Why We Invested: Captur
On-device AI that instantly verifies user-submitted photos.
Captur is edge-AI infrastructure that instantly verifies user-submitted photos for quality and accuracy, helping mobile-native businesses reduce disputes and improve customer experience. The technology runs directly on a user’s phone, analyzing images in milliseconds without needing an internet connection or racking up large token costs. Captur processes tens of millions of images monthly across delivery, mobility and e-commerce platforms.
The company recently raised a $6 million seed round led by Rally Ventures, with participation from existing investor Sure Valley Ventures (SVV). Below, Founder & CEO Charlotte Bax discusses the shift to on-device AI, building trust in automated systems and why the hardest problems in AI are about people, not technology.
1. You’ve spent nearly a decade building applied AI and computer vision products. When did you realize real-time image verification was a major unsolved problem, and how is Captur addressing it?
I spent years watching computer vision advance while smartphones became increasingly powerful. That convergence made it possible to move AI processing from the cloud to the device, unlocking faster performance, offline capability and eliminating per-transaction cloud costs. The problem was that companies didn’t have the tools to make that shift.
Mobile businesses rely on user-submitted photos for critical operations such as proof of delivery, parking compliance, identity verification and insurance claims. When images are low-quality or do not show what they should, it creates serious operational and trust issues. Disputed deliveries and compliance errors require costly investigations, and these problems add up quickly across millions of transactions.
Captur solves this by giving users immediate, actionable feedback as they take photos. Everything happens in real time on the device. When a scooter is parked incorrectly, for example, the rider knows instantly — before a city fine is issued. For GoBolt, a last-mile delivery company, if a driver accidentally photographs their foot instead of the delivery at the customer’s door, Captur flags the mistake on the spot, preventing complaints and disputes. That instant, specific feedback does not exist anywhere else.
2. Captur runs powerful computer vision models directly on-device, rather than in the cloud. Why does that architectural choice matter, and what does it unlock for customers?
Running on-device unlocks three major advantages: speed, cost and privacy. Our models process images in 30 milliseconds—about 10x faster than the blink of an eye. There’s no delay from sending data to the cloud and waiting for a response. It works even without the internet, and user images stay on the device, which is critical for privacy and compliance.
Before Captur, companies had two options: slow, costly cloud AI that charges per image and can lag or fail when users are offline, or manual checks for every image. Neither scaled well, and both left businesses exposed to fraud, mistakes and poor customer experiences. For a driver or field technician, there’s no time to wait for a photo to be checked. If it isn’t approved on the spot, it’s too late. On-device processing solves all of that: verification happens instantly, it scales without added infrastructure and costs stay flat regardless of volume.
The tradeoff is that building for phones is much harder upfront. You’re constrained by the device’s processing power, and the models have to work reliably across thousands of phones, from budget Androids to the latest iPhones. We support more than 6,000 device types, each with different chips, cameras and performance profiles. Delivering human-level accuracy at 30-millisecond speeds across that range is a real technical challenge, and it requires deep expertise in both AI and mobile engineering.
3. You’ve said the hardest problems in AI aren’t technical, they’re about people. How has that belief shaped the way you’ve designed Captur’s product and customer experience?
There’s a quote I love: if AI innovation stopped today, it would take 10 years for adoption to catch up. Technology advances exponentially, but people don’t. That idea shaped Captur from day one. We knew that if AI was going to run in real time without human oversight, especially in high-stakes situations, it had to be trusted.
For us, that meant setting the bar at human-level or better accuracy. Anything less wouldn’t earn confidence from product teams or their users. We work closely with customers to help them understand performance, define what “good” looks like and build the right feedback loops so the system is reliable enough to deploy without manual review.
We built the team around that standard. Our founding AI engineer has years of experience building high-scale production systems (i.e. machine learning that needs to work with millions of users in the real-world, not just in the lab). Our lead mobile engineer came from Qualcomm with deep embedded systems experience. The hardest part isn’t just building strong models. It’s building systems that product teams trust and users can rely on every time.
4. Before Captur, you founded Mars Needs Women to drive awareness and funding to women in STEM, which went viral and landed partnerships with NASA and others. What inspired it, and how did that experience shape your approach to leadership and building companies today?
The idea for Mars Needs Women came while I was at an early-stage startup in New York. I saw the talent gap for female engineers firsthand and wanted to do something about it. It started small—selling hats at my WeWork—but quickly snowballed into a viral brand. People really connected with the message.
That experience taught me the power of community-first building. Meetups, hackathons and spaces where people could connect showed me that the best companies and movements are built around communities, not just products. When you create something people genuinely care about, it takes on a life of its own.
The viral moment and partnerships with NASA validated that there was real hunger for STEM advocacy. But more importantly, it taught me that leadership isn’t about having all the answers, it’s about creating space for others and building momentum around ideas that matter.
With Captur, I’ve applied the same lessons. We’re not just selling software; we’re educating a market on on-device AI and building trust in automated systems. Who knows—now that I’m back in New York, maybe Mars Needs Women will see a revival!
Ben Fried, Partner at Rally Ventures, shared his thoughts on the investment and what makes Captur such a compelling opportunity:
This investment checks every box. Charlotte is a talented, multi-time founder, and Captur is tackling a hard technical problem that creates a deep competitive moat. The team achieved human-level accuracy at 30-millisecond speeds across 6,000+ devices—something larger teams at top companies couldn’t replicate. That combination of difficulty and specialized talent makes this a compelling opportunity.
Captur also highlights why we remain bullish on enterprise software. This isn’t a simple database app. It requires deep expertise in computer vision, mobile optimization and on-device AI, creating durable advantages. The ChatGPT wave legitimized a broader field called deep learning—building specialized neural networks that can solve problems traditional software can’t. We believe deep learning is a much richer area for innovation than just chatbots, no matter how impressive those are.
Captur exemplifies this. There’s no chatbot running on your phone—there’s a specialized vision model optimized for real-time, on-device performance. This is similar to Incept AI, another portfolio company where the founder’s years of experience building customized audio processing models became a huge differentiator. Small, specialized teams with deep expertise can beat much larger teams at big tech companies. We’ve seen it happen repeatedly.
We like that Captur has proven traction in very different industries—mobility and food delivery—and it’s easy to imagine other applications. The unit economics are particularly attractive: cloud-based AI scales costs with usage, but Captur’s on-device approach keeps costs flat while margins grow as customers process more images. With strong initial revenue, impressive customers and a product that’s genuinely hard to replicate, Captur combines technical excellence, market need and favorable economics—the exact kind of opportunity we look for at Rally.
