Dense Slides: Deterministic Infra for Non-Deterministic AI Agents - