Videos

Popper Character Design | The Philosophical Agent for AI Validation |Open Source Framework by Humanitarians AI" https://youtu.be/v0CVQJ1T7Iw Meet Popper, our 1940s-inspired character embodying computational skepticism in AI! Named after philosopher Karl Popper, this character represents our open-source framework for systematic AI validation through evidence-based assessment. The Popper Framework applies scientific skepticism to artificial intelligence, embracing Popper's principle that theories gain strength by surviving attempts at falsification. Through specialized validation agents orchestrated by a central verification layer, we're helping developers discover what approaches to AI validation actually work in practice. Our character's vintage aesthetic reflects how we're bringing time-tested philosophical principles to the cutting edge of AI development. His penetrating gaze and determined expression symbolize unwavering commitment to critical thinking in an age of increasingly powerful AI systems. Get involved with Popper: GitHub: https://github.com/Humanitariansai/Popper Project Website: https://www.humanitarians.ai/popper Connect with Humanitarians AI: Website: https://www.humanitarians.ai/ YouTube: https://www.youtube.com/@humanitariansai LinkedIn: https://www.linkedin.com/company/105696953/ Music on Apple: https://music.apple.com/us/artist/humanitarians-ai/1781414009 Spotify: https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV Exact Prompts Used for Character Design: "Create a 1940s-era intellectual character inspired by philosopher Karl Popper, with a penetrating gaze and determined expression that reflects unwavering commitment to critical thinking and scientific skepticism." #AIValidation #OpenSourceAI #KarlPopper #ComputationalSkepticism #AIEthics #ResponsibleAI #HumanitariansAI #AIAgents #AIOrchestration #PhilosophyOfScience #AIFramework

Meet Dayhoff: The 1950s Scientist Character Behind Our Biological AI Framework | Humanitarians AI https://youtu.be/ASfQlTZN2zE Introducing the Dayhoff character - the visual embodiment of our agentic computational bioinformatics framework at Humanitarians AI! Inspired by pioneering scientist Margaret Bell Dayhoff, we've designed a 1950s era researcher with a brilliant smile and determined expression that reflects her unwavering commitment to scientific innovation. The Dayhoff framework is our open-source platform for biological intelligence that orchestrates specialized AI agents to tackle complex challenges in life sciences. Just as Margaret Dayhoff revolutionized biology with her pioneering work in protein sequence analysis and her development of the one-letter amino acid code, our Dayhoff framework coordinates intelligent agents that work in concert to solve critical biological problems. Currently supporting groundbreaking projects like PredictaBio, which creates recipes for novel proteins with specific properties, and the Raman Effect project, which uses AI to analyze wastewater spectroscopy data for disease surveillance. Join our community to help develop these agentic systems that are already revolutionizing protein synthesis and public health monitoring. The Dayhoff framework helps scientists discover what approaches to biological agent computing actually work in practice. Learn more about Dayhoff: https://www.humanitarians.ai/dayhoff Follow us: Apple Music: https://music.apple.com/us/artist/humanitarians-ai/1781414009 Spotify: https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV Website: https://www.humanitarians.ai/ YouTube: https://www.youtube.com/@humanitariansai LinkedIn: https://www.linkedin.com/company/105696953/ Please like and subscribe to Humanitarian's AI YouTube to follow our progress and learn more about our transformative work in computational biology! #HumanitariansAI #DayhoffFramework #ComputationalBiology #AIinScience #BioInformatics #OpenSource #AIagents #MargaretDayhoff #BiologicalIntelligence #ProteinSynthesis #PublicHealth #AIforGood

