Adam Feldman

👉 Capital Factory is hosting Arca Continental AC Ventures at Austin Tech Week, showcasing six breakthrough companies - check out the curated list here! 🏭 Creative 3D Technologies helps manufacturers build finished products with one machine so they can ship faster, cut costs, and make parts you can’t with traditional tools. Their “factory-in-a-box” prints, mills, lasers, and assembles in one system. 👉 creative3dtechnologies.com 📝 Intellectible helps nonprofits write winning grant proposals so they can secure funding faster. Upload your info and requirements -they turn it into a polished, ready-to-submit document in hours, not weeks. 👉 intellectible.com 📱 Skyllful helps warehouse and delivery teams learn the exact taps and swipes in their work apps so they can pick, ship, and service without mistakes.  👉 skyllful.com 🚆 Glīd improves freight logistics via cargo trucks that seamlessly transitioning between road and rail. They replace long-haul trucking with trucks that can travel on train tracks. It's faster and cheaper.  👉 glidrail.com 🏟️ Zippin helps stadiums sell food & drinks without lines so fans get back to their seats faster. Their checkout-free stores boost throughput and cut labor headaches - already rolling out at venues like NRG Stadium and Petco Park. 👉 getzippin.com 🔩 Emmet helps large manufacturers buy steel and automate their supply chain so they can keep lines running and cut costs. They plug into your systems to auto-reorder, track deliveries, and prevent stockouts. 👉 emmethq.com Who do you know that one of these entrepreneurs needs to meet?

47

🚀 API Watchdog – Building in Public Update This week’s lesson: your API key can absolutely get robbed. I spent days this week convinced I was the problem. Token counts too high. Prompts too verbose. Models too expensive. So I did what any good builder does: ✂️ trimmed prompts 🧮 optimized context 📉 refactored calls 🤔 questioned my life choices And then I checked the logs more closely. Turns out… Someone had stolen one of my OpenAI API keys and was hammering it with math competition problems and foreign-language queries — on the newest, most expensive models — at times I wasn’t even online. Attached are some examples. (Seeing LaTeX-heavy number theory prompts at 3am is… a vibe.) So yeah — my “optimization sprint” was actually a security incident. What I did immediately: 🔒 Revoked every API key 🔁 Rotated keys + separated prod / staging / dev 💸 Added hard usage caps and alerts 🏷️ Tagged every request with metadata (never skipping this again) 🕵️ Audited logs until the pattern was undeniable Lessons learned (the hard way): • Token optimization doesn’t matter if your key is public • Spend limits are not “nice to have” • Logs without metadata should make you suspicious • If usage looks weird, assume compromise before rewriting your app API Watchdog progress didn’t stop — but this week’s unexpected feature was: detecting suspicious API usage in my own system. Back to building the fun stuff next: Flows, alerts, scheduling, and yes — security guardrails baked in by default. If you’re building with LLMs: Rotate keys. Set caps. Tag requests. It’s not paranoia — it’s production. (And if you’ve got an API horror story, please tell me so I feel less alone 😅)

11

1 Yorum

Breaking: Claude Code just introduced Sub-Agents You can run your own custom Agents. Using the /agent command Based on an agent file format, written in markdown. Custom system prompts, custom tools. Making it very easy to create your own agents. Previously I made a REST api for Claude Code, meaning so I could script things from a browser. Now, I can create teams of sub agents, working together on a problem. More info on the feature: https://lnkd.in/g3H4_q7E You would think they are paying me, I'm just hyped because this allows Spec Driven Development to be even easier.

