Six Sessions at QCon AI Boston 2026 That Take Productionizing AI Seriously

2026-05-21 · Source: InfoQ · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Advanced, short

Summary

QCon AI Boston 2026, scheduled for June 1–2 at Boston University, will feature over 40 sessions, with six highlighted sessions focusing on the practical challenges of productionizing AI. Martin Spier from OpenAI will discuss optimizing ChatGPT performance, emphasizing that latency issues extend beyond GPUs to client work, context assembly, and agent-operated investigation. Ajay Prakash of LinkedIn will present CAPT, an MCP-based context layer for AI agents, detailing its organizational rollout and achieving 70% faster issue triage. Vinoth Govindarajan from OpenAI will highlight the "agent harness" for reliability, covering control planes and auditability using OpenClaw. Susan Chang from Elastic will share insights on building reusable evaluation frameworks for agentic AI products, drawing from two years of production experience. DoorDash's Siddharth Kodwani and Swaroop Chitlur will explain their GenAI platform, including an LLM Gateway and Agentic Gateway, to consolidate common LLM infrastructure. Finally, Andrew Swerdlow of Roblox will address engineering an autonomous SDLC at scale, introducing Exemplar Alignment for agent-driven codebase migrations and maintenance.

Key takeaway

For AI Engineers and MLOps teams deploying agentic AI, recognize that production readiness demands robust system engineering beyond model performance. Your focus should extend to multi-layer latency analysis, building agent harnesses with control planes and auditability, and establishing reusable evaluation frameworks. Consider consolidating common LLM infrastructure into shared platforms to accelerate development and ensure consistent quality, especially when scaling autonomous SDLC processes.

Key insights

The core challenge in productionizing AI agents lies in robust system engineering beyond model capabilities.

Principles

• AI application latency is a multi-layer system problem.
• Agent reliability stems from external control planes and auditability.
• Shared GenAI infrastructure prevents redundant development.

Method

LinkedIn built CAPT, an MCP-based context layer, to ground AI agents in internal services and workflows. OpenAI moves performance engineering to agent-operated investigation using telemetry. Roblox uses Exemplar Alignment for agent-driven SDLC redesign.

In practice

• Implement multi-layer performance monitoring for AI apps.
• Design agent "harnesses" with control planes and audit trails.
• Consolidate LLM plumbing into shared platform components.

Topics

• AI Productionization
• Agentic AI
• Performance Engineering
• Context Engineering
• Evaluation Frameworks
• GenAI Platforms
• Autonomous SDLC

Best for: AI Architect, AI Product Manager, AI Engineer, Machine Learning Engineer, MLOps Engineer

Related on AIssential

• QCon AI Boston’s Early Program Focuses on the Engineering Work Behind Production AI — Production AI demands robust engineering for reliability, explainability, security, and scalability beyond initial demos.
• Why AI Agents Still Fail in Production (and Why Better Models Won’t Fix It) — AI agent unreliability stems from multiplicative system errors and architectural flaws, not primarily from model intelligence.
• AI latency is a business risk. Here’s how to manage it — Enterprise AI latency is a system-level business constraint, not merely a model-tuning problem.
• QCon Previews 20th Anniversary Conferences: Production AI, Resilience, and Staff+ Engineering — QCon's 20th anniversary conferences focus on practical production challenges in AI, system resilience, and senior engineering roles.
• Evaluating AI agents in production: A practical framework — Effective AI agent evaluation requires a multi-layered, continuous approach adapting to non-deterministic LLMs and complex architectures.
• The TechBeat: Why "Build an AI Agent" Is the Wrong Starting Point for AI Systems (4/16/2026) — Production-ready AI systems demand robust architecture, integrated design, and multi-model approaches, not just advanced prompting.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by InfoQ.