Addams — Your Humanitarians AI SherpaGenerative AI broke the artifact as proof of learning. Addams documents what the artifact can't: the friction, the judgment calls, the moments you had to supply what the machine could not. Addams is the weekly learning documentation system for OPT volunteers at Humanitarians AI — built to capture the cognitive work that a polished report no longer proves happened. Named after Jane Addams, who understood that good work without documentation disappears, this tool asks specifically for what most documentation systems ignore: where you got stuck, what the AI got wrong, what you had to reformulate from scratch, and what you're still figuring out. Every week, you file a /hai report. Addams walks you through a structured intake — objectives, work sessions, hours, artifacts, Substack publication, and the irreducibly human moments that are the actual evidence of learning. It enforces gates, flags thin answers, and produces a record structured for three readers: you, your program coordinator, and Humanitarians AI management. The artifact can be produced without the learning. Addams documents the learning. In this video: Why AI permanently decoupled artifact production from the learning it used to require What the actual proof of work looks like in the AI era How Addams captures it — and why it enforces the friction section How to set up your Claude Project and file your first /hai report this week Get started: → Addams: https://www.humanitarians.ai/tools/Addams #HumanitariansAI #Addams #LearningDocumentation #OPTFellows #AIEra #IrreduciblyHuman #EdTech TAGS: Addams learning documentation, Humanitarians AI OPT, learning record AI era, OPT volunteer documentation, AI and proof of learning, irreducibly human, weekly documentation system, Claude project setup, learning engineering, AI decoupled from learning, OPT fellows program, documentation for international graduates, Humanitarians AI 501c3, what AI cannot do, learning friction documentation

Dewey: Humanitarians AI's Educational AI Framework https://youtu.be/egZRNybs0Yo The Dewey Project: Our design for the Dewey character is the embodiment of our agentic educational AI framework. Inspired by progressive educational philosopher John Dewey, we've created a turn-of-the-century professor with an inquisitive gaze and thoughtful demeanor that reflects his unwavering commitment to educational innovation. The Dewey framework is our open-source platform for educational intelligence that orchestrates specialized AI agents to transform how we learn and teach. Just as John Dewey revolutionized education with his experiential learning philosophy, our Dewey framework coordinates intelligent agents that work in concert to create personalized, interactive educational experiences. Currently supporting projects across education, including: The Medhavi projects: Intelligent AI Books that adapt to individual learners, solving the "chicken and egg" problem by guiding conversations through dynamic Socratic dialogue Ada: A calculus learning assistant providing step-by-step guidance through complex mathematical concepts Newton: A physics chatbot incorporating dynamic simulations to visualize scientific principles Grace: An algorithms chatbot that simulates computational processes with interactive visualizations Botspeak: Teaching effective communication with AI through practical tutorials Case Crackers: Enhancing management education with customized case studies and simulations The Dewey framework demonstrates how intelligent agents can transform education. Our orchestration layer ensures these specialized agents collaborate effectively, just as Dewey himself integrated multiple educational approaches. By joining our community, you'll help develop these agentic systems that are already enhancing personalized learning and democratizing access to quality education. Learn more: https://www.humanitarians.ai/dewey Follow Humanitarians AI: Website: https://www.humanitarians.ai/ YouTube: https://www.youtube.com/@humanitariansai LinkedIn: https://www.linkedin.com/company/105696953/ Apple Music: https://music.apple.com/us/artist/humanitarians-ai/1781414009 Spotify: https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV #DeweyProject #AIEducation #HumanitariansAI #EducationalAI #AIAgents #PersonalizedLearning #SocraticMethod #OpenSource #EdTech #FutureOfEducation #ComputationalEducation #LifelongLearning Please like and subscribe to Humanitarian's AI YouTube to follow our progress and learn more about our transformative work in educational technology

