David Blonski
San Francisco, California, United States
3K followers
500+ connections
About
Engineer, turned investor, turned founder. I'd rather be in front of a customer than…
Activity
3K followers
-
David Blonski reposted thisDavid Blonski reposted thisSnowflake Summit is less than a month out, and it’s a critical event at a time when every enterprise is wondering how AI will become REAL and deliver VALUE. Two keynotes worth the trip on their own. Emmanuel Frenehard, Chief Digital Officer at Sanofi, is on the Opening Keynote with Sridhar Ramaswamy. What Sanofi has built with Elementum AI and Snowflake is the clearest picture of governed agent orchestration to run global processes across IT, Procurement and GTM at scale in a regulated industry. Patrick Duroseau, Chief AI Officer at Under Armour, is on the Platform Keynote with Christian Kleinerman . UnderArmour's Snowflake data foundation and Elementum workflows take the agent era deep into the era of business value. Then there's our SAASpocalypse party - the one room at Summit where nobody's pretending per-seat has another decade in it. What’s real and what’s hype? Bring your SaaS renewal horror stories. We'll bring the drinks.
-
David Blonski reposted thisDavid Blonski reposted thisEvery legacy enterprise app you bought in the last fifteen years is four things stacked together. Three you bought from the vendor. The fourth you had to staff yourself. It's a UI. It's a logic engine. It's a database… And — because the system was designed for an army to push the buttons — it's a lot of people processing the invoices, the tickets, and the exceptions. You paid one vendor for the stack and a separate services firm for the people. That was never about engineering. It was about pricing power. Here's what's happening to each layer right now. The “button pushing” layer is collapsing. For every dollar of license spend, large enterprises typically spend another one to two dollars on the people who were once necessary to run the system. AI is collapsing that business model by an order of magnitude. One of the largest line items in your legacy SaaS total cost is dropping fast. Now, let’s talk about the UI. Legacy UI is a liability—it’s why everything is going Headless. Your corporate GPT is now the front door. Employees aren't going to be opening legacy ITSM systems to file a ticket — they're asking an agent your AI team built in two weeks. The logic layer is splitting. About 30% of enterprise workflows are genuinely agentic. The other 70% are deterministic. Modern orchestration runs both. Your legacy vendor's engine wasn't built for either. The data layer is moving. Once your operational data lives in a modem cloud like Snowflake, the question isn't “how do I get data out of my legacy SaaS vendor.” It's why should ANY of my data be held hostage at all? Take a $20M legacy ITSM contract. Add the BPO spend wrapped around it — usually another $20 to $40M. The license was never the real ticket. It was the entry fee. Now strip the stack. UI replaced by your corporate GPT. Engine belongs on your data cloud. Data layer in your cloud. BPO layer collapsed by AI doing the routing. What's left? Maybe 30% of the work, priced as if the vendor still owned all four. That math doesn't survive a procurement review. Not in 2026. The vendors who sold you the bundle have eighteen months to convince you it still makes sense. They'll add an AI layer. They'll launch an agent. They'll tell you the data integration is “deeper than ever.” The bundle is still breaking—no stopping progress now. https://lnkd.in/gyUWQJwT
-
David Blonski posted thisOrchestration without replacement is useless. A familiar story: Careers made—and ended—on infrastructure that mostly admired itself. First, it was CDOs, building their beloved teck stack: Informatica, Alation, Monte Carlo. Millions spent. Beautiful dashboards. Perfect lineage. Business impact? Marginal. Result: CDOs having the highest turnover in the C-Suite. Now it's with CIOs in the hot seat. "Agent orchestration" by itself is the new MDM—expensive, impressive, useless. The new Alation is LangChain and LangGraph—frameworks to design elegant agent systems no business unit asked for. The new Informatica is the hyperscaler “dream” stack: Microsoft, Amazon, Google. Massive platforms. Massive services. Massive lock-in. None of it replaces a single legacy SaaS contract. None of it cancels a contractor’s SOW. None of it gives a CFO back a dollar. Orchestration only matters if it replaces something: - Replace ServiceNow ticketing workflows - Replace Coupa intake and approvals - Replace Salesforce lead routing and pipeline updates - Replace Workday HR requests and approvals Otherwise, orchestration isn't transformation. It's architecture as a hobby. At Elementum AI, orchestration is important, but as the means, not the end. The end is replacing legacy SaaS with AI-native capabilities (personalized concierge, headless UI, latest LLMs); 50-70% cost reduction; and new workflows launched in days not quarters. No system gatekeepers. No data silos. No hostage negotiations for your business logic. Not more software. Less. Forget the dream architecture. Cancel a contract. Your colleagues—and your career—will thank you for it. #EnterpriseAI #AI #DigitalTransformation
