QCon Celebrates 20th Anniversary with Focus on Production AI and Resilience

Moving beyond Chatbots to Agentic Systems

QCon’s 20th anniversary conferences will mark a significant shift in the industry’s approach to AI, with a focus on moving from experimental use of LLMs to “agentic systems” that integrate models, tools, and workflows. Senior engineers like Hien Luu, Sr. engineering manager at Zoox, will lead tracks on AI Engineering, discussing the rigorous testing and validation required when probabilistic models enter production traffic.

Why This Matters

The move to production AI requires a fundamental shift in how we approach system design, with a focus on handling non-determinism, observability, and security in critical paths. This is a significant challenge, with many organizations struggling to validate the ROI of their AI investments, and the cost of failure can be substantial, with some estimates suggesting that the average cost of an AI project failure is over $1 million.

Key Insights

• 75% of AI projects fail to deliver expected results, according to a report by Gartner in 2022
• Agentic systems require a new approach to system design, with a focus on integration and workflow, as seen in the work of companies like Zoox and UBS
• Tools like Temporal are being used by companies like Stripe and Coinbase to improve the reliability and scalability of their AI systems

Working Example

# Example of a simple agentic system using Python and the Temporal framework
import temporal

# Define a workflow that integrates multiple models and tools
@temporal.workflow
def my_workflow():
    # Call a machine learning model to make a prediction
    prediction = my_model.predict()
    # Use the prediction to make a decision
    decision = my_decision_maker.decide(prediction)
    # Take action based on the decision
    my_action_taker.take_action(decision)

# Run the workflow
temporal.run(my_workflow)

Practical Applications

• Use Case: Companies like Zoox and UBS are using agentic systems to improve the reliability and scalability of their AI systems, with a focus on integration and workflow.
• Pitfall: One common anti-pattern is to focus too much on the technology, without considering the broader system design and workflow, which can lead to significant costs and delays.

References:

• https://www.infoq.com/news/2026/02/qcon-previews-20th-anniversary/
• https://www.gartner.com/en/newsroom/press-releases/2022-02-15-gartner-says-75-percent-of-ai-projects-will-not-achievestyleType