On day two of CES 2026, Havas CEO and Chairman Yannick Bolloré announced AVA during the company’s C Space Keynote, a proprietary AI platform that connects the agency’s 23,000 employees to multiple leading models – GPT-5, Claude, and Gemini – through a unified interface with security and compliance embedded throughout. The launch came alongside confirmation of €1 billion in total AI investment and a company-wide mandate: every employee must achieve AI certification, with 550 leaders required to complete it before attending the upcoming leadership meeting in Paris.
Havas’ approach provides a promising blueprint for organisations seeking to maintain long-term strategic control in volatile market conditions, as the company prioritises becoming AI-first as a necessity not just for competitive advantage today, but for the agency’s operational future. What emerged from the keynote was a clear example of proactive AI integration: a creative organisation architecting AI capabilities to augment rather than replace the strategic thinking and creative judgement that define its value, delivering lasting outcomes for clients.
Below, we explore Bolloré’s announcements and what they reveal about successful AI deployment at enterprise scale, a challenge that demands architecting systems that enhance operations while protecting the strategic assets that determine competitive positioning.
Building operational foundations for AI-native work
Havas’ announcements at CES extend beyond capability adoption, instead representing a recognition of the restructure necessary for AI-native workflows to deliver outcomes while preserving the human strategy required for agency value, protecting these assets as the foundation of competitive positioning.
AVA, the organisation’s newly launched unified infrastructure, enables consistent access to client data while maintaining control over security and compliance, rather than individuals selecting tools independently or routing private IP through public services. Supporting this, strategic partnerships with Verva and Eko address direct operational constraints, taking a problem-first approach to AI integration by transforming expensive and time-consuming persona research into immediate, scalable insight generation and enhancing media planning capabilities.
This targeted adoption of AI capabilities is already producing tangible results, emphasised by the company’s CEO as he confirmed clients such as Renault have seen 50% reduction in production cost by shifting to AI-enabled workflows.
However, AI integration at this scale introduces risks alongside opportunities, and Bolloré identified data privacy as potentially the most significant: if proprietary client intelligence becomes exposed through AI deployment, the competitive advantages these systems enable would be completely undermined.
Security as a strategic imperative
Bolloré stressed data privacy as one of the biggest threats facing agencies deploying AI at scale, noting: “If one day one of our agencies were to upload client data in the cloud and it were to become public, that would be a catastrophe”.
This concern reflects a challenge we observe across industries but find to be particularly acute in creative services, whereby the integration of AI creates an entirely new category of exposure points for proprietary intelligence. For advertising agencies, these assets include client strategies, audience insights, campaign performance data, and creative IP – all of which lose value completely the moment they’re no longer exclusive. The opportunity for AI to enhance these capabilities is clear, but the critical question becomes whether organisations can deploy such technology without compromising the intelligence that differentiates them in the first place.
Most organisations approaching AI deployment currently focus on capability access without adequately architecting for intelligence protection, and for those with value to lose, the risk in this approach is immediate. Organisations pursuing AI-enabled efficiency gains while inadvertently exposing the unique insights that enable premium positioning in the first place face a strategic paradox: increasing operational capability while eroding value at the same time.
Architecting for intelligence protection
Therefore, purpose-built infrastructure that maintains strict boundaries around proprietary intelligence while facilitating AI integration at scale is now a strategic imperative. Platforms like AVA represent significant progress toward IP protection, but sustained future advantage will demand embedding security across the complete AI stack, integrated throughout entire AI systems from the ground up.
This requires treating security as an architectural imperative from the outset: designing infrastructure where AI capabilities operate effectively while allowing proprietary assets to remain isolated, and building systems where security enables innovation to scale sustainably rather than constraining it.
The organisations maintaining strategic control are those designing for this tension, not choosing between capability advancement and security protection, but rather architecting systems that enable both.
Building foundations for sustained competitive control
Havas’ CES announcements demonstrated what proactive AI integration looks like: substantial investment, operational mandates, and explicit recognition of the risks alongside the opportunities.
Yet what emerged from Bolloré’s keynote, and echoed across CES discussions, is a consistent pattern: the organisations positioned to capture lasting value from AI are those treating deployment as an architectural challenge requiring systematic redesign, not simply capability adoption alone.
As the Stelia team continues to connect with partners and customers at CES this week, this is the transition we’re focused on – helping organisations to architect AI systems that enable both capability advancement and IP control over the very assets that determine competitive positioning. Havas’ approach provides one blueprint, but the broader imperative remains: success demands building foundations that support both innovation and strategic control, with governance, security, and compliance embedded from the ground up.
