Conference Presentation, Fireside Chat
Building AI That Gets Work Done | Gautier Cloix, H Company | RAISE Summit 2026
- Strategic Pivot: Gautier (CEO) transitioned from deploying AI at Palantir in the US to leading H in France a year ago, addressing the "wall" enterprises hit when legacy systems, missing APIs, and unclean data prevented traditional AI integration.
- Third-Generation AI Definition: H's core product, "computer use," represents the third generation of AI deployment, designed to operate directly on any software interface without requiring code changes, API integration, or months of engineering preparation.
- Unlike first-generation chatbots (efficiency/summary) or second-generation agents (which struggle with legacy systems lacking APIs), H's solution interacts with messy, pre-90s software environments by mimicking human actions on screens.
- This paradigm shift reduces the time from deployment to measurable Return on Investment (ROI) from months to a few weeks.
- Deployment Methodology: The company utilizes "forward deployed engineers" to implement solutions on the ground, prioritizing speed and adaptability over theoretical system perfection.
- Proven Use Cases & ROI:
- Construction Sector: Deploying swarms of agents in procurement to negotiate pricing on the "long tail" of purchases (e.g., light bulbs, cables) rather than major commodities.
- Result: Achieved a 4–5% reduction in purchasing costs, translating to hundreds of millions of dollars in savings for large enterprises within weeks.
- Banking & Insurance: Streamlining complex processes such as new corporate customer onboarding and claims handling for damaged goods.
- Broad Customer Base: Securing major French enterprise clients including BNP Paribas, FDG (the largest lottery group in France), and other large caps.
- Construction Sector: Deploying swarms of agents in procurement to negotiate pricing on the "long tail" of purchases (e.g., light bulbs, cables) rather than major commodities.
- Strategic Value Proposition: H aims to transform companies by removing the "execution layer" (mouse, keyboard, screen interaction), allowing human capital to focus entirely on "intention" and strategy.
- The company explicitly targets use cases with measurable ROI on the balance sheet or P&L, rejecting adoption metrics like token usage or number of deployed solutions as primary KPIs.
- CEO Gautier warns clients to demand ROI within three months, otherwise, they should replace the vendor.
- European Competitive Advantage: H positions itself to win globally by focusing on niche, high-efficiency applications rather than competing in the "arms race" for massive, expensive frontier models.
- Cost Efficiency: H's models cost approximately $3 per million tokens compared to competitors like Anthropic at $125 per million tokens, resulting in 40x lower execution costs and significantly reduced CO2 emissions.
- Market Performance: H achieved an 80% score on the OS World Verified benchmark (the standard for computer use), outperforming OpenAI, Anthropic, and Chinese models.
- Ecosystem Integration: The company is part of Nvidia's Nemotron Coalition; leveraging Nvidia's Nemotron updates allows H to increase computer use performance by 20–30% within three days of a new release.
- Model Development: While H utilizes open-source models (including Nemotron and French models) for foundational concepts, the company builds its own models from scratch, deriving value primarily from proprietary reinforcement learning, post-training data, and specific environment examples.
- Future Outlook (12-Month Horizon):
- The company aims to shift industry discourse from "AI capabilities" to "AI-enabled outcomes," where repetitive tasks are fully automated.
- With the execution layer removed, H expects to see a surge in business ideas previously deemed impossible to implement due to technical complexity, leading to unprecedented organizational transformation.
- Growth trajectory: The company anticipates significantly higher Annual Recurring Revenue (ARR) by the next conference, driven by the scaling of strategic assets rather than just efficiency gains.