Keynote - Beyond the Chatbot: Generative AI’s Next Big Leap
Kalle Saarikannas — Senior GTM Lead, AI at Work & Modern Work, Microsoft Finland
This keynote explores how AI is moving beyond simple text-based interfaces into multimodal AI, AI agents and sophisticated role-specific AI solutions. We’ll dive into how the evolution of AI, including solutions like Microsoft 365 Copilot, is transforming productivity, collaboration, and the future of work itself.
"By building specific agent AI capabilities for specific tasks makes things a lot easier and also kind of reduces the hallucination."
Summary
- GPT-3.5's chat interface revolutionized AI interaction. - AI aids in data integration and presentation creation. - GitHub Copilot helps developers with code generation. - AI can transcribe meetings and assist with note-taking. - Domain-specific AI agents enhance productivity and reduce errors.
Article
Beyond chatbots: generative AI evolves into specialized agents
Kalle Saarikannas reveals AI's next phase at VERGE conference in Helsinki
At the VERGE AI Frontier conference in Helsinki, industry expert Kalle Saarikannas delivered a compelling keynote that charted the rapid evolution of generative AI and offered a glimpse into its future trajectory. Moving beyond the now-familiar chatbot interfaces, Saarikannas described how AI is transforming into specialized agents capable of orchestrating complex tasks across multiple domains.
From chat interfaces to integrated workflows
"GPT three point five became publicly available. That kind of started the whole generative AI hub that we've been living through," Saarikannas told the audience gathered at the Helsinki event on Monday.
The presentation highlighted how the chat interface has become the primary method of AI interaction, whether through Google services, Microsoft Copilot, or OpenAI's ChatGPT. However, the true power of these systems is emerging in their ability to integrate diverse data sources.
"We can interact with the different data sources in a single point of view," Saarikannas explained, demonstrating how AI can combine web research with internal documents like Excel spreadsheets to provide comprehensive insights for business decisions.
Revolutionising content creation and development
The keynote showcased practical applications of generative AI across various professional contexts. In content creation, tools like PowerPoint Copilot can draft entire presentations based on natural language prompts, which users can then refine using their existing materials.
For software development, the evolution of GitHub Copilot revealed significant improvements in code generation capabilities. What began as a simple code completion tool has evolved into a sophisticated reasoning system requiring less human guidance to complete complex programming tasks.
Addressing team collaboration challenges
Despite productivity gains at the individual level, Saarikannas acknowledged persistent challenges in team environments. Citing Microsoft's Work Trend Index, he noted that 64% of workers report lacking energy and time for their daily tasks, with the number of team meetings having tripled in recent years.
"AI could be doing the note taking, making sure that we can actually focus on the agenda and the AI would do the notes and let us be natural," Saarikannas suggested, pointing to meeting transcription and summarization as valuable applications. He also highlighted how AI-powered live translation is breaking down language barriers in international teams.
The rise of agentic AI
The most forward-looking portion of the keynote addressed the emergence of AI agents – the next evolutionary step beyond chatbots and assistants.
"With agents as well, it kind of takes it one step further," Saarikannas explained. "AI can be in the layer background doing the kind of orchestration on behalf of the user."
These agents operate with increasing autonomy in specific domains, such as retail operations where they can monitor inventory levels and sales data, or in strategy development where reasoning models draft market entry plans based on multiple data sources.
Domain-specific vertical agents
The future of AI, according to Saarikannas, lies in developing highly specialized vertical agents focused on specific industries or functions. This approach reduces information overload and minimizes the risk of AI hallucinations – instances where AI generates plausible but incorrect information.
By embedding generative AI capabilities directly into business processes, organizations can benefit from AI's efficiency without requiring direct user interaction. The technology effectively fades into the background, handling orchestration tasks while humans focus on more creative and strategic work.
As the VERGE conference continued in Helsinki, Saarikannas's keynote established a framework for understanding how generative AI is evolving beyond the novelty of chatbots toward more sophisticated, specialized systems that promise to transform productivity and collaboration across industries.
Part of VERGE | The AI Frontier: Creativity, Security & Collaboration