Back Office vs. Front Office
Hey there, investors and innovators!
This week's crop of launches out of the current batch makes an unusual pair: on one side, founders are going after the unglamorous, judgment-heavy back-office work that keeps large, regulated organizations running; the kind of work nobody photographs for a pitch deck. On the other, founders are going after culture and communication directly; content, language, the stuff people actually watch and say to each other. It's a reminder that "AI company" now spans an enormous range, from a drone quietly optimizing a solar farm's output to a platform producing entertainment an audience binges by the hundreds of millions of views. Different audiences, different stakes, same underlying bet: that AI is now good enough to take on both the operational grind and the front office frontier at once.
What the Pattern Says
- The back office is becoming AI's next frontier, not just the front end. Proxy voting, hardware procurement, supply chain logistics: these are exactly the kinds of specialized, judgment-heavy functions that used to require dedicated headcount and institutional memory. This batch treats them as automatable, which says something about how far founders think AI's reasoning has come, not just its content generation.
- Physical infrastructure is getting its own optimization layer. A drone clearing clouds over a solar installation isn't just a novelty, it's a sign that founders are chasing marginal efficiency gains in physical, capital-intensive assets that were previously left to the weather. Expect more of this: AI applied not to replace physical infrastructure, but to squeeze more performance out of what's already built.
- Real-time multimodal AI is closing the gap between people, not just people and machines. Between AI-native entertainment and technology aimed at collapsing the language barrier in real time, this batch is using multimodal capability for something more ambitious than a better chatbot: making communication itself faster, cheaper, or entirely automated in areas from live commerce to telehealth.
- Deep technical pedigree keeps showing up as the differentiator. Founders with backgrounds spanning institutional finance operations, published research in human animation, and specialized hardware supply chains are building the moat as much through expertise as through the product itself. This is a pattern that continues to show up across this cohort.
- Batch-wide context: S26 is tracking its record-setting predecessor's shape; YC's 2026 batches are running roughly 60% AI, with B2B still dominant. What stands out this week is the range within that: automating institutional-grade back-office functions and reshaping consumer content and communication are both fully in scope for a single AI-heavy batch.
On the Radar
Demo Day: September 10, 2026. Expect the pace of launches to accelerate meaningfully through August as founders get in front of investors ahead of it.
Eight Capital tracks every YC batch, cohort to cohort. Reach out if you'd like the full running list of this batch's launches or a deeper cut on any theme above.
Fund Manager Picks
Pennant
The Corporate Governance OS for Public Markets
Why it's interesting
Pennant is a Y Combinator S26 company founded in 2022 and headquartered in Minneapolis, MN. The company provides a unified SaaS platform for proxy voting and governance intelligence, specifically designed for institutional investors. Pennant enables portfolio managers, governance teams, and compliance officers to analyze, coordinate, and execute proxy voting in alignment with fund strategies, offering features such as governance-risk surfacing, policy-aligned recommendations, and automated compliance workflows. The founding team brings deep domain expertise from institutional asset management and proxy voting. Pennant’s mission is “to empower financial institutions and institutional investors with innovative governance and voting intelligence solutions that enable them to transform, adapt, and thrive in the digital era.”
Waybill
Managed procurement & inventory for hardware teams.
Why it's interesting
Waybill represents a bet on the ongoing digitization of courier and delivery operations, targeting a large and growing market segment that remains underpenetrated by modern SaaS solutions. The company’s platform addresses a clear operational bottleneck—manual and inefficient dispatching—by automating key workflows and integrating real-time tracking and financial management. The founder brings relevant operational and technical experience, and the YC S26 signal provides additional validation. The “why now” is supported by accelerating e-commerce and rising customer expectations for delivery transparency and efficiency.
Meteoric
Drones that clear clouds over solar farms
Why it's interesting
Meteoric is a London-based startup developing drone technology to increase the energy output of solar farms by reducing the reflectivity of clouds overhead. The company’s drones autonomously fly into low and mid-altitude clouds above solar installations, altering cloud droplets to allow more sunlight to reach solar panels. Meteoric’s approach requires no chemicals or new infrastructure, offering a measurable increase in solar energy generation. The founding team consists of Eric Nilsson (CTO) and Mete Karslioglu (CEO), both with engineering backgrounds from the University of Cambridge. The company’s mission is: “To open up the skies over solar plants and increase renewable energy output by reducing cloud reflectivity using drone technology.”
Mochi.TV
1 Minute Anime
Why it's interesting
MOCHI.TV is a vertical, mobile-first platform delivering bite-sized anime content, optimized for daily consumption on smartphones. Founded in 2025 and headquartered in San Francisco, the company is led by Robert Qiu, a UC Berkeley alumnus with prior experience scaling gaming and NFT ventures. MOCHI.TV offers daily updated, easily consumable anime stories that users can watch, engage with, and create, aiming to broaden anime’s audience through a mobile-centric approach. Its mission is to make anime accessible and engaging for the next generation of mobile users.
Familiar
World Models for Accurate Humans
Why it's interesting
Familiar is going after real-time video translation; dubbing content into any language while preserving the original speaker's voice, face, and delivery, live. Founders An Zhu Liu, Mingi Kwon, and Xu Zheng lead a team of PhDs and researchers who helped originate the field of joint audio-visual human modeling and co-authored the current state of the art in real-time human animation. That's a meaningfully different starting point than most dubbing startups, which stitch together off-the-shelf speech and video models. Familiar is building the underlying world model for human faces and voices from first principles. The team has published benchmarks claiming to outperform ElevenLabs, YouTube, HeyGen, and Meta on translation accuracy, while being the only solution in the set that runs in real time rather than requiring post-processing.
