Stories

Junevity is a new biotech co-founded by entrepreneur Rob Cahill, UCSF scientists Janine Sengstack and Hao Li, and veteran CEO John Hoekman, with a mission to reprogram aging cells using small interfering RNA (siRNA). Cahill’s path from software founder to bioinformatics and lab work was shaped by family experiences with aging and Parkinson’s, pushing him toward preventive, disease-modifying therapies. From day one, Junevity made AI part of its operating system: every employee gets ChatGPT, and OpenAI models augment the company’s genomics and machine-learning pipelines. The team estimates its first development candidate, a liver-targeted siRNA program for Type 2 diabetes and obesity, can reach key milestones at 2–6x lower cost and 2–3x faster timelines than typical industry norms.

Probably Genetic funds free at-home genetic testing and counseling, pairing it with an AI-driven symptom intake that converts long, free-form narratives into structured phenotype data and candidate diagnoses. For Elizabeth “Betsy” Minium and her daughter Kali, that approach ended a 31-year search marked by inconclusive labels and dead ends, culminating in a Pitt-Hopkins syndrome diagnosis tied to loss of function in one copy of the TCF4 gene.

Immunologist Dr. Oral Alpan is using GPT-5 Pro to surface new uses for existing FDA-approved drugs by starting from real-world clinical vignettes and working backward to mechanism. A patient with severe eczema and Food Protein Induced Enterocolitis Syndrome (FPIES) stopped reacting to wheat after beginning dupilumab, then relapsed when coverage lapsed and improved again after restarting, prompting Amerimmune to document additional cases. To test whether AI could connect the dots before the observation became part of the literature, Alpan’s team provided GPT-5 Pro the de-identified case and received dupilumab as the top-ranked candidate, alongside mechanistic rationale and risk considerations. The same workflow is now being applied to other conditions where immune subtypes may be hiding in plain sight, helping prioritize repurposing candidates such as fevipiprant for prospective study in areas like IBS and POTS.

Clinical Mind AI is a Stanford-origin platform and research program led by Marcos Rojas Pino that trains clinical reasoning through realistic, decision-based simulations delivered by text and real-time voice. Built on OpenAI’s API and developed from an early laptop prototype into a full platform, it lets instructors author cases and gives learners a telehealth-style interface with an electronic health record for documenting histories, reviewing images, and ordering tests. The system targets two core gaps that separate novices from experts: asking effective questions to gather data and recognizing which findings matter for diagnosis, treatment, referral, and disposition decisions.

At Scripps Research, PhD candidate Marco Uytiepo studies how neural circuits store experience by analyzing immense electron-microscopy datasets and reconstructing brain wiring across millions of synapses. Trained primarily as an experimental scientist, he taught himself to code and now uses ChatGPT to accelerate the unglamorous work that makes modern neuroscience possible: building data pipelines, debugging scripts, and automating 3D reconstruction workflows across tools like MATLAB, Python, and Blender. Work that previously required days of iteration can be reduced to rapid cycles that produce usable code and clearer visualizations, supporting faster hypothesis testing and more experiments within a single project.

Dr. Fleurique Franke, a general practitioner in the Netherlands, uses custom GPTs to streamline routine work in her clinics while improving the quality and tone of patient communication. By drafting clear, guideline-based responses to hundreds of e-consult messages and tailoring explanations to different reading levels, ChatGPT helps her team respond faster without losing empathy or precision. The time saved, roughly a third of her daily message workload, is reinvested in care, training, and follow-up, showing how AI can modernize clinical workflows while keeping medicine humane and accessible.

Navy veteran and educator Stephen K. Hudson founded AI Ready Veteran to give service members and their families scalable, just-in-time career coaching during the high-friction transition to civilian work. Drawing on two decades as a Navy data analyst, he built a suite of custom GPT tools that translate military roles into civilian competencies, generate tailored resumes, simulate multi-round interviews with voice practice, and coach newcomers through workplace norms that can feel like culture shock after a chain-of-command environment. Launched in May 2024, the nonprofit now serves more than 1,000 users, helping veterans identify target roles and employers, prepare more effectively for interviews, and adapt faster once hired. By turning mentorship into an always-available, highly specific guide, the platform aims to reduce early job churn and help veterans convert experience into confident, mission-ready impact.

