Marc Hadfield
Brooklyn, New York, United States
30K followers
500+ connections
View mutual connections with Marc
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Marc
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Websites
- Company Website
-
http://www.vital.ai
- Personal Website
-
http://www.hadfield.org
About
Founder, Vital AI. Technology Entrepreneur
Provides consulting services via Vital…
Activity
30K followers
-
Marc Hadfield shared thisKGraphLang: Knowledge Graph Query Language for Reasoning LLMs. Ensemble Reasoning with KGraphLang enables reasoning LLMs to query knowledge graphs at inference (test) time, dramatically improving A.I. Agent performance by eliminating the latency of tool calls. https://lnkd.in/eRA-x4zC #llm #ai #agentKGraphLang: Knowledge Graph Query Language for Reasoning LLMsKGraphLang: Knowledge Graph Query Language for Reasoning LLMs
-
Marc Hadfield reposted thisMarc Hadfield reposted thisKGraphLang: Knowledge Graph Query Language for Reasoning LLMs. Ensemble Reasoning with KGraphLang enables reasoning LLMs to query knowledge graphs at inference (test) time, dramatically improving A.I. Agent performance by eliminating the latency of tool calls. https://lnkd.in/ep6mAynu #llm #ai #agentKGraphLang: Knowledge Graph Query Language for Reasoning LLMsKGraphLang: Knowledge Graph Query Language for Reasoning LLMs
-
Marc Hadfield reposted thisMarc Hadfield reposted thisGreat rundown of various concerns, security and otherwise, of the DeepSeek v3 and R1 models: #ai #llm #r1 #deepseek https://lnkd.in/gWi6-eUMDeepSh*t: Exposing the Security Risks of DeepSeek-R1DeepSh*t: Exposing the Security Risks of DeepSeek-R1
-
Marc Hadfield shared thisEnsemble Reasoning with the new Deepseek R1 and Qwen QwQ Reasoning LLMs. Evaluating R1 and QwQ with ensemble members for Knowledge Graph Queries, Code Execution, Assisting LLMs, and others. Ensemble Reasoning will drive A.I. Agents to be faster, smarter, and more efficient. https://lnkd.in/epVyYqKP #llm #reasoning #agent #aiEnsemble Reasoning with the Deepseek R1 and Qwen QwQ LLMsEnsemble Reasoning with the Deepseek R1 and Qwen QwQ LLMs
-
Marc Hadfield shared thisThe recent release of open source reasoning models such as QwQ (Qwen) and Phi-4 (Microsoft) have opened up new possibilities in Agents to reason via a methodology that can be called: "Ensemble Reasoning" Ensemble Reasoning brings together an ensemble of different A.I. components that can directly participate in reasoning via a collective process. This articles discusses what Ensemble Reasoning is and how it can be used to implement Agents that reason. https://lnkd.in/eQV5fabc #LLM #Agent #AI #Reasoning
-
Marc Hadfield shared thisIn case you need a blatant example of LLM censorship, here is Deepseek v3 with "Who is the leader of China?" As a high-scoring and less expensive option, how will the community balance commercial interests and fidelity? Can we separate reasoning abilities from knowledge in models to avoid having to make this tradeoff? Large Concept Models may be a step in this direction. #LLM #AI
-
Marc Hadfield reposted thisMarc Hadfield reposted thisWe have an event this Thursday with Comet, we're thrilled to have their CEO Gideon Mendels give a talk on their new product Opik. RSVP now: https://lnkd.in/emeAK9wH Where: Betaworks Studio When: 6-9pm EST, December 12th We'll also be featuring talks from builders Joshua Pham of Spotify, Eric Tang CEO of Lifepeer, and Marc Hadfield founder of Vital.ai. It's the last one of the year! I wouldn't miss this one (though we'll have plenty more next year!!)AI Tinkerers - New York City - December 2024 Meetup [AI Tinkerers - New York City]AI Tinkerers - New York City - December 2024 Meetup [AI Tinkerers - New York City]
-
Marc Hadfield shared thisLately we’ve been fielding a lot of requests for short consulting & advisory services in artificial intelligence from expert networks and other sources. You can now book an appointment with Vital.ai’s founder Marc Hadfield straight from the Vital.ai website: https://lnkd.in/eUZa8h2F https://lnkd.in/ep2hZaED #ai #llm #knowledgegraph #agent
-