Lyrical Literacy - Music a workout for your brain? https://youtu.be/1ErOpyD5Q1o Did you know that singing is like a full-body workout for your brain? Our Lyrical Literacy project provides songs and tutorials for anyone wanting to exercise their brain through music! When we sing, we activate more brain regions simultaneously than almost any other activity. Through our collection of tongue twisters, lullabies, and nursery rhymes, we're offering tools to engage multiple neural pathways at once. Just like balanced physical training, singing creates balanced cognitive development. Remember: 'Use it or lose it' applies to our brains too! Visit us to start your musical brain workout today! Lyrical Literacy is a program developed by Humanitarians AI that uses music, poetry, and artificial intelligence to support literacy, memory, and emotional development. Drawing on research in neuroscience, it leverages music's ability to activate multiple brain regions at once—language, memory, emotion, and motor control—to create personalized, engaging learning experiences for a range of learners. Our initiative includes subprojects focused on: Multilingual Melody (language acquisition) Rhythm & Recall (memory support) Therapeutic Tones (cognitive rehabilitation) Materials are tailored to different developmental and cultural needs. This YouTube series documents our work and invites collaboration from educators, researchers, and creatives working at the intersection of music, education, and AI. 🔗 CONNECT WITH HUMANITARIANS AI: https://music.apple.com/us/artist/humanitarians-ai/1781414009 https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV https://www.humanitarians.ai/ https://www.humanitarians.ai/lyrical-literacy https://www.youtube.com/@humanitariansai https://www.linkedin.com/company/105696953/ #LyricalLiteracy #BrainDevelopment #MusicalLearning #NeuroscienceOfMusic #CognitiveDevelopment #EducationalMusic #AIForEducation #BrainWorkout #LanguageDevelopment #MemoryEnhancement #ProfessorBear #HumanitariansAI #SingingForLiteracy #MusicAndNeuroscience #EarlyChildhoodEducation

Madison: Open Source AI Marketing Framework | Join the Project https://youtu.be/JL3d_ShVnSc Madison: Open Source AI-powered marketing intelligence framework that's transforming how America does business! Madison is a sophisticated open-source marketing system with five specialized agent layers, each handling different aspects of marketing challenges with unprecedented efficiency: 🔍 Intelligence Agents - Monitor brand reputation across thousands of sources simultaneously 📊 Research Agents - Analyze data from thousands of customers to identify market segments ✍️ Content Agents - Generate personalized marketing materials for multiple channels in minutes 👥 Experience Agents - Create digital concierges and personalized customer recommendations 📈 Performance Agents - Continuously test and optimize marketing decisions with advanced algorithms This isn't science fiction - this is an open-source research project you can join today! Whether you're interested in combining agentic AI with generative AI, verifying truth in AI, or exploring optimization through multi-armed bandits and reinforcement learning, Madison offers incredible opportunities. Join us in building the electronic marketing brain that's fast, versatile, and gives you the RIGHT answer every time! Follow Humanitarians AI: 🎵 Apple Music: https://music.apple.com/us/artist/humanitarians-ai/1781414009 🎧 Spotify: https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV 🌐 Website: https://www.humanitarians.ai/ 📺 YouTube: https://www.youtube.com/@humanitariansai 💼 LinkedIn: https://www.linkedin.com/company/105696953/ #AI #Marketing #AgenticAI #MarketingIntelligence #OpenSource #ArtificialIntelligence #MachineLearning #MarketingFramework #DigitalMarketing #AIResearch #MarTech #BusinessInnovation #HumanitariansAI

What is Agentic AI in Finance? Meet Mycroft https://youtu.be/tlyUimlR6KM It's Mycroft here! I'm excited to chat about agentic AI and how it's changing finance through our Mycroft project. Unlike basic AI chatbots that just answer questions, agentic AI takes things to a whole new level - especially for making smart investment decisions. Agentic AI works through four key steps: It perceives - gathering data from market feeds, financial databases, and regulatory filings It reasons - using large language models to coordinate specialized financial functions It acts - connecting with financial tools to execute tasks based on its plans It learns - improving through feedback loops that make it smarter over time We named our project after Mycroft Holmes - Sherlock's brilliant older brother. I'm working with Professor Nik Bear Brown on this open-source educational experiment in AI-powered investment intelligence, "Using AI to Invest in AI." Connect with Humanitarians AI: https://music.apple.com/us/artist/humanitarians-ai/1781414009 https://open.spotify.com/artist/3cj3R4pDpYQHaWx0MM2vFV https://www.humanitarians.ai/ https://www.youtube.com/@humanitariansai https://www.linkedin.com/company/105696953/ #AgenticAI #FinancialAI #InvestmentIntelligence #Mycroft #AIFinance #HumanitariansAI #AIInvesting #FinTech #MachineLearning #AIAgents #FinancialInnovation #AIResearch #OpenSourceAI #AIStrategy