Aidan McLaughlin, a researcher on OpenAI’s core models team, uses Codex to translate analytical questions into working front-end tools that make training data easier to see, share, and act on. By describing an investigative goal and letting Codex scaffold a browser app, he can iterate quickly on dashboards that would have taken weeks to build by hand, even without deep front-end experience. In one recent project, an end-to-end visualization surfaced a previously ambiguous signal clearly enough to shift the team from asking whether it existed to deciding how to train on it, spawning a new line of training techniques. Codex also reduces day-to-day friction by helping him navigate and explain unfamiliar parts of the codebase, accelerating alignment and experiment design across collaborators.

Casey Cuny, a 10th-grade Honors English and Senior Mythology teacher at Valencia High School and California’s 2024 Teacher of the Year, uses ChatGPT to convert students’ interests into stronger writing and deeper inquiry. His “Elaboration Conversation” prompt turns ChatGPT into a coach that offers a claim and evidence on any student-chosen topic, then pushes students to defend the idea using reasoning and rhetorical techniques taught in class. Run as a twice-weekly warmup, the activity coincided with students scoring 23% higher than the school average on a district writing assessment, while shifting classroom energy toward engaged, self-driven work. Cuny’s framework keeps the relationship-centered core of teaching intact: students draft, ChatGPT supports practice and structured questioning, and the teacher provides the human feedback and final guidance.

Sean Harrington, who teaches AI and law at the University of Oklahoma and leads technology and innovation for the law school, is using ChatGPT to replace take-home exams with live, individualized assessment that reveals what students actually understand. He built a Socratic quizbot that ingests assigned readings and conducts a short one-on-one exchange with each student at the start of class, adapting questions in real time and producing transcripts that show how each student applies rules, distinguishes cases, and reasons through hypotheticals. The transcripts also function as a feedback loop for curriculum design, helping him adjust content based on student interest and confusion, and he has open-sourced the tool for other educators. Beyond assessment, Harrington uses GPT-based assistants to support legal-tech training and reduce administrative overhead, demonstrating how AI can preserve rigor while modernizing both instruction and operations.

Iowa State software engineering student Brandon Pieczka is moving fast, stacking heavy course loads, working part-time, and applying relentlessly for internships in a competitive job market. ChatGPT helps him convert that pressure into progress by filling the gaps when he misses lectures and by turning course notes into a responsive, personalized tutorial. He uploads materials, asks for clear explanations of difficult concepts, and uses back-and-forth questioning to test and correct his understanding as if he were in office hours. The result is practical leverage: he keeps pace with demanding classes while building career momentum, landing internships early and staying on track for an accelerated path into applied AI.

Micah Blum, a sophomore at American University double-majoring in Justice & Law and Finance, uses ChatGPT to manage competing academic demands while pursuing multiple career directions. He relies on the tool to tailor resumes and cover letters to distinct roles, surfacing the most relevant coursework and experiences for legal, finance, and campus safety positions, an approach he believes helped him land a public safety supervisor role. In class, ChatGPT supports research and writing by helping locate niche datasets, suggesting visualization approaches, and providing targeted feedback across essay drafts. For planning, it helps him estimate timelines and sequence tasks, while he keeps authorship and judgment firmly human, using AI to accelerate structure and focus without replacing his voice.

Conor Grennan, NYU Stern’s chief AI architect and CEO of AI Mindset, is working to help educators use ChatGPT to make learning more vivid, memorable, and individualized without adding workload for teachers. His partnership with his teenage son Finn illustrates the approach: while studying 19th-century German immigration to New York, Finn used ChatGPT to roleplay a fictional immigrant and “walk” through mid-1800s Manhattan, turning historical facts into a lived narrative he could recall like a film scene. The same method can recast complex topics into tailored scenarios, from cell biology framed as a superhero story to Hamlet reimagined as a modern conversation, helping students connect emotionally and cognitively with the material. Grennan’s broader thesis is that conversational AI can finally make practical the kind of personalization classrooms have long wanted, aligning instruction to each student while keeping rigorous content intact.

