Past event

AI Agentic Architectures

Monday, May 11, 2026

📍 Kendall Square, Cambridge MA

Two practitioners on the production engineering of AI agents. Ray Liao (Inkbox) on rethinking identity and trust when agents self-register. Michael R. Schulte (HBS) on the three questions every AI deployment dies on, with a live deploy into a real codebase.

RSVP on Luma See agenda
Speakers

Who shared what

Ray Liao
Ray Liao
Cofounder · Inkbox
Talk

Tradeoffs and Learnings with AI Agent Self-Signups

As AI agents increasingly become first-class actors on the internet, every existing auth model breaks down, they all assume a logged-in human at the start of the flow, but agents now arrive at endpoints before any human does. At Inkbox, Ray and his team recently enabled agent self-registration, which forced them to rethink identity, trust, and permissions from the ground up. A lesson from one of the first production teams solving this problem.

Speaker profile →
Michael R. Schulte
Michael R. Schulte
AI Builder · Harvard Business School
Talk

The Greenfield Delusion: Shipping AI Inside Real Organizations

Everyone can now build impressive AI demos in a weekend. Almost none of that work ever makes it into production at a real organization. This session walks through the three questions every deployment dies on, then puts the framework to work: we'll deploy an AI agent into an existing codebase in real time, not a blank folder. The gap between demo and production, made concrete.

Slides ↗
Speaker profile →
Agenda
  1. Doors open · informal mingling
  2. Welcome, Kranthi Manchikanti
  3. Talk 1 · Ray Liao, AI Agent Self-Signups
  4. Talk 2 · Michael R. Schulte, The Greenfield Delusion
  5. Open discussion · networking
  6. Close
Platform partner
Hackersquad Platform Partner
Recap

What happened

A full room in Kendall Square for two production stories on AI agents. Ray Liao walked through what breaks when agents register themselves before any human is in the loop, and Michael R. Schulte put the demo-to-production gap on screen with a live deploy into a real codebase. Photos below.

  • Ray Liao Every existing auth model assumes a logged-in human at the start of the flow. Agents arrive at endpoints before any human does, which breaks identity and trust from the ground up.
  • Michael R. Schulte Almost no weekend AI demo survives contact with a real organization. The gap shows up as three questions every deployment dies on, made concrete with a live deploy into an existing codebase.

Scroll for more, tap a photo to enlarge.

Subscribe

Missed it? Get the next one.

Subscribers get the recap, slides, and the next Boston event before the public Luma drop.