-
David Blonski reposted thisDavid Blonski reposted thisEvery AI change. On the record. Now that AI is core business, not just experimental—every change matters. Two new security primitives are getting elevated treatment in the Elementum AI platform. Activity logs for AI Providers and Agents are getting a big upgrade. Every change — provider swap, model update, agent instruction, new tool — is captured with the user, the field, the timestamp. Production gating for AI configuration. No silent fallbacks. Customers must explicitly choose the provider and model that runs each workflow. You can't govern what you can't see. And you can't trust what you can't trace. This is the audit trail your security and compliance teams have been asking for.
-
David Blonski posted thisBuffett Is Exiting AI Risk. CFOs Are About to Inherit It. Berkshire Hathaway, along with Chubb and Travelers, is excluding AI-related damages from corporate liability policies. That risk didn’t disappear. It moved onto your balance sheet. Every liability category starts the same way A real, recurring business risk exists. At first, it’s handled operationally. Then it becomes material. So auditors start asking questions. Audit committees are forced to get involved. Eventually, accounting catches up. That’s how we got: - ASC 460 (warranties) - ASC 450 (loss contingencies) - ASC 606 (sales returns) - ASC 326 (credit losses) Each became inevitable once losses were probable and estimable. AI is already crossing that threshold. AI errors are financial exposure --> Copyright violations from generated content (active lawsuits) --> Unauthorized contract terms --> Regulatory breaches in healthcare/finance --> High-volume operational errors (refunds, claims, approvals) Individually, these are edge cases, but at scale, they become loss distributions. Once losses are systematic, they are reservable. The math every CFO will do AI Liability ≈ Volume × Error Rate × Cost per Error × Degree of Autonomy More AI usage → higher error rate More autonomy → higher severity and variance Less control → less predictability, harder to bound exposure For Fortune 500 companies running AI deeply into regulated workflows, this is tens to hundreds of millions of dollars of reservable exposure. For CFOs, AI architecture is now a financial reporting decision - Where do you allow AI to act independently? - Where do you enforce rules? - Where do you require humans? These choices determine: - whether losses are estimable - how wide the risk band is - how large the reserve becomes Audit committees will ask questions. Auditors will mandate action. Most vendors are optimizing for the wrong thing ServiceNow, Workday, Salesforce, Microsoft, Anthropic, OpenAI all sell AI on usage, seats, and throuhput. They don’t optimize for risk. They offer no concept of liability containment. Instead, they increase: volume; autonomy; and opacity All three expand the reserve. Companies that win will use AI, but manage it aggressively > Rules where outcomes must be exact > AI where exposure is acceptable > Humans where judgment matters > Full auditability across every step This is the architecture we built Elementum AI around - and it's exactly why public companies and regulated industries are choosing Elementum to run their critical workflows and replace their legacy SaaS systems. Auditors don’t reserve against narratives. They reserve against what they can measure and defend. The insurance market just made AI accounting inevitable. AI is coming to your balance sheet—as a liability. Design for it now, or reserve for it later. #CFO #EnterpriseAI #AgenticAI #AIRisk #AIGovernance
-
David Blonski reposted thisDavid Blonski reposted thisHow do you ensure maintaining your AI powered business apps is not as tedious as maintaining legacy SaaS? The reality CIOs are facing: Building apps is getting easier. But keeping your AI apps current often mean days of tedious work — rebuilding every agent, every automation, one at a time. Not anymore. Elementum's bulk model migration moves your entire agent and automation estate onto the latest model in a single action. So when that new model comes out—or when you want to change model providers—you can do so easily and safely. Innovative — stay on the frontier. Fast — migrate in seconds, not in prompt editing sprints. Optimized — pick the best model by cost and accuracy Implement the latest models. Without the hours of work. Because the AI era should be easier than the legacy SaaS era.