Minnesota’s new Enterprise Translations Office (ETO) was created in 2022 to serve 25 state agencies with a six-person linguist team, a scope that made traditional translation workflows hard to sustain. Under director Adam Taha, ETO adopted ChatGPT Enterprise to produce a first-pass translation that human linguists then review and polish, feeding corrections back in a weekly improvement loop that has lifted Somali accuracy from roughly 90% to about 94–95%. After launching operations in mid-2024, ETO now handles 400 to 600 requests per month, prioritizing Spanish, Somali, and Hmong, while delivering what Taha estimates is $120,000 to $140,000 per month in cost avoidance by reducing reliance on external vendors. The practical impact is broader than savings: more timely, consistent translations for high-stakes materials such as public health notices, voter information, and driver licensing documents, helping diverse communities access information they can use.

mRelief co-founders Rose Afriyie and Genevieve Nielsen are using ChatGPT to expand a nonprofit that helps families claim SNAP benefits quickly and with dignity, reporting 5.2 million people served and $2.1 billion unlocked to date. With frequent partnership meetings across the country, the team uses AI to tailor pitches to each partner’s geography, mapping impact to specific ZIP codes and reframing results in the metrics that resonate, cutting preparation time while improving alignment. On the service side, AI helps eligible families follow through: an outreach chatbot answers questions and nudges applicants back into the flow, and a document-checking model improves photo submission quality, increasing first-try completion by 20% and reducing thousands of manual reviews in a South Carolina pilot. The result is faster, more reliable benefit access in a system where delays can mean missed meals.

United Way adopted OpenAI Enterprise to improve quality and productivity across its organization, pairing the rollout with governance and an internal AI Ambassador program led by CTO Lawana Jones. A standout use case is contract review: ChatGPT now helps the team extract key terms, renewal dates, and non-standard clauses in seconds, turning long agreements into actionable checklists and helping leadership avoid expensive mistakes. In one case, the tool surfaced a little-noticed termination window that could have triggered an automatic renewal, and it also supports version comparisons when vendors submit “revised” contracts to ensure changes match what United Way requested. Beyond legal workflows, teams use ChatGPT for communications, research briefings, and rapid deck drafting, reclaiming hours while reducing risk and keeping more attention on vendor strategy, security, and mission delivery.

When the Carl B. and Florence E. King Foundation received more than 200 grant applications in a cycle that typically brings about 90, president Kim Davis used GPT-5 to help evaluate submissions and clarify which proposals best fit the foundation’s goals. Rather than producing blunt rankings, the model explained why certain mission statements worked better and offered alternative rewrites tuned to different contexts, from a bold, board-memorable version to a warmer, public-facing one, each with clear rationale. Davis noted that earlier models often required repeated prompting to stay consistent, while GPT-5 delivered faster, more precise, and more instruction-following support from the outset. The result was a workflow that preserved judgment and standards while helping the foundation communicate more clearly and move through a higher-volume grant cycle with confidence.

When the Greenville Arts Council learned at 4:48 PM that a $20,000 grant proposal was due by 5:00 PM, board president Mary Catherine Brooks used ChatGPT to turn urgency into a winning submission. She uploaded prior successful proposals and had the model draft a new application fast enough to file before the deadline, securing funding for arts and education programming for children and families across the Mississippi Delta. After participating in OpenAI’s Nonprofit Jam, the council applied AI in a high-stakes moment where time was the main constraint, showing how small bursts of support can translate into real community impact.

Kaija Pack turned grief after her husband’s sudden death into Break Life Houston, a 10,000-square-foot “rage room” where people safely vent stress by smashing objects in themed spaces, including a dedicated scream room. After attending OpenAI’s Small Business Jam, she began using ChatGPT to research competitors, refine pricing conversations, and boost weekday traffic, helping a small family-run team operate more efficiently. The tool also helped her craft clearer messaging about why Break Life costs more than smaller rivals by emphasizing the bigger, better-equipped experience. For Pack, AI became a practical force multiplier that supports growth while staying rooted in the business’s core purpose: making it acceptable to feel anger and express it safely.

Facing a tight two-week window to sell out a leadership summit in Bangkok, Woody Milintachinda turned to GPT-5 for a practical growth plan. After reviewing his public-facing materials, the model produced a detailed, step-by-step strategy he could share immediately with partners and staff, giving the team clearer next actions and renewed confidence. Woody then extended the workflow to content acceleration, using GPT-5 to extract podcast highlights, turn them into short-form scripts with captions, and align posts with what was trending on Thai social media. The result is a faster path from insight to execution, adding AI-driven planning and distribution to the toolkit of one of Thailand’s best-known hosts and producers.

Diagon, founded by longtime manufacturing procurement leader Will Drewery, applies GPT-5 to one of the most consequential bottlenecks in industrial growth: finding and buying highly specialized machines. Drewery’s experience sourcing equipment across global factories, including managing roughly $3.5 billion in capital spending at Tesla during a period of explosive workforce growth, shaped a clear insight. Plants create jobs and revenue only after the right machines are installed, yet procurement remains slow, opaque, and reliant on fragmented information and expert memory.

