right here in Texas.The conversation covers why the US has lost two battery technology leads to Asia already (lithium and sodium), why this is America's third shot and it's staying home, and why AI data centers have completely changed the demand equation. A few years ago it was EV madness. Now it's NVIDIA, Google, Microsoft, and Amazon all needing electrons and battery stability at infrastructure scale, faster than Group1 can grow. The insight? One person with AI agents is now as effective as ten. And startups don't die from starvation. They die from ingestion.Key questions answered:Why is the battery problem a manufacturing and supply chain problem, not a chemistry problem?What is potassium-ion chemistry and why does it matter for AI data centers?How much energy storage does the world actually need, and how far behind are we?Why did the US lose the lithium and sodium battery leads to Asia?How is Group1 keeping this technology from leaving the country again?What does re-industrialization actually look like at a battery manufacturing plant in Austin?How did AI data centers completely change the demand equation for battery storage?How does Alexander use AI agents to make one person as effective as ten?Why do startups die from ingestion, not starvation?What does a special ops team have to do with pre-product market fit hiring?What does "CEO as chief enabling officer" actually mean in practice?How do you cross the gap between a Nobel laureate's lab and a manufacturing plant?If you're in AEC wondering why data centers keep coming up in every infrastructure conversation, a founder trying to figure out how to scale manufacturing without scaling headcount, or just curious what deep tech commercialization actually looks like from the inside, this episode will show you why the battery problem is closer to solved than you think, and why the hard part was never the chemistry.Listen now.Catch Alexander live at AEC Summit on October 19th at The Weylin.

What if the reason construction robotics never scaled wasn't the technology, but the approach?In this special September edition of KP Unpacked, guest host Tyler Sellhorn sits down with Nadav Shuruk, co-founder and COO of Okibo, ahead of their appearance at AEC Summit on October 19th in Brooklyn. Nadav breaks down why every other construction robot has failed to reach real scale, and why Okibo's EG7 took a fundamentally different path. No BIM. No site prep. No markers. No pre-programming. You unload it from the truck, press start, and it finishes roughly 1,000 square feet of drywall or paint in an hour.The conversation covers how Okibo was born from a family robotics business in Israel, why none of the founding team had construction experience (and why that actually helped), and how the robot's multi-task platform keeps utilization high every single day of a project. Mudding on Monday, sanding Tuesday, priming Wednesday, painting Thursday. The robot always has something to do, and that's exactly how you build trust in a conservative, low-margin industry that has been promised hardware solutions for decades and never got them.Key questions answered:Why have construction robots failed to scale and what did Okibo do differently?How does the EG7 navigate a job site without BIM, GPS, or expensive sensors?What's the right split between what the robot does and what the operator does?How did a family robotics business in Israel end up building drywall robots?Why does multi-task utilization matter more than raw performance in construction?How long does it take to train an existing painting or taping crew to operate Okibo?What data does Okibo collect daily and how does it help subcontractors bid better?Why does the GC's buy-in determine whether a robot succeeds or fails on site?How does robotics connect the value chain from owner to GC to subcontractor?Why does one plus one equal more than two when operators and robots work together?What does project success as a team sport mean for robotics on a job site?Where can you see Okibo in person at AEC Summit in Brooklyn?If you're a subcontractor wondering whether robotics actually works in the field, a GC trying to understand what buy-in really means for new technology, or an owner curious about what data access from a robot looks like on your project, this episode will show you why the hardware the trades were promised is finally here.Listen now.Catch Nadav live at AEC Summit on October 19th at the Wayland in Brooklyn.