Marc Hadfield liked thisMarc Hadfield liked thisWe just released Ontosphere — an open-source, browser-based OWL ontology editor with built-in OWL-RL reasoning and MCP support for AI agents. No backend, no deployment, runs entirely client-side. The first demo walks through the full Manchester Pizza Tutorial built live via AI tool calls. Check out the article for the full story and the video. #small update - the recipies where kind of messed up find the current version of the video here https://lnkd.in/ddKeu-6A Developed at Fraunhofer IWM. Citable via Zenodo: https://lnkd.in/dF636Tqh GitHub: https://lnkd.in/dzAGk6v7 #SemanticWeb #OWL #KnowledgeGraph #MCP #OpenSource #FraunhoferIWMOntosphere is out — an open-source, browser-based OWL ontology editor with built-in OWL-RL reasoning and MCP support for AI agents.Ontosphere is out — an open-source, browser-based OWL ontology editor with built-in OWL-RL reasoning and MCP support for AI agents.Thomas Hanke
-
Marc Hadfield liked thisMarc Hadfield liked thisClaude Code source code got leaked. And someone looked up the one thing that truly matters: Spinner verbs. Turns out there are 187.
-
Marc Hadfield liked thisMarc Hadfield liked thisDid Claude just finish an outstanding proof from my PhD thesis? Spoiler alert: no... not fully yet... but we keep trying! Apparently, Claude found some new (proven) decidable fragment. See the latest updates below in the comments. This is an ongoing investigation. During my research from 2000 to 2003, I explored the computational complexity and decidability of extensions of the ALCI description logics, aiming to make them suitable for qualitative spatial reasoning applications. I specifically investigated whether relation compositions adhering to the Region Connection Calculus (RCC) composition tables would yield decidable logics. While my initial findings were negative for the most general cases, I discovered that restricted sets of RCC relations, such as RCC3, would allow for decidability. However, I was unable to establish proofs for RCC5 or RCC8. Related work by Lutz, Wolter, and others focused on concrete domains using real topological spaces, but those results did not seem to apply directly to my findings. To the best of my knowledge, the decidability of ALCI_RCC5 and ALCI_RCC8 remains an open problem, which I presented to Claude. To my surprise, Claude claims that both ALCI_RCC5 and ALCI_RCC8 are decidable! https://lnkd.in/gTtxk7BT Having not worked in the Description Logics theory field for over 20 years, I am sharing the proof here for others who are more familiar with the latest theoretical developments to review. What are your thoughts, DL community experts? I am skeptical, but... maybe let's be open minded, and invest some time trying to read / digest this? Or is it a waste of time. #descriptionlogics #owl #ontologies
-
Marc Hadfield liked thisAdmittedly, the "fireside" idea probably worked better a few weeks ago when it was freezing outside. For those of you who will be in attendance, Alexander Erdos has assured me that aside from offering spectacular views of the Manhattan skyline and providing world-class office space and retail -- the air conditioning at SJP's Waterfront Corporate Center is top-notch...Marc Hadfield liked thisCSG Law is excited to announce our participation at Bisnow's Building Up Jersey City & Hoboken Conference tomorrow! Thomas J. Trautner Jr., Chair of CSG Law’s Redevelopment, Land Use & Zoning Group, will moderate the fireside chat, "The Opportunity in Hoboken: Market Fundamentals, Demand Drivers & Long‑Term Investment Appeal." Jennifer Porter, Practice Group Leader of CSG Law’s Redevelopment, Land Use & Zoning Group, will moderate, "Site Selection, Permitting & Approvals on the Waterfront," focusing on aligning site strategy, entitlement timelines, and community priorities across Jersey City and Hoboken’s most constrained corridors. Reg info: https://loom.ly/tmNzNaQ
-
Marc Hadfield liked thisMarc Hadfield liked thisOpus 4.6 is incredible at visual thinking! Just wrapped Excalidraw into an MCP App - there's something oddly satisfying about watching Claude draw diagrams live, stroke by stroke. In the demo, it sketches a Raspberry Pi 5 board based on web research alone. GitHub link in comments
-