-
David Blonski posted thisI don’t like outcome-based pricing. There, I said it. Sure, when Adobe’s President, Anil Chakravarthy, says "Tokens don't equate to value," I agree. That’s the easy part. In an ideal world, cost and value would perfectly align. But that’s not how it works in real life. Pretending otherwise is short-sighted and naive. Yet, Adobe, Sierra, Zendesk, Intercom, Salesforce are all moving in this direction: Charge when the agent "works." Charge per "agentic work unit." Charge per successful resolution. It sounds aligned. It isn't. - Who decides the "outcome?" The vendor? The customer? Think you’ll both agree? What counts as a successful resolution? The ticket the customer reopens twice? The 30% of cases where a human stepped in? If the vendor writes the contract, and the vendor audits the outcome, will the customer trust the results? If the outcome is a financial metric, will the customer open up their books and be transparent? - Back to fees with no ceiling. Token pricing at least had a token count. Outcome pricing has a vendor invoice. "We resolved 140,000 cases this quarter" - prove otherwise. One unpredictable meter swapped for another. Unless upside is capped, but then that’s not real outcome-based pricing, and no vendor wants to assume all the risk with no reward. - Budgeting is impossible. CFOs have to forecast operating costs. How do you model a line item where the vendor determines volume and "success" after the quarter closes? Variance analysis becomes archaeology. Margins blown because "the agent worked harder than expected." No public company CFO can sign up for that. Customers shouldn’t get punished for their own success. They grow - for many reasons - and the vendor gets to come along for the ride, taking an increasingly bigger cut of the “value” delivered. No doubt the vendor contributed (or their contract would be cancelled), but is that really fair? Seems more like a royalty than a partnership. The reason outcome-based pricing is being promoted is that AI vendors built products that burn tokens they can't defend. So now they're hunting for a new meter to justify the same bill. Elementum AI, we price differently. Fixed. Based on volume and complexity. No clawbacks. No outcome audits. No revenue share on your operations. We can price that way because our platform isn't incentivized to burn tokens in the first place. Hard rules when it matters. AI when it helps. Stacked in a priority order admins control. Rules are 10x cheaper than tokens — and that savings goes to customers, not our top line. ServiceNow, Workday, Salesforce, Microsoft, Adobe - every one of them is racing toward pricing models where their revenue scales with yours - whether they call it tokens, outcomes, or 'agentic work units.’ We built toward the opposite. "Fair" shouldn't be a differentiator. But if it is, we’ll side with our customers every time. #EnterpriseAI #AgenticAI #AIAgents #CFO
-
David Blonski posted this"𝘔𝘺 𝘸𝘪𝘧𝘦 𝘭𝘪𝘬𝘦𝘴 𝘥𝘰𝘨𝘴. 𝘞𝘩𝘢𝘵 𝘴𝘩𝘰𝘶𝘭𝘥 𝘐 𝘣𝘶𝘺 𝘩𝘦𝘳?" Sorry, Agentforce can’t help you. That's according to Pandora's CTO. Salesforce's flagship AI agent - the company with all your sales data - "hasn't been able to reliably recommend products on its own." Meanwhile, Marc Benioff's telling the WSJ "People think we have our back against the wall when in fact the opportunity has never been greater." The numbers tell otherwise. Just 15% adoption after eighteen months. 