73

23 Yorum

8 books every builder should read before launching anything. I've read some of these. Others are on my list. But every product person I respect keeps recommending the same ones. 𝟭. 𝗭𝗲𝗿𝗼 𝘁𝗼 𝗢𝗻𝗲 — Peter Thiel Stop competing. Build something new. This book rewires how you think about differentiation. Every founder I've talked to has read this. 𝟮. 𝗧𝗵𝗲 𝗠𝗼𝗺 𝗧𝗲𝘀𝘁 — Rob Fitzpatrick People lie when you ask "would you use this?" This book teaches you how to validate ideas by asking better questions. 𝟯. 𝗜𝗻𝘀𝗽𝗶𝗿𝗲𝗱 — Marty Cagan The bible for product managers. If you build anything for users, this is the starting point. The product discovery chapters alone are worth it. 𝟰. 𝗧𝗵𝗲 𝗟𝗲𝗮𝗻 𝗦𝘁𝗮𝗿𝘁𝘂𝗽 — Eric Ries Build. Measure. Learn. Sounds simple until you realize you've been building for 3 months without measuring anything. This book fixes that instinct. 𝟱. 𝗛𝗼𝗼𝗸𝗲𝗱 — Nir Eyal Why some products become habits and others get deleted after a week. Short read. Completely changes how you think about user retention. 𝟲. 𝗦𝗽𝗿𝗶𝗻𝘁 — Jake Knapp Solve big problems in 5 days. The Google Ventures framework for testing ideas fast instead of overthinking them for months. 𝟳. 𝗧𝗵𝗲 𝗛𝗮𝗿𝗱 𝗧𝗵𝗶𝗻𝗴 𝗔𝗯𝗼𝘂𝘁 𝗛𝗮𝗿𝗱 𝗧𝗵𝗶𝗻𝗴𝘀 — Ben Horowitz No motivational talk. Just the brutal reality of running a company. Everyone recommends this one for a reason. 𝟴. 𝗖𝗿𝗼𝘀𝘀𝗶𝗻𝗴 𝘁𝗵𝗲 𝗖𝗵𝗮𝘀𝗺 — Geoffrey Moore Why great products fail after early adopters. Most builders never think about this gap. This book makes you rethink who you build for first. If you only pick one, start with The Lean Startup. It'll save you months of building the wrong thing.

2

I founded Bolt Data & Energy because energy is the single biggest constraint on scaling AI. We can build better models and better algorithms, but the physics of computation are unavoidable: advanced AI requires enormous amounts of power. For the United States to remain competitive, we have to solve the infrastructure problem—not someday, but now. Scaling compute to meet virtually limitless demand requires three things beyond technical expertise: cheap and reliable energy, vast land, and abundant water. That is why Bolt has partnered with Texas Pacific Land Corporation (TPL). TPL provides unparalleled access to abundant resources and some of the lowest energy costs in the world, thereby allowing us to build world-class infrastructure to meet global demand at scale. Our goal is to build the largest and most efficient network of AI data centers in the world, powering the next generation of artificial intelligence and fueling the Fourth Industrial Revolution. We have a responsibility to shape AI in ways that positively impact the world and reflect democratic values.  Whoever leads in providing the energy that powers AI will help define its trajectory.  We are just getting started! Read more here: https://lnkd.in/ejrNUuU6 #AI #Data #DataCenters #Energy #Compute #Technology #AI #Innovation

737

82 Yorum

Austin Network...this is going to be a great one! 🚀 On March 30, the Flutter Austin + GDG Austin communities are hosting a very special meetup with Marcus Twichel (Flutter/Dart GDE, Tech Lead at Very Good Ventures) live...in the flesh! He’ll be walking through a live demo of a coffee shop platform running across 8 different screens from a single Flutter codebase - mobile, web, kiosks, POS, kitchen displays, menu boards, and more - all syncing in real time. Then he’ll break down the architecture behind it! If you’re a Flutter engineer, curious about serious cross-platform architecture, or a tech professional in Austin, this will be a really valuable session. 📍 Voltron Room - Capital Factory 🕡 Doors open at 6:30 PM ⚠️ Limited to 200 spots (in person) 🏝️ Virtual attendance for those outside of the area Food, drinks, networking, and some very cool engineering discussion! Grab a spot here 👇 GDG Event: https://lnkd.in/gcTY-t-w Meetup: https://lnkd.in/gHzmfWCu #Flutter #AustinTech #SoftwareEngineering #GDGAustin

4

3 Yorum

125 OpenClaw startups now listed on TrustMRR. $274,397 of verified revenue generated in the last 30 days. If this represents ~1% of the market, @steipete's open source project helped developers make around $30M in February.

76

4 Yorum

Vibe coding on V0 vs Replit. Who's the winner? 🤖 My verdict: 🔧 Build on Replit for functionality, versatility, extendability. Shorthand: B2B SaaS 🚀 Launch on v0 by Vercel for speed, lose structure, specialized offerings like Composable Commerce, Marketing Sites. Shorthand: D2C & E-com The design of the prompt box says it all. ------ SPEED You can build fast on V0 and collaborate with it. But if you are more clear on what you are building, Replit is a better investment of your time. ACCURACY In the build phase, Replit consistently makes better design choices, more grounded product decisions, and fewer hallucinations. V0 sometimes takes leaps into generating visuals and text instead of using original assets. EXTENSIBILITY Replit builds more durable code. Small details like: "The filename of the image you uploaded has spaces in it. I will rename the file." These details will add up over time. In general Replit has better "awareness" of what it is building. SPECIALIZED USE CASES Vercel is rapidly specializing in conversion optimization, SEO, e-com stack. If you want to give your application built-in feedback loops like A/B testing so it can evolve on it's own, V0 is a better tool. What's your experience so far?