After breast cancer, Allison Leeds found that medicine could explain a surgical plan but offered little guidance on what recovery would feel like day to day, emotionally. She began using ChatGPT in early 2025 to synthesize more than 11,000 patient experiences drawn from public forums, then paired those insights with research in color science and neuroaesthetics to explore how visual environments can shape stress and pain. The result became a new tool for recovery: weekly art cards that combine color palettes with mantras, designed to arrive in cadence with the hardest parts of healing. Shared with support groups, the cards quickly resonated with patients and caregivers and grew into Leeds’s business, built between surgeries with AI acting as a research partner and product scaffold. Her broader body of work, The Healing Spectrum, is now exhibited at SFMOMA, extending the same idea: recovery needs information for the heart, not only the body.

When Jasmine Hudson and her team at Black Paper Party wanted to reimagine “The 12 Days of Christmas” through a Black cultural lens, they used ChatGPT to brainstorm lyrics that became the creative anchor for a new holiday product line. The resulting remix, full of vivid references from Tuskegee Airmen to gold grills, helped the team expand a spark into a cohesive collection that refreshes familiar seasonal imagery with authentic representation, now carried by Walmart. Beyond that signature collaboration, Jasmine uses ChatGPT to accelerate product ideation, refine brand voice in the company’s “rich auntie” tone, build seasonal plans, and support operational work like inventory forecasting, partner discovery for pop-ups, and nonprofit outreach, letting a lean team move faster without burning out.

North Carolina’s Department of State Treasurer ran a 12-week pilot with OpenAI to train staff on using ChatGPT with public information to speed audit reviews and help identify unclaimed property that could be returned to residents. Preliminary results reported by the Treasurer’s office show strong user reception, with 85% of participants describing their experience as positive and estimated daily time savings of roughly 30 to 60 minutes, particularly for drafting and editing work.

After a life-threatening rare autoimmune clotting disorder left Rich Kaplan facing kidney damage, his prescribed medication stalled for months because an insurer demanded evidence the drug worked for a condition as uncommon as catastrophic antiphospholipid syndrome. Kaplan used ChatGPT to assemble a referenced synthesis of trials, case reports, and studies, paired with a plain-language note explaining what was at stake; the packet ultimately helped secure approval through third-party arbitration. Since then, he has used ChatGPT as a practical tool for navigating the ongoing complexity of chronic illness, checking potential drug interactions against his medication list, flagging risks tied to kidney disease, and translating dense medical records into clear summaries and questions for clinicians. For Kaplan, the value is agency: turning scattered information and bureaucratic friction into actionable support that helps him pursue better care and stay actively involved in his own health decisions.

Physicist Alex Lupsasca approached AI the way he approaches big claims in science: by stress-testing it. In early 2025, ChatGPT proved useful for academic housekeeping and for turning rough, multilingual drafts into clear scientific prose, but he did not view it as a tool for producing verifiable predictions.

Mathematician Ernest Ryu, one of more than 1 million weekly ChatGPT users working on advanced science and math topics cited in a new OpenAI report, began testing whether language models could translate real-world scenarios into rigorous optimization problems and learned early systems still missed key constraints. After reasoning models arrived, he used ChatGPT as a fast, iterative collaborator to solve an open problem related to Nesterov acceleration across three late-night sessions, verify the proof to publication standards, and later bring that same math-first mindset to OpenAI’s synthetic data team.

In Miles City, Montana (pop. 8,400), family physician Dr. Marjorie Albers practices the kind of wide-scope medicine rural America still depends on, and she uses AI tools to make her workload sustainable.

With ChatGPT, Ayrin Santoso found a way to navigate care for her mother, Fifi, who lives in the Indonesian city of Surabaya. Ayrin, a tax professional working for OpenAI in San Francisco, had already been using ChatGPT regularly, driven by a duty she feels to her aging parents: “How can I actually still be there and help them when they need something?”

Matt Rosenberg and his wife, Chef Kamonwan, used ChatGPT to evaluate neighborhoods, model restaurant unit economics, and ultimately pivot from an unaffordable lease to a legal home-kitchen path that let them test the concept with real customers. By translating dense regulations into checklists, pricing and packaging guidance, and a daily operating plan inside ChatGPT Projects, they soft-launched Bangkok Rush Thai Kitchen, served dozens of meals with five-star feedback, and opened a runway toward a future brick-and-mortar or franchise.