10% revenue growth, down from 35% a decade ago. Investors see it too. Salesforce’s stock down 28% year to date. Chicago Capital, a major shareholder, telling the WSJ "they need to show revolutionary jumps" — right before selling their shares. Agentforce’s best claim to fame, a PenFed case study using agents for password resets. A neat trick in 2024, but this is 2026. Anthropic just killed Tableau with live artifacts, and Salesforce really believes this isn’t a moment of truth? Of course not. Which is why Salesforce released Headless 360 - it’s new “agent-first architecture.” Salesforce cofounder Parker Harris asks triumphantly: "Why should you ever log into Salesforce again?" I don’t know, Parker, why would you? We all know the signs of someone going through a midlife crisis: get a makeover, buy a motorcycle. New look, same problems. Headless 360 is what a 27-year-old platform looks like going through a midlife crisis. New look, same problems. The pitch: agents "inherit" decades of Salesforce business logic. Translation: we’ve still got your data on lock-down, trapped within various data silos. The new euphemism to describe custom configurations required for every Agentforce deployment: "scaffolding." It's sysadmin in a hoodie. They invented a new metric - Agentic Work Units - because ARR and seat counts no longer tell the story they need. When financial KPIs stop working, make up new ones. Rings of WeWork’s “community adjusted-EBITDA.” They even relaunched Einstein under a new name — Agent Albert — because the first 12 years didn't land. As if the name was the problem. You can’t blame Salesforce. They’re running the same playbook every incumbent runs at the end: push distribution to hide the lack of innovation. Make noise. Invent metrics. Rebrand failures. Hope the renewal gets signed before the customer notices. Elementum AI was built from the ground-up with AI infused in its DNA. LLM-first architecture, agentic deployments, and Day 1 go-lives. We also made the explicit business decision to never hold our customers hostage: never persisting customer data, never hiding business logic behind “scaffolding,” never charging for tokens or seats or credits to misalign incentives. Now, public companies are cutting over from ServiceNow, Salesforce, Coupa, and Workday. One agentic front door. One platform to manage agents. As for Salesforce, you know what they say about putting lipstick on a pig. #EnterpriseAI #DigitalTransformation #Innovation
-
David Blonski reposted thisDavid Blonski reposted thisEvery LLM call that a simple rule could have handled is money burned, and a chance that it’s routed incorrectly. Elementum AI Agent Gateways route every conversation to the right agent, deliberately: → Deterministic rules for the logic you already trust → Probabilistic rules for everything fuzzy …stacked in a priority order admins control. Hard rules when it matters. AI when it helps. Agent Orchestration, done right.
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
-
-
-
-
-
-
Volunteer Experience
-
Board of Directors - Alumni Association
Rochester Institute of Technology
- Present 5 years 5 months
Patents
-
SELECTIVELY GRANTING COMPUTER SYSTEM ACCESS CREDENTIALS TO EXTERNAL USERS AND NON-USERS
Issued US 12,204,667
Multi-party, role-based incident management in a single SaaS platform with the ability to assign tasks to third-parties while sharing only relevant, permission-based information to the specific role(s) within the assigned parties.
-
SYSTEMS AND METHODS FOR A DATA-DRIVEN WORKFLOW PLATFORM
Issued US 12,158,894
See patentThe ability to automatically initiate (i.e. trigger) a workflow from within a cloud-based data warehouse without requiring data to be moved, transferred, or copied outside from the warehouse to a third-party application. Also referred to as cloud-native automation.