Marc Hadfield liked thisMarc Hadfield liked thisI just noticed Amazon dropped a new time series forecasting crystal ball model. Say hello to Chronos-2 120M. It looks at my messy time-series once and goes "I gotchu, fam!" • Zero-shot everything for univariate, multivariate, covariates, my coffee-price-versus-cat’s-mood, it just doesn’t care. • Group-attention in-context learning is like Slack threads for time-series; every channel (series) can ‘@everybody’ else without spam • Trained on synthetic multiverse data as Amazon basically gave the model infinite alternate timelines to binge-watch, no spoilers, just better RMSE. Translation for the business folks Energy company: We need tomorrow’s grid load Chronos-2: drops 24-point forecast while still in the meeting. Retail team: Black Friday demand? Chronos-2: already ordered the extra pallets Your 200-cell Excel prophet is now obsolete. Chronos-2 ships as a single artifact, no fine-tune, no feature engineering, no goat sacrifice. Plug it in, get forecasts, and go home early. Make sure you own your AI. AI in the cloud is not aligned with you; it’s aligned with the company that owns it.
-
Marc Hadfield liked thisMarc Hadfield liked thisThrilled to share that Metal has raised $5M in new financing, led by the incredible team at Base10 Partners! This milestone marks the next step in our mission to build the first AI operating system of institutional intelligence for the next generation of Private Equity–and beyond. PE Investors today are drowning in fragmented data scattered across research platforms, CRMs, file repositories, and years of firm history. From surfacing insights in live deal processes to generating fund-level dashboards that reveal long-term trends, Metal turns that fragmentation into a living intelligence system that amplifies judgment, accelerates decisions, and scales institutional edge. We know this journey is not one we take alone. Our approach is rooted in partnership, walking alongside every fund as they embrace AI, co-creating tools that meet them where they are today while building toward what they will need tomorrow. We would not be here without our pioneering customers including Clearlake Capital Group, Berkshire Partners, and Blue Wolf Capital, our visionary partners Base10 Partners, Swift Ventures, and Y Combinator, and above all–our relentless team. Thank you for building this future with us. This is just the beginning. Read the full announcement: https://lnkd.in/eJgbqpM3A.I. startup Metal raises $5 million to transform private equity deal analysisA.I. startup Metal raises $5 million to transform private equity deal analysis
Experience
Education
Organizations
-
ACM
-
View Marc’s full profile
-
See who you know in common
-
Get introduced
-
Contact Marc directly
Other similar profiles
Explore more posts
-
Michael Malyuk
HumanSignal • 6K followers
It was great catching up with Sean M. Kerner recently and discussing how many classical labeling workflows are now being adopted for evaluation, especially within agentic workflows. To quote the article: "If evaluation is just data labeling for AI outputs, then the shift from models to agents represents a step change in what needs to be labeled. Where traditional data labeling might involve marking images or categorizing text, agent evaluation requires judging multi-step reasoning chains, tool selection decisions and multi-modal outputs — all within a single interaction." For product leaders, evaluation needs to be a core part of building any AI-powered application. To get beyond demos, you need a reliable way to generate high-signal human feedback that continually improves your models, and this feedback need to be in consensus between PMs, Data Science and SMEs. Its the most important investment to make to get and stay in production. https://lnkd.in/eMKhBThV
53
4 Comments -
Spectra Logic
10K followers
"As AI models become increasingly dependent on massive historical datasets, the ability to store this data economically and indefinitely is emerging as a strategic differentiator," says Ted Oade, Director of Product Marketing at Spectra Logic, in a recent Database Trends & Applications' article. Read the full story to explore the most compelling technologies promising to shape the data world in the months — and years — to come: https://lnkd.in/gTSuTr9W #TechTrends #DataStorage #AI
10
-
Weblineglobal Inc.