8

1 Yorum

🧨 The Hidden Cost of Your Mono-Repo: A $200K+ Annual Tax on Your Business First in a series on untangling monolithic codebases and reclaiming your team’s velocity 💥 The Silent Cost Your Engineering Team Lives With Your mono-repo might be the most expensive “free” decision your team ever made. Once, it helped with simplicity. Now? It’s slowing everything down. Every day your engineers “just deal with it,” while your business loses time, money, and momentum. 🚩 What You’re Losing 🚀 Delayed Time to Market Simple features that should take 2 days now stretch to 2 weeks: • Long build times • Coupled deployments • Dev teams waiting on unrelated merges or CI passes Meanwhile, your competitors ship 3–5x faster. 💰 Productivity Drain in Plain Numbers Let’s say you have 10 engineers. If each loses just 1 hour/day fighting CI/CD issues, broken builds, or merge hell: → That’s 2,500 hours/year → At $80/hr = $200,000/year wasted And that’s not even counting the opportunity cost of missed features, lost deals, or churned users. ⚠️ Increased Risk, Slower Recovery • One config change breaks production • Frontend tweaks touch infrastructure • Every release feels like a gamble 🏃‍♂️ Your Best Engineers Will Leave Top talent doesn’t want to untangle scripts or wait 40 minutes for a test run. They want to ship value, not fight their tools. Exit interviews often include: “Too much tech debt” “Too slow to deliver” 🔍 Why This Happens What started as “simple” became a bottleneck: • App logic, infra (Bicep/Terraform), configs, and pipelines all tangled together • No boundaries • No team ownership • One repo, hundreds of problems ✅ But It Doesn’t Have to Be This Way We’ve helped mid-sized companies turn this around - without halting delivery. This series shares a battle-tested roadmap for splitting mono-repos into clean, modular components - each with its own purpose and ownership. 💬 Let’s Talk Seeing signs of mono-repo friction? Drop a comment or DM me - I’ll share practical advice tailored to your setup. And if you’ve tackled this before, I’d love to hear what worked (or didn’t) for your team. #SoftwareArchitecture #DevOps #MonoRepo #AzureDevOps #CI/CD #PlatformEngineering #TechnicalDebt #MidMarketTech #EngineeringVelocity

16

5 Yorum

Interesting insights about AI (you never learn enough) by Blair Garrou of Mercury Fund at the Plug and Play Tech Center Fall Expo 2025 here in Sugarland Texas Key Insights: 1) B2B over a 10 year span was more favorable towards middle market and SMBs. AI However, is now all in on enterprise 2) Revolutionizing/Modernizing older industries is where AI will be seamless (in construction, certain types of engineering and older processes in energy) 3) It's better to invest in AI that equips workers with another skillset than ones that result in layoffs #startups #venturecapital #ai #emergingtech

6

1 Yorum

New from Buyers x Builders 🚀 Dispatches from the frontier of enterprise AI. Every week, Dave Ambrose interviews 2-3 startup founders building the enterprise workflows and infrastructure of the future. First up: Paul Klein IV from Browserbase on automating agentic workflows. It's called Dispatch. It's weekly. And if you want to know what's actually being built (not just what's being hyped), this is it. Link in comments. Let us know what you think 👇 #BuyersxBuilders #EnterpriseAI #AgenticAI #Procurement

17

3 Yorum

Apprentage fulfills DOL requirements and guidance for shared infrastructure for data collection, performance reporting and compliance. We’re doing it so you don’t have to. * WIOA Section 121(h)–(i) (29 U.S.C. § 3151 - Performance under § 3151 requires: - Unified data reporting - Cross-program outcome tracking - Integrated case management - Shared participant records * TEGL 17-16 - A unified performance system (data collection, validation, reporting) assumes shared infrastructure * TEGL 23-19 (Change 3) - Updates documentation guidance (e.g., wage sources, functional gain definitions) tied to performance reporting. * TEGL 10-16 - a foundation for shared infrastructure across programs. #apprenticeships