Languages
-
Spanish
Limited working proficiency
Recommendations received
1 person has recommended David
Join now to viewOther similar profiles
Explore more posts
-
Paul Perrett
Big milestone for Firmable. We’ve raised $14m Series A led by Airtree. Sales has moved through a few big waves: intuition-led, CRM-led, data-led. We’re now entering the next one – intelligence-led sales. The opportunity isn’t just better data. It’s turning that data into clear direction and action, without adding more work for sales teams. That’s what we’re building at Firmable: a foundation of trusted external data, layered with intelligence that helps sellers know who to focus on and when. Led by Airtree, this round supports our expansion across Asia and into the US – and accelerates the build-out of AI agents that take the admin work off sales teams so they can focus on what they do best. Proud of the team, grateful to our customers and investors. We’re just getting started. Read the exclusive in the AFR. https://lnkd.in/gr66uknb Leigh Jasper | Tara Salmon | Karthik Venkatasubramanian| Chester Thompson| Chath Widanapathirana
20 Comments -
Jeff Perry
Seth Levine nailed it! The concentration problem isn't just a GP pain point... it's choking innovation. When capital pools around the same 20 funds, emerging managers and founders get shut out. Love that he's using his platform to call this out. Foundry has backed 50+ emerging managers. That's the diversification the ecosystem needs. This is exactly why Carta exists — making capital allocation visible and accessible. Capital Evolution hits at exactly the right moment. Thanks for having us Daniel Dart. Team Carta loves the community of FUTURE TITANS you have built 🚀
6 Comments -
Peter OBrien
Every hire, product feature, and GTM experiment is a capital allocation decision, most teams just don’t call it that. I’m kicking off a 3-part series on capital allocation for founders, operators, and finance leaders with a simple question: ➡️ Will the next $1 you deploy become worth more than $1? Series overview Part 1: defines the core concept + the math Part 2: where capital goes + how to measure whether it’s working Part 3: AIMS framework for communicating allocation decisions to management teams and your board Part 1 (attached here) lays the practical foundation: 🔹 The $1 invested test and ROIIC vs. hurdle rates (scoreboard vs. decision lens) 🔹 A lightweight “investment brief” to evaluate bets (GROW / BUILD / BUY) 🔹 The small metric toolkit that translates finance into operating decisions (NPV / payback / ROIIC / incremental margin) 🔹 Early-stage proxies when DCF inputs are unknowable (burn multiple + revenue quality) 🔹 Operator “vital signs” to spot drift early (profit engine, leverage, cash conversion, optics) Series Roadmap ✔️ This series (Parts 1–3): breadth-first. shared, lightweight framework to define value creation, choose decision-grade metrics, and communicate tradeoffs clearly ✔️ Next series: depth-first. momentum drivers + operational decisions that need near real-time measurement to spot drift early and course-correct fast Read Part 1 here and Part 2 and 3 on substack. Next Series will be out next week. 💰 What’s one bet you’re funding right now, and how will you prove it’s working?
-
Aviel Ginzburg
While there has never been a more exciting time to be a founder building dev tooling or next-gen infra, it has also never been less investable at seed/pre-seed. I'm either really missing something or a lot of my peers are lost. As someone who has not just written, but also SHIPPED, about 75k lines of code in the past 6 months I can tell you that the evolution of how to build products has changed as much in the past year as it did in the entirety of 2007-2017. The complete rise and fail of frameworks, platforms, methodologies, etc... paved over and forgotten... that is of course except for the 1 company that gets a 1000x return from a wildly overvalued hyper-scaler or drunken growth stage investor obsessed with compounding at scale. Imagine a world where any seed investor in trends like Openstack, Hadoop, PaaS, etc all took a full loss on their investment. That's what we're looking at right now. I personally know of over a dozen well-funded seed-stage companies building in these spaces, with years of runway, scrambling to get acquired for a return of capital + several million personally while they're still relevant. If you're not seeing this unfold in front of you, you either aren't paying attention or you're satisfied playing the lottery instead of investing.
