18 / 07 / 25 – Obvious Future Team
Why Building for Privacy Matters: The Case for Resident AI
A New Urgency Around Data Privacy
The global shift to cloud computing promised speed, scale, and flexibility. But as artificial intelligence becomes central to digital transformation, a new reality is setting in. According to IBM’s Cost of a Data Breach Report 2024, 40% of breaches involved data stored across multiple environments. The highest average cost? Public cloud breaches, with losses reaching $5.17 million per incident. The same report noted a 10% year-over-year increase in global breach costs—the highest ever recorded.
Meanwhile, Verizon’s annual Data Breach Investigations Report found that back in 2021, for the first time, external cloud assets were involved in more incidents than internal infrastructure. Since then, the risks of relying on shared, remote infrastructure have only grown.
At Obvious Future, we believe it’s time for a different approach—one that puts ownership, control, and data security back into the hands of organizations. We call it Resident AI.
Recent Data Breaches: A Pattern of Risk
Cloud vulnerabilities aren’t hypothetical—they’re making headlines. Take AT&T’s April 2024 breach, when hackers accessed its Snowflake cloud environment using stolen credentials and no multi-factor authentication. As reported by Hackread, the attackers compromised the metadata of nearly 110 million customers, targeting over 160 companies in the process. Within a month, an additional 86 million records were leaked.
According to the 2025 Verizon Data Breach Investigations Report, 30% of breaches now involve third-party environments—double the rate of the previous year. Many incidents exploited vulnerabilities in unpatched software or stemmed from long-remediated secrets in repositories like GitHub (with a median delay of 94 days to patch). More concerning still: 17% of breaches were linked to espionage motives, and 28% of attacks involving state-sponsored actors were financially driven.
Together, these patterns show the escalating complexity—and cost—of handing over sensitive operations to shared infrastructure. As enterprises rely more on SaaS and cloud storage, these cases serve as urgent reminders that security vulnerabilities can escalate rapidly—and the cost isn’t just financial. It’s also reputational, legal, and operational.
Owning the AI: Why On-Prem Matters
As Eddi Weinwurm, CEO of Obvious Future, explains: “Our AI can run entirely on-premise, without relying on cloud infrastructure. That’s crucial because while the cloud may seem convenient in the short term—it becomes expensive and problematic in the long run.”
While cloud AI services may seem efficient in the short term, they come with major downsides over time—especially for industries that handle sensitive data. Shared infrastructure, limited customization, variable latency, and exposure to vendor instability all pose significant risks.
When using a model like ChatGPT, users are borrowing time on a platform shared by thousands. This often leads to slower, less predictable responses and removes the ability to fully tailor the experience. Worse, if a service discontinues its API or changes its terms, companies can lose access to critical systems overnight.
CaraOne takes a different route. It’s designed with a modular AI architecture that runs locally. Organizations can start small and scale as needed, without investing in massive data centers or dealing with complex procurement processes. It’s a way to bring cloud-like flexibility without sacrificing control.
Customization and Control: AI That Adapts to You
Most cloud-based AI systems are built to serve everyone—which means they’re rarely ideal for anyone in particular. They offer limited customization, and adapting them to highly specific use cases can be nearly impossible.
By contrast, Obvious Future develops its own models in-house, allowing for precise tailoring to each client’s needs. In one case, CaraOne was trained to understand the entire fictional universe of a science fiction franchise—from planets and species to ships and environments. In another, it was taught to identify individual motocross riders despite them wearing nearly identical helmets and gear.
This level of customization is only possible when organizations fully own and operate the AI infrastructure. It allows for purpose-built tools that deliver real value instead of generic features that require teams to compromise. And because Obvious Future builds the models internally, it also means clients get long-term support and the freedom to evolve.
A Smarter, Safer Future
Resident AI isn’t just a technical innovation—it’s a philosophical one. It shifts power away from distant providers and back into the hands of teams who need reliability, control, and security.
At Obvious Future, we believe that data privacy should not be a luxury. And that real innovation must respect the boundaries of trust, compliance, and creative ownership.
If your organization is rethinking its cloud strategy—or if you’re building the next generation of media workflows—it’s time to ask: Could Resident AI work for you?
Discover More
Why Building for Privacy Matters: The Case for Resident AI
A New Urgency Around Data Privacy
LLMS: Start to Think Small
Why Huge LLMs Aren’t the Solution for Everything
Building an AI Startup
A COO’s Perspective on Product, People and Process
Discover More
