AppsFlyer, a company with over 1,300 employees, took an unconventional approach to AI adoption by temporarily pausing its usual business objectives and making AI education the central focus for all staff. For a full month, employees were not asked to fit AI learning around their existing work; learning and experimenting with AI became the work itself. This initiative, shared by Chief Product Officer Barak Witkowsky in a recent podcast, shows how a large organization can implement AI transformation from the ground up.
The Challenges of Conventional AI Integration
Many companies treat AI as a top-down initiative. Executives define key performance indicators, design processes, and instruct teams to adopt AI solutions. While leadership may be optimistic about the potential of AI, employees often experience uncertainty and hesitation. Witkowsky refers to this as the "AI hesitancy gap," where employees feel unsure about their own AI skills, even as the organization pushes for adoption. This dynamic can create imposter syndrome and slow meaningful engagement with AI tools.
Combining Leadership Vision with Team-Level Innovation
AppsFlyer’s strategy combined clear executive guidance with bottom-up innovation. The company introduced AI-driven objectives known as ATOMs while simultaneously empowering employees across all departments to explore and experiment with AI solutions. Internal hackathons, recognition of AI champions, and dedicated time for experimentation allowed employees to contribute ideas and insights that directly influenced the company’s AI capabilities. According to Witkowsky, much of AppsFlyer’s AI knowledge now comes from its own workforce rather than external consultants or vendors.
CEO Oren Kaniel was instrumental in backing the initiative, committing the company to a model where AI competence is broadly distributed rather than centralized. Witkowsky recalls initial skepticism among staff, but the approach has now reshaped the company culture.
Evolving the Product Alongside AI Adoption
Alongside internal transformation, AppsFlyer evolved its product offerings. Originally an attribution platform, the company now positions itself as a modern marketing cloud, integrating AI agents that support marketers with tasks such as audience analysis, campaign strategy, and creative recommendations. This evolution responds to growing pressures on marketing teams to demonstrate measurable results, manage complex omnichannel strategies, and adopt AI tools capable of autonomous workflows.
The AI agents function as extensions of the marketing team, enabling clients to experiment with data-driven strategies and improve campaign efficiency. By embedding AI into its core product offerings, AppsFlyer simultaneously helps clients transition toward AI-first operations.
Fostering Trust in AI Insights
One of the central challenges in AI adoption is ensuring that users trust AI-driven decisions. AppsFlyer measures confidence through user behavior rather than vanity metrics. For example, clients are observed making rapid, high-stakes decisions, reallocating ad spend, or adjusting campaigns after only a few days based on AI analysis. Witkowsky explains that these actions are a more reliable indicator of confidence than surveys or usage statistics. The company supports clients in gradually adopting autonomous marketing practices, reinforcing trust in AI agents as part of the decision-making process.
Implications for Marketing Professionals
With AI handling more operational tasks, performance marketers are expected to shift toward strategic and creative responsibilities. Witkowsky envisions a model where marketers manage AI agents, which in turn manage other agents, enabling teams to scale creative and strategic output. Rather than reducing the role of marketers, AI extends their capabilities and allows them to focus on higher-level planning and user engagement. Those who adapt to this environment are likely to remain competitive, while those who resist risk falling behind.
Lessons for Companies Considering AI Transformation
AppsFlyer’s experience offers several takeaways for organizations looking to implement AI: it demonstrates the value of making AI learning the core of employees’ work, distributing AI knowledge broadly, and pairing executive guidance with grassroots innovation. The investment is significant, and the ROI is still being measured, but the cultural impact is evident. By prioritizing AI skill-building across the workforce, AppsFlyer has positioned both its employees and clients for long-term adaptation in the AI era.
Source: Deconstructor of Fun
Frequently Asked Questions (FAQs)
What does it mean to have an AI-first culture?
An AI-first culture prioritizes AI knowledge and adoption across all employees rather than limiting it to specific teams or executives. At AppsFlyer, every employee was trained to understand and use AI tools as part of their regular work.
How did AppsFlyer train 1,300 employees in AI?
The company paused standard business objectives for a month and provided a structured AI builder program. Employees participated in hackathons, received guidance from internal champions, and were encouraged to experiment with AI as part of their daily tasks.
Why did AppsFlyer pause regular business goals for AI training?
The company aimed to remove competing priorities so employees could fully focus on understanding AI. Making AI the central work ensured engagement and allowed the organization to distribute knowledge across all departments.
How has AppsFlyer’s product changed with AI adoption?
AppsFlyer transformed its product from an attribution platform into a modern marketing cloud. The platform now includes AI agents that assist marketers with analysis, campaign management, and strategic recommendations.
What is the long-term benefit of an AI-first workforce?
An AI-first workforce can innovate more efficiently, improve internal processes, and help clients adopt AI solutions effectively. Employees gain skills that enhance productivity, while the organization benefits from distributed AI knowledge and faster adaptation to new technologies.