From 500,000 URLs to Real Jobs: Inside the 80 Days to Stay Engine https://youtu.be/ZSJ96v8uOBM https://www.humanitarians.ai/80-days-to-stay Okay, quick 80 Days to Stay update. The bot is still running – it’s been running all night – and right now it’s chewing through about 50 sites in parallel. Each row you’re seeing here is a website where we managed to grab something. Some of them are duds, some are gold. In this video I walk through a few examples. One site looks like it needs a login or just isn’t giving us anything useful, so that goes into the “retry or ignore” pile. The next one is exactly what we want: a real company with solid about pages, detailed information, and multiple career and job listing pages. The plan is to take this huge initial pool — roughly 500,000 pages — and narrow it down to around 5,000 high-quality leads. We’ll annotate what each company offers, what kinds of roles they post (even if they’re labeled “senior”), and then push all of that into a database. Longer term, I want this to power an intelligent search: you upload your resume, and the system automatically scans everything we’ve collected and matches you to likely opportunities. That’s going to take about a month to build into a proper tool. For this week, the goal is simpler: get usable lists into people’s hands so they can start applying, then we’ll slow down and build the tool that really should already exist. It doesn’t, so I’m building it. #80DaysToStay #jobsearch #AIjobs #webscraping #careersearch

Domain Inference Complete: 62% Hit Rate | Now Using LLMs to Analyze Hiring Likelihood https://youtu.be/CGRnq-sniOY Our domain verification bot finished running! Results: 62% success rate - we found working websites for 25,748 out of 41,299 funded startups. What We Found: Companies like "Crossbow Therapeutics" → crossbowtherapeutics.com ✓ "Agent Smith Inc" → agentsmith.com ✓ "Corrine NX" → corrineNX.com ✓ Because these company names are so specific, there's a very high probability these domains actually belong to the companies. But we'll verify. Next Steps - The Scraping Pipeline: Visit each verified domain - Grab up to 50 pages per site (to prevent infinite loops) Save as HTML - Store locally in organized directories under WWW/domain.com/ Convert HTML → Text - Strip all HTML tags, keep just the content Feed to LLM for analysis - Give the language model: Raw text from the website SEC metadata about the company (funding, industry, location) Ask it to analyze like I would analyze a LinkedIn profile LLM Analysis Questions: What does this company actually do? Is this really the company from the SEC filing? Is this an industry that hires international students? Why? Is this an industry that hires recent tech/research graduates? Why? Are there job pages? People pages? What signals do we see? The LLM will read these websites the same way I read LinkedIns - looking for patterns that indicate hiring likelihood. For example: Do they list technical roles? Do they mention visa sponsorship? What's their team composition? What degrees/backgrounds do they emphasize? Why This Matters: This is for the "80 Days to Stay" initiative through Humanitarians AI - helping international students find funded startups that might sponsor work visas. By processing SEC Form D data and analyzing actual company websites, we can create a curated list of real opportunities rather than just guessing. The Pipeline So Far: ✓ Downloaded SEC Form D filings (568K companies) ✓ Filtered to relevant states/industries (41K companies) ✓ Inferred domain names from company names ✓ Verified domains with HTTP requests (25K working sites, 62% success) → Next: Scrape website content (up to 50 pages each) → Then: LLM analysis to identify hiring signals Testing the LLM prompts by hand first to get it right before automating the analysis across all 25,000 companies. #InternationalStudents #VisaSponsorship #WebScraping #LLM #AI #DataScience #StartupData #SECData #HumanitariansAI #80DaysToStay #DomainVerification #AsyncProgramming #Python #MachineLearning #TechForGood #JobSearch #F1Visa #OPT #H1B #InternationalCareer #TechJobs

