​Toronto Tech Week is a citywide celebration of the people building what’s next. From May 26–29, 2026, founders, investors, and builders come together for hundreds of community-led events across Toronto, connecting tens of thousands of people around Canadian tech. Torontotechweek.com.

​The Rise Is Real. So Is the Risk.

​AI-generated code is already a majority of what's being committed in a lot of repos. The same is happening to reports, analyses, compliance outputs, and customer-facing decisions. That shift is real, it's accelerating, and most teams are shipping it without any meaningful governance layer underneath. The gap isn't capability — the tools are remarkable. The gap is control. Agentic workflows now span dozens of tool calls, cross API boundaries, trigger real-world side effects, and execute multi-step plans that no single human is watching end-to-end. Prompt injections, unintended scope creep, unauthorized actions, and opaque model-switching aren't edge cases anymore — they're failure modes showing up in production systems at enterprise scale. And as the agents get more capable, the blast radius of a bad run gets larger.

What We're Building:

🤖 IBM Bob (cherio, if you got the reference) — Governance-First, Multi-Model Coding Agent
Bob is not another Copilot wrapper. It's a coding agent built from the ground up with auditability, policy enforcement, and multi-model flexibility at its core. Bob tracks which model made which decision, why, and under what constraints — giving teams the transparency they need to actually trust AI-generated code in regulated environments. It works across heterogeneous LLM backends, so you're not locked into a single model provider making decisions your compliance team can't see.

🧠 Agentic Control Plane — Orchestration with Guardrails
When you're running multiple agents across complex enterprise workflows, you need more than an orchestration layer — you need control. The Agentic Control Plane manages agent lifecycles, enforces policy and permission boundaries, maintains observability across multi-agent pipelines, and gives teams a single plane of glass to audit what's running, what it touched, and what it decided. Built for the scale and compliance demands of financial services, insurance, and other regulated industries.

The Agenda:

Doors Open + Networking — Meet the IBM team and fellow builders over drinks and bites
Talk 1: The Governance Gap — Why most agentic systems are one prompt injection away from a bad day, and what the architecture should actually look like
Talk 2: IBM Bob, Under the Hood — A technical deep-dive into how a governance-first coding agent is designed, what tradeoffs were made, and what it looks like in a real enterprise deployment
Talk 3: The Agentic Control Plane — Orchestrating multi-agent workflows at scale with real observability, policy enforcement, and auditability baked in


Who Should Come
- Software engineers and architects building or evaluating agentic systems
- ML/AI engineers thinking about production reliability and safety
-Platform and DevOps engineers responsible for what runs in their environment
-Technical leads in financial services, insurance, or any regulated vertical
-Anyone who's asked the question: "How do I actually trust what this agent is doing?"

Space is limited. Registration subject to review.