11 Comments -
Lakshmi Shankar
Thrilled to announce that Together Fund is investing in Sentra, alongside a16z speedrun! You track results in Jira. Decisions in Notion. Conversations in Slack. But the reasoning, the debates, trade-offs, and context behind why you chose A over B, disappears into what we call "Dark Matter." A decision made in March looks insane by July because no one remembers the constraints that made it smart. I lived this firsthand at Twitter scaling from 800 to 8,000 employees, and at Google while launching AI Overviews to billions at planet scale. The problem isn't process. Process is compensation for something deeper: organizational amnesia. An organization’s "Systems of Record" doesn’t solve this, they encode it. They store what happened, never why. That's why we are investing in Sentra. Sentra is the always-on collective memory that eliminates organizational amnesia by maintaining accurate context for all members and agents, functioning as an operational nervous system. It connects to every channel where work happens, meetings, Slack, email, code commits, docs, calendars, and treats them not as artifacts to search, but as living signals to synthesize. The fleeting and the permanent, unified into a memory that understands. The founding team is built for this: - Jae Gwan Park (CEO): Product-first founder, memory systems research at UofT and MIT - Ashwin Gopinath (CSO): Former MIT professor, created "Reflexion" (NeurIPS 2023), agents that learn from mistakes, 2x founder - Andrey Starenky (CTO): Early Vapi engineer, ex-IBM, built to process enterprise-scale data firehose Together is an operator-led fund. We invest in problems we've lived. This is one of them. Many congrats Jae, Ashwin and Andrey, we are so excited to partner with you! Read the full thesis: https://lnkd.in/gixj9cE4 Book a demo: https://www.sentra.app/ #OrganizationalMemory #AI #Sentra #TogetherFund #a16z #ContextGraphs
3 Comments -
Matt Rappaport
Don't Build a Better Wheat Farm" - Why Defensibility Stakes Are Higher in Deep Tech Just published a new piece on my "Ignore the Confusion" blog, building on thoughtful insights from Eric Ver Ploeg at Tunitas Ventures about startup defensibility. Eric's core thesis: Too many startups pitch like wheat farmers - "huge TAM, slow incumbents, growing market, domain expertise" - but fail to think through long-term defensibility until it's too late. From a deep tech perspective, the stakes are even higher: ** Unlike software, deep tech founders must commit to defensibility strategies from day one - their funding depends on it ** Patent vs. trade secret decisions are often difficult to reverse and shape your entire competitive strategy ** Even "picks and shovels" providers (the tools that make industries more efficient) become commodities without proper moats The key insight that resonates: Defensibility can't be retrofitted. Whether you're building software or deep tech, your moat must be architected into the business model from the start. Thanks to Eric Ver Ploeg for sharing these insights on startup strategy and letting me build on his framework from a deep tech lens. Read the full post: https://lnkd.in/dEj_iF-Q #DeepTech #StartupStrategy #Defensibility #VentureCapital #Innovation
2 Comments -
Amber Illig
🎙️ From Square to Mercury: How Rohini Pandhi became one of fintech’s top product leaders—and how velocity of learning at high slope companies shaped everything. Rohini dropped so many gems in our first episode of First Builders: 💎 Generalist → Specialist: In her early career, she was a generalist and tried everything. This led her to critical discoveries. She specialized in product and fintech when the “time was right” 💎 Follow the Engineers and Designers: Square was a high slope environment for Rohini, packed with talent density and learnings. She finds high slope environments by studying where smart engineers and designers are going. 💎 Fake News about PMs: PMs don’t just “move fast and break things.” Her team runs 30-50 customer interviews per quarter, providing rigor behind every decision. 💎 It’s Usually Too Early to Hire a PM: A top question she gets from founders is whether they should hire a PM. She usually says: “it’s too early.” 💎 Hiring a World Class Team is Like Tennis: When finding a tennis partner, you want someone a little better than you who can challenge you to improve your game. Give us a listen and leave a review on your favorite podcast location (see comments for links)! #FirstBuilders #StartupPodcast #ProductLeadership #TechCareers #Leadership
4 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content