Finding Domains for 41,000 Companies with Domain Inference Script https://youtu.be/46CipO_dtFM In this video, I run a domain inference script on 41,299 companies and achieve 100% success - meaning we found at least one working website for every single company in the dataset. Here's how the domain discovery and verification process works. What Domain Inference Does: The script takes a company name and intelligently guesses potential domain patterns: cloud4.io cloud4.co cloud4.com cloud-4.com And other common variations Then it checks each pattern: Does this domain actually have a live website? If we get a legitimate HTTP response, we save that URL. Sometimes companies own multiple domains, so we might keep 2-3 URLs per company. The Results: ✅ 41,299 companies processed ✅ 100% had at least one working domain ✅ Ready for verification crawling Why This Isn't Done Yet: Just because a domain responds doesn't mean it's the RIGHT company. For example, "Whisker Acquisition" could be: A pet care company A hedge fund A whiskey company Something else entirely We need to verify by actually crawling the sites. Next Step: Web Crawling & Verification The bot will: Start at the homepage - Visit the discovered domain Follow internal links - Find pages within the same domain Extract text content - Gather all the text from these pages Look for key pages - Jobs, careers, team/people pages, about pages Verify company match - Does the text mention "Clever Care Health Plan" or whatever the company name is? Infer company purpose - What does this company actually do? Assess hiring likelihood - Based on industry, size, content - likely to hire international students? Example Walkthrough: Take "Clever Care Health Plan": Domain inference finds: clevercare.com (or similar) Crawler visits homepage Follows links: /about, /careers, /team, /contact Extracts all text from these pages Verifies: Yes, mentions "Clever Care Health Plan" Finds: Jobs page at clevercare.com/careers Infers: Healthcare company, likely Series B stage, probably hires Saves: URL, career page location, company description What We'll Get From Crawling: 🔹 Job page URLs - Direct links to career opportunities 🔹 ATS platforms - Greenhouse, Lever, Workday, etc. 🔹 Company descriptions - What they actually do (in their own words) 🔹 Team pages - Size signals, diversity signals 🔹 About pages - Mission, values, funding stage mentions 🔹 Contact info - Additional validation 🔹 Verification text - Confirm this is actually the right company The Crawling Process: Crawling means: Start at page 1 (homepage) Find all links on that page Check: Are these links on the same domain? (not external links) Follow those links Find links on THOSE pages Keep going until you've mapped the whole site This gives us comprehensive text from ver.com/jobs, ver.com/about, ver.com/team, etc. Why This Matters: Some company names are ambiguous. "Cloud Four" could have: cloud4.io (tech startup) cloud4.co (consulting) cloud4.com (different company) By crawling and reading the actual content, we can verify which domain belongs to which company from our SEC filings. Current Status: Script completed: 41,299 domains found Next run: Web crawler to verify and extract content ETA: Next day or two (because it's late tonight!) For Developers: This is a standard web scraping pipeline: Domain inference (pattern matching + HTTP checks) Web crawling (recursive link following) Content extraction (HTML parsing) Verification (text matching) Enrichment (finding job pages, inferring details) Tools & Technologies: Python for domain inference HTTP requests for domain validation Web scraping libraries (BeautifulSoup, Scrapy, or similar) Pattern matching for domain generation Recursive crawling for site mapping Example Companies Found: Clever Care Health Plan Whisker Acquisition Cloud Four And 41,296 more with working domains Timestamps: 0:00 - Running the Domain Inference Script 0:30 - How Domain Pattern Matching Works 1:00 - 100% Success Rate: All 41K Companies 1:30 - Why Multiple Domains Per Company 2:00 - What Verification Means 2:30 - Next Step: Web Crawling Explained 3:15 - Example: Clever Care Health Plan 3:45 - What We'll Extract From Crawling 4:20 - Job Pages, ATS Platforms, Descriptions 5:00 - Why This Takes Time (It's Late!) Key Concepts: Domain inference: Guessing domain patterns from company names HTTP verification: Checking if domains are live Web crawling: Recursively following internal links Content extraction: Gathering text for analysis Company verification: Matching scraped text to expected company #WebScraping #DomainDiscovery #DataScience #PythonProgramming #WebCrawling #HTTPRequests #DataEnrichment #CompanyResearch #JobSearch #InternationalStudents #DataPipeline #Automation #ClaudeAI