2

We’re thrilled to share that Usul (YC S24) has raised $3.3M in Seed funding led by Scout Ventures with participation from BVVC, Teamworthy Ventures, True Ventures, Y Combinator and more. Usul is the modern platform for defense contract capture and federal growth. Our customers have won hundreds of millions in defense contracts for capabilities across Air, Land, Maritime, AI/ML and more. When Oliver Gomez, Joonghyun Lee and I joined forces during Stanford’s Hacking 4 Defense class, we interviewed 300+ defense companies to intimately understand how difficult it is for national security solutions to get deployed quickly at scale. In normal capital markets this is called dead weight loss. In warfighting, this costs lives. We're building an elite team of hackers and defense enthusiasts in San Francisco. If you’re interested in powering the next 1000 defense primes, email us at founders@usul.com or dm me. Read more about our journey in Erin O'Brien's Tectonic article below in the comments. Lastly, thank you to our amazing cap table for joining the ship. Bradley C. Harrison, Cody H., Nate George, Zach Beecher, Michael Keane, Joe Musselman, Gene Ebersole, Nick Nowlin, Thomas Lehrman, Stephen Schmalhofer, Michael Montano, Gustaf Alströmer 🇺🇦, Jared Friedman, Steve Blank, Jacqueline Tame, Jack Shanahan, Peter Newell, Steve Weinstein, Jeff Decker, Ivan Kirigin, Jeff Heitzman, Keith W. Gibson, CFCM, SCPM, John Tincoff, Michael Malis, Austen Allred, Tom Hsieh, Donald Donckers

448

77 Yorum

Your startup doesn’t need a dev team. Why? It needs someone who actually understands healthcare. THE $6M HEALTHCARE MELTDOWN: Recently I saw a A VC-funded startup hired 12 “top engineers” to build their AI-powered RPM platform. ✔They shipped features. ✔They built dashboards. ✔They even had real patients onboarded. But 6 months later? The entire platform was dead. Why? Because not one engineer knew: What PHI meant • How HIPAA audit logs work • Why FHIR validation breaks at scale • That HL7 v2 is STILL used in real hospitals That one wrong API leak = $1.5M fine They didn’t need engineers. They needed interpreters of the chaos that is healthcare tech. Here’s what I’ve seen after working with 50+ healthcare orgs:   Regular devs: “What’s the spec?”   Our devs: “Where’s the compliance risk?”  Regular devs: “Let’s integrate with GPT-4.”   Our devs: “Where’s your BAA? What’s your de-identification process?”  Regular devs: “We built the feature!” Our devs: “Clinicians can’t use this on mobile — you just killed adoption.” The truth? Most dev teams are dangerous in healthcare. They move fast, break things and leave founders holding the liability bag. If you're building in healthtech, here’s the playbook: •Don’t hire code monkeys. •Don’t hire architecture obsessives. Hire compliance-native, FHIR-fluent, clinical-aware engineers who know the battlefield. Your startup doesn't need speed. It needs precision in a regulated minefield. •Build safe. •Build smart. Or watch your platform die in beta.

17

1 Yorum

A couple Ah-ha's this week. Adlib hosted customer roundtables at SSON's Shared Services Outsourcing event in Houston. Theme - "The AI Payback Clock: Why Accuracy Upstream Matters". Tagline - "Every $1 spent cleaning docs saves ~$5 in AI model performance, HITL rework, and compliance remediation". Three takeaways from customer conversations: 1. AI is Only as Strong as Its Inputs: Customers stated a shocking % of enterprise docs/data are unusable by AI. In regulated industries, unstructured data often arrives incomplete, inconsistent, or locked away in legacy formats. Feeding this into an AI system creates more cost than value...resulting in stalled ROI, rework, or worse, regulatory exposure. Takeaway: Before investing in new AI use cases, orgs must quantify what % of their critical docs are actually AI-ready...and calculate the cost of the gap. 2. Biggest ROI Lever Is Upstream: AI budgets are often over-invested in downstream model training or workflow integration, while upstream doc accuracy is underfunded. Takeaway: Address doc/data intake standards, standardization, provenance checks...enables companies to unlock faster cycle times, higher first-pass yield, lower exception handling costs and ultimately better end-customer experiences 3. Future-Proofing AI Means Being Board-Ready: Execs stressed AI success isn’t measured in pilots (recent MIT study = noise); it’s measured in boardrooms by showing sustained ROI, reduced compliance risk, improved CX and scalable impact. Takeaway: Orgs that future-proof AI will be those that can demonstrate measurable payback, not just experimental wins. Closing Thought: Deep customer conversations are invaluable. They are the single most important thing that helps me position Adlib to support customers today, but also invest smartly to help them get the W tomorrow. AI is proving to be a powerful innovation cycle and I'm proud of Adlib's participation in helping enterprises position for success by turning unstructured, complex docs/data into AI-ready, trusted inputs that are compliant and auditable...critical for regulated industries. p.s. Thanks to our fiercly loyal customer LyondellBasell for sharing their Adlib story with the group. p.s.s. Very proud of how the Adlib team engaged with customers and represented our brand - true professionals...Craig R., Molly Morgan, Alex Murray, Mike Chasteen and Kunal Bargotra. #IDP #AI #SSON