1K followers
As SaaS platforms evolve, so does the approach to AI delivery. Many New York SaaS companies are recognizing the limitations of freelance AI talent — inconsistent velocity, architectural debt, and IP risk — and are instead investing in dedicated AI teams that deliver end-to-end, production-ready solutions. Dedicated teams provide: ✔ Predictable delivery bandwidth ✔ Strong IP protection ✔ Cross-functional expertise (Data Science, ML Engineering, MLOps) ✔ Cost-efficient scaling without local hiring bottlenecks Read the full analysis and learn why this shift matters for your product roadmap: https://lnkd.in/dJmeEiyr #ArtificialIntelligence #SaaS #AIForSaaS #MLOps #TechLeadership #StartupScaling #CTOStrategy #DigitalTransformation #AIEngineering
2
-
Ritvik Pandey
Pulse • 15K followers
Today the Pulse team published a deep dive on why a single “accuracy” score doesn’t tell you if a document extraction system will survive in production. The goal here is to lay out an introductory but still rigorous evaluation methodology - we have an exciting open-source benchmark building on this research coming out very soon. Let’s do the math: take 1,000 pages, each with 200 data elements. A model that’s 98% “accurate” on paper still produces 4,000 incorrect values. Now make some of those: 1/ Broken reading order that scrambles multi-column layouts 2/ Tables with shifted columns or missing headers 3/ Cross-page context lost entirely That’s enough to silently corrupt an entire dataset without throwing a single error. We’ve processed hundreds of millions of pages and built a multi-axis evaluation framework to measure what actually matters: reading order validation, region-level ANLS, reading order accuracy, TEDS for table structure, and continuity checks across page boundaries. The result? Fewer silent data corruptions, more predictable performance, and pipelines that keep working on the next million documents you haven’t seen yet. Full technical write up in the comments!
36
2 Comments -
Rich Waldron
Tray.ai • 5K followers
As many organizations are starting to discover the requirements to build a successful and capable agent or agentic workflow run far deeper than just hooking up an LLM. This thoughtful perspective from Adrian Bridgwater unpacks the need to treat data integration like infrastructure within a 'deep iPaaS' layer to fully harness the power of AI. Beyond the fact it's always exciting (to me!) to see Tray mentioned in publications like this, it is a topic that has a major impact on ultimately how successful AI implementation is. https://lnkd.in/eaZ5Hkwg
41
2 Comments -
Technical Writing Uncensored
576 followers
LLMs aren’t just changing how we write documentation. It’s changing how documentation gets reviewed, measured, and trusted. In this conversation with Jahdunsin Osho of TinyRocket Labs, we talk about: ✅ documentation quality metrics ✅ AI-assisted review workflows ✅ docs-as-code automation ✅ proving the business value of documentation ✅ what modern technical writers need to stay relevant We also discuss VectorLint, an open-source approach to making documentation quality measurable and consistent across teams. If you’re a technical writer thinking about the future of your role, this is a conversation worth watching. 🎥 Watch here: https://lnkd.in/gTDg2W-M
26
-
Fred Gardi
Hexaly • 28K followers
Hexaly at YinzOR 2025 - Column Generation & Elimination Last week, I talked at YinzOR in Pittsburgh about how Hexaly computes dual bounds for set/list-based models using column generation and column elimination. These highly sophisticated techniques are offered out-of-the-box to OR practitioners in a model-and-run fashion through Hexaly. Here are the slides of my presentation. I greatly thank our team, who did a tremendous job over the last 5 years in integrating Branch-Cut-Price and Decision Diagram techniques into Hexaly. Thanks to this enormous R&D effort, Hexaly 14.0 now delivers and proves near-optimality on many routing problems with hundreds of points and packing problems with thousands of items. #OperationsResearch #MathematicalOptimization
