AI Makes Waves. Inverted AI will Keep the Ship Afloat

If AI is going to change the world, how will the world need to change?

The wave of AI companies taking the world by storm has been driven largely by the groundbreaking technology itself – the foundation models. A secondary wave of excitement has been driven, like a gold rush, to be the first to apply these near-miraculous language generation tools to specific industries – customer service and coding to start, but also 100 other white-collar service jobs like lawyers, accountants and HR.  

But we believe that as AI makes contact with reality, a third wave will emerge: companies that succeed not because they harness AI directly, but because they contain and structure the power of AI. We call this the 'inverted AI' thesis.

What is 'inverted AI'?

'Inverted AI' refers to companies that benefit in direct correlation to AI-adoption, but whose core business does not use AI. A simple example from public markets is the way that energy demand makes utility stocks an AI investment.  
‍
These companies are the mirror image of the foundation model companies; they stand out by addressing the unique challenges, opportunities and disruptions created by AI’s proliferation. Where the latter builds the AI-driven future, 'inverted AI' companies enable or adapt to the consequences of that future. They will succeed because LLM-driven workflows create new problems, require new cultural understandings, and enable new business models.

Venture opportunities

While venture investors are not going to start investing in public utilities, AI energy demand has driven big tech to become capital heavy as they invest in infrastructure, energy solutions and utility partnerships. Within the startup ecosystem, this can mean hardware innovations like the diamond cooling technology at Akash, or the ways that Claros is re-architecting the way that power flows to chips. The same story is at the heart of renewed interest in nuclear power.  

Nascent 'inverted AI' opportunities

Beyond energy, the 'inverted AI' thesis is already at play across various industries. This trend exemplifies how AI creates entirely new categories of risk and responsibility, which then leads to novel solutions. We’ve previously explored areas like:
‍

• Merchant acquiring for AI-driven agents: As AI agents proliferate, tailored financial infrastructure will become essential.
• Insurance products for AI-driven robots: Scaling robot adoption will require new approaches to liability and risk management. Waymo and other self-driving leaders have pioneered bespoke solutions, but the broader market remains underserved.

Blending AI usage with 'inverted AI'

There are companies that are blending direct AI usage with an 'inverted AI' strategy to deliver outcomes to their customers. Ntropy (QED portco) provides data cleaning at scale, overcoming a roadblock for financial institutions to move from AI experiments to production. Spellbook (Moxxie portco) provides a complete AI suite for transactional lawyers and they streamline the drafting process with expert-created content integration. The work of curating and prepping the content engine is key here.
‍
Ocrolus (QED portco) has transitioned much of its internal data extraction and analytics to LLMs. It’s not just that they provide data extraction and analytics, they offer their customers peace of mind that they’ll remain at the forefront of AI adoption. But as they’ve gone through their journey, they’ve seen just how critical and difficult benchmarking LLMs against each other can be. Now Ocrolus is working towards products that will continuously monitor and benchmark LLM performance. Pharos Health (Moxxie portco) automates the work of data gathering and reporting for hospitals. By automatically pulling insights from patient charts, they give hospital admins visibility without adding to IT backlogs or clinical reviewer workloads.
‍
The internet is being rebuilt by AI. The cost of content creation is dropping to zero. But this creates new challenges for creators seeking to monetize their work. For example, affiliate links must still be embedded manually or with separate tools. Another QED portco, Wildfire, just launched Revenue Engine. Using masterprompting, they automate affiliate embedding and enable creators to close the revenue loop in AI-driven content creation. While not an AI product in the traditional sense, it exemplifies the inverted AI thesis by empowering a new wave of small-scale entrepreneurs thriving in the AI-transformed internet.
‍
In addition to that, here are more opportunities that are broadly emerging:
‍

• Human-in-the-loop platforms: AI systems still struggle with edge cases, requiring human oversight for critical functions like content moderation, model fine-tuning and decision support. Companies providing services that integrate humans with AI systems can scale alongside AI adoption.
  ‍
• Model validation: Model validation is a core regulatory requirement in financial services and other highly regulated spaces. The workflow tools, testing methods both pre-deployment and in production will require deep industry knowledge and trust that the LLM safety teams are unlikely to be able to conquer on their own. This is more than just explainability or ‘alignment’!
  ‍
• Privacy-preserving technologies: AI systems are consuming vast datasets. And with this trend, privacy concerns are growing. But business users aren’t equipped to navigate this new world. For example, small companies must use AI to keep up, but they can’t possibly exert market pressure on the LLMs or adapt open-source tools themselves. Meanwhile, their customers are requiring boilerplate legal templates that ask businesses to prevent even “indirect” processing of personally identifiable information and to guarantee “secure environments.” 'Inverted AI' companies can thrive by offering solutions for secure data sharing, anonymization and privacy-preserving techniques.
  ‍
• AI governance: Large companies have vast footprints, so they need to make sure they can account for all their behavior. Auditing AI systems and ensuring that they’re compliant is a key use case for them.
  ‍
• AI data supply chains: AI requires immense amounts of structured and unstructured data. Companies that specialize in sourcing, curating, labeling and managing these datasets are essential to AI workflows. Blockchain solutions have been proposed to handle this, but this is a complex coordination problem. Will industry find ways to manage this flow or will regulation (as usual Europe will move first) be required?
  ‍
• Cybersecurity for AI systems: AI introduces unique attack surfaces including model poisoning, adversarial inputs and data breaches. Companies building AI-specific security tools will play a critical role in safeguarding businesses deploying AI.
  ‍
• Deepfake detection: We need to address the rise of synthetic media and its implications on trust and authenticity. Already we’ve seen deep fake videos get past selfie-based liveness tests, and the cat-and-mouse game will start to run even faster.
  ‍
• Talent development: The idea that “prompt engineer” would be a career or a college class was initially touted as a joke, but look at the rise of Cursor – the days of “programming in English” are already here. Moreover, it continues to amaze how variable outputs are to seemingly subtle changes in prompts. Companies offering specialized training programs, bootcamps and upskilling platforms will help enterprises transition to AI-driven workflows.

The future of 'inverted AI'

We don’t yet know all the ways AI will reshape commerce and culture. Looking at history, we know that the results may be unpredictable and unexpected. For example, electricity didn’t really improve manufacturing until it changed how factories were designed. Inverted AI companies are the powerful latent layer of this revolution. They provide the scaffolding for a world where AI is a ubiquitous force.
‍
As investors, identifying these companies requires a broader lens – one that looks beyond the technology itself to the ripple effects it creates. The AI revolution isn’t just about integrating ChatGPT or raising billions for the next foundation model. It’s about understanding and investing in the ecosystem that will thrive because of them.

‍

By QED Investors Partner Amias Gerety and Moxxie Ventures' Prateek Joshi.