human-in-the-loop-ai
(17)
Fin Built an Assistant. The Real Product Was the Tape.
Fin tried to sell busy people an extra pair of hands. The hands were expensive, but the machinery built to study them became a second company - and a useful warning about where human-in-the-loop AI actually earns its keep.
Fullintel Put People Back in the Machine
Media monitoring promised to automate the morning briefing. Then the alerts became another job. Fullintel built its business on a counterintuitive fix: keep the software, and put trained readers back in the loop.
PublicRelay and the Seven Stories That Matter
Most media software promises to find everything. PublicRelay built a business around a fussier proposition: find the few things worth acting on, then put a human name behind the answer.
Machinify’s $670 Million Bet: Fix America’s Medical Bills Before They’re Paid
Machinify started with a broad promise to turn enterprise data into decisions. A decade later, it is making a narrower, harder wager: teach machines the rules of American healthcare, then catch costly claims errors before the money leaves the building.
Empathy Turned Grief's Worst Chore List Into a 50-Million-Person Benefit
The startup began with an app for the brutal paperwork after a death. Its bigger invention was persuading insurers and employers to put practical, human-guided care inside benefits people already have.
Jeremy Bellinghausen Keeps Building Marketplaces Where People Still Matter
The Scripted CEO has spent two decades tuning marketplaces - from hotel rooms and music to business intelligence and freelance writing. His durable idea is simple: software should remove friction without removing the people who create the value.
John Kim’s Bet: Let AI Do the Chasing, and Let People Make the Match
Paraform’s co-founder spent midnight-to-lunch shifts in a university library learning an unfashionable lesson: when software makes outreach abundant, judgment and trust become scarce.
DeweyLearn Put an AI Assessor in the Room - Now It Has to Earn the Expert's Trust
The New York startup moved beyond chatbot-style tutoring to judge what people actually do - from knife work to therapy practice. A $5 million Series A gives it room to prove that expert feedback can scale without becoming generic surveillance.
The Company That Turned Video Games Into a Cancer Lab
An Allen, Texas company spent years convincing gamers that tracing a line in Minecraft could help fight cancer. Then it had to figure out how to get paid for it.
The Startup That Turns Drone Photos of Cell Towers Into Money
Anne Zink built a company that lets you inspect a 300-foot cell tower without climbing it. The trick isn't the drone - it's what happens to the pictures after they land.
TaskUs Built a 63,200-Person Business for the Work AI Still Can’t Do
The outsourcing company grew up answering tickets for tech startups. Now it is training models, testing robots and moderating the messiest corners of the internet - while teaching its own 63,200-person workforce to work beside AI.
Fetcher
Fetcher is a New York-based AI recruiting platform that automates the top of the hiring funnel - finding, verifying, and reaching out to qualified candidates so talent acquisition teams can spend their time interviewing instead of searching. It pairs machine learning against a database of roughly 500 million professional profiles with human sourcers who quality-check every batch, then delivers vetted candidates and automated, personalized outreach directly into recruiters' inboxes and applicant tracking systems.
Accend
Accend builds human-in-the-loop AI agents that automate commercial credit underwriting for banks and fintechs. The San Francisco startup, founded in 2023 by ex-Brex risk operators, ingests financial documents and tax filings, spreads financial statements, models cash flow, and drafts credit memos - with every AI-generated spread reviewed by human analysts. Customers report application processing times cut by up to 80%, and the company positions itself as a replacement for offshore BPO spreading teams.
Recidiviz
Recidiviz is a nonprofit technology company that stitches together the fragmented data trapped inside America's prison, parole and probation systems and turns it into usable tools. Its platform - and its newer AI-powered Case Planning Assistant - helps corrections agencies in roughly 19 states surface people eligible for early release, automate case triage, and connect returning citizens to housing, treatment and jobs. Founded in 2019 as a volunteer project at Google, Recidiviz's mission is to accelerate progress toward a smaller, fairer, safer justice system.
Pranjal Daga
Pranjal Daga is the co-founder and CEO of Accend (YC S23), a San Francisco fintech building human-in-the-loop AI agents for commercial credit underwriting and compliance at banks and fintechs. Before Accend he led AI product on the Risk team at Brex, where his work helped prevent roughly $20M in fraud losses, and earlier helped build Cisco Innovation Labs from a single person to a 35-person team. He dropped out of a PhD in AI/ML at Purdue to chase the build, has done research at Adobe and IBM, and is racing to replace slow, manual back-office compliance work with audit-ready AI.
Mark Atkinson
Mark Atkinson is the Co-Founder and CEO of Mursion, an AI-powered immersive learning platform that blends live human performers with artificial intelligence to create realistic practice simulations for professional development. An Emmy Award-winning television producer turned serial edtech entrepreneur, Atkinson has spent over two decades building technology ventures at the intersection of human capital development and innovation. Before Mursion, he co-founded Teachscape, a pioneering video-based teacher observation platform that became central to the Gates Foundation's Measures of Effective Teaching project. Under his leadership, Mursion has grown to serve more than a third of Fortune 100 companies, delivering 18 million-plus minutes of simulation-based training across healthcare, education, corporate, and hospitality sectors.
Spence Green
Spence Green is the co-founder and CEO of LILT AI, a San Francisco-based enterprise AI translation platform that has raised $95.5M across three funding rounds. A Stanford PhD in computer science and University of Virginia computer engineering graduate, Green built LILT in 2015 alongside co-founder John DeNero after both worked on Google Translate. His research career spans Northrop Grumman defense systems, Johns Hopkins NLP research, and Google, where he developed English-to-Arabic translation models. LILT now serves government agencies including the U.S. Air Force and global enterprises, combining adaptive machine translation with human oversight. Green's guiding philosophy - 'the world belongs to the discontented' - reflects his conviction that universal multilingual information access remains an unsolved problem.