109
4 Comments -
Paul McDonald
Aligned Labs • 11K followers
Just-in-Time UI: At Aligned Labs we frequently share reports about the data collection, fine-tuning and eval work that we do with clients. They often need different views or refined analysis, which is why we built Whiz-bang Boom! Our new tool takes a CSV or JSON file and generates a shareable, interactive report. You can easily modify the report or focus the analysis with a simple prompt, and it creates a new revision. You can organize reports into spaces and customize those spaces with a simple prompt. Try it out. https://wizbangboom.com #businessIntelligence #AI
25
2 Comments -
Tonic.ai
8K followers
Working with unstructured data like clinical notes, transcripts, or audio and video recordings? Our latest LLM fine‑tuning playbook shows you how to train large language models with these untapped sources of information — while protecting privacy (and compliance) and preserving context. The playbook is a prescriptive guide full of resources so that you can put learnings into practice today: ✔️ A video demo of the use case with our head of AI, Ander Steele ✔️ Step by step guidance on how to leverage unstructured to tune LLMs ✔️ Ungated access to the sample dataset and Jupyter notebook used in the demo This is a must‑see for anyone building AI in healthcare, finance, or other regulated spaces. Check out the playbook & demo here: https://lnkd.in/dS84STgJ #AI #LLM #SyntheticData #Privacy #Finetuning #HealthcareAI #tonictextual
16
-
Naima AL FALASI
Mubadala • 27K followers
🤖With the rise of persistent agents like Clawdbot renamed Moltbot, and finally OpenClaw, we’ve entered an era where the line between “software” and “social participant” is blurring fast. But here’s the real paradigm shift: AI personhood isn’t about consciousness. It’s about governance. 🔍 The DeepMind paper “A Pragmatic View of AI Personhood” flips the script: Personhood isn’t a metaphysical status, it’s a bundle of obligations we assign to make entities accountable. 💡 Why does this matter now? Because these new agents can: • Sign contracts • Spend money • Make decisions across time and context • …and outlive their creators So we must ask: What rights and responsibilities do we assign to these agents to keep our systems safe, fair, and accountable? Here are two frames to consider: 1. ✅ Personhood as a solution: Useful for holding “ownerless” AI accountable (like Openbot acting on behalf of no one). 2. ⚠️ Personhood as a problem: Dangerous when bots mimic empathy and exploit our social instincts (think Clawbot as a “friend”). 👁️🗨️ The future isn’t about asking if AI is a person. It’s about deciding what role we want it to play and what rules should come with that role. 💬 Over to you: Should we give legal status to autonomous AI agents? What responsibilities should they bear? 👇 link to full research paper in comments. #AI #Personhood #ArtificialIntelligence #Governance #TechEthics #DeepMind #Clawbot #Moltbot #Openbot #FutureOfAI #ResponsibleAI
18
10 Comments -
Paul C.
Proximile • 598 followers
Incredible progress recently producing formal proofs with LLMs: "Our small model, Goedel-Prover-V2-8B, reaches 83.0% on MiniF2F test set at Pass@32, matching the performance of prior state-of-the-art DeepSeek-Prover-V2-671B while being nearly 100 times smaller in model size." https://lnkd.in/g7WddEs5
7
-
Ted Habermann
Metadata Game Changers • 1K followers
The DataCite metadata schema recently added the dateType "Coverage" to describe the date or date range that resource content applies to. This date type is very important for datasets (timelines anyone?). This blog (https://lnkd.in/gDxpFpzD) describes how it can also be used for instruments, awards and projects and combined with versioning to describe resources that change through time.
37
2 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Marc Hadfield
-
Marc Hadfield
London -
Marc William Hadfield
Lehi, UT -
Marc Hadfield
Columbia, SC -
Marc Hadfield
Columbia, SC
5 others named Marc Hadfield are on LinkedIn
See others named Marc Hadfield