38

2 Yorum

We are moving quick with our Railroad Commission of Texas MCP. We should have all datasets by the end of the week, sans unstructured data. Here’s what’s avalible right now. searchWells searchWellsByArea searchInspections listPipelines searchPipelineSegments searchGasProcessingPlants

6

Qualify for AI/Robotics R&D Tax Credits If your businesses are: • Developing AI systems • Building robotics prototypes • Creating proprietary software • Improving algorithms • Designing automation processes • Developing new service platforms • Building robotics hardware integration You likely qualify. How It Reduces Your Taxes It is a dollar-for-dollar tax credit. Example: Annual Income : $700k If your federal tax liability is $180,000 and you generate a $75,000 R&D credit Your federal tax drops to: 👉 $105,000 That’s real cash retained. #AI #Robotics #Innovation #TaxCredits #Taxes #IRS #Startups #Entrepreneurs #Founders #CTOs #HighNetWorth #HNW #Wealth #FamilyOffice #FamilyBusiness #TaxStrategy

6

Would your product still pass HIPAA if tested tomorrow?   Most startups believe they will, until a hospital audit proves otherwise.   In healthcare, trust begins and ends with data protection. If your product can’t guarantee patient privacy, no amount of innovation will get it through the door.   Why HIPAA compliance matters? - Protects patient data from unauthorized access and misuse - Builds credibility with healthcare providers, insurers, and enterprise partners - Enables integrations with hospitals and large healthcare systems - Shields your business from costly legal and reputational damage   How to achieve true HIPAA compliance? - Encrypt all data both in transit and at rest - Limit access by role and enforce multi-factor authentication - Sign Business Associate Agreements (BAAs) with every third-party partner - Conduct regular risk assessments and update policies accordingly - Train your team on HIPAA protocols, data handling, and incident response   HIPAA compliance is not a technical formality, it is the foundation of trust in healthcare.   Is your product built on that foundation?

13

1 Yorum

If You're a builder you need to read this You can build software for only three types of people. Everything else is distraction. 1. Time-rich, highly technical users (Developers, hackers, builder community, Startup employees) * Early adopters * Very high standards * Care deeply about perception, purity, and flexibility * Prefer free or low-cost tools * Can pay, but usually choose not to * Time-rich → they’ll stitch together five tools and call it “better” If something can be customized or hacked, they will do it. 2. Time-constrained, outcome-driven users *(SMBs, founders, influencers, operators)* * Often underserved by most software * Adopt later, but with clearer intent * Need simpler, outcome-first products * Comfortable paying if value is obvious * May not explore every feature * Decisions are driven by promised results and urgency They don’t want tools. They want results with minimal thinking. 3. Skilled freelancers and prosumers *(Designers, marketers, creators)* * Spoilt for choice * Already have many capable free options * Will adopt only when:  * There is clear differentiation over free tools  * Peer adoption is visible  * It’s packaged as a system or workflow  * There is a clear category leader They follow momentum more than features. A common founder mistake When you build primarily for group 2, most criticism will come from group 1. Questions like: “This can be done for free.” “Why not just use the native tool?” “Why build this layer at all?” Examples we hear often GrowEasy Why not use Facebook Ads Manager directly? DesignEasy Why not use Nano Banana, Midjourney, Canva, or something similar? -The correct framing Because we are not building for group 1. We are building for group 2 people who don’t want control, flexibility, or technical depth. They want: * Fewer decisions * Lower cognitive load * Faster outcomes * Someone else to think through complexity Abstraction is not a limitation for them. It is the product. The quiet truth If a product is loved by technical users, it often feels complex to SMBs. If SMBs are paying happily, technical users will usually question it. That’s not a flaw. That’s clarity of ICP. Choose your audience. Then build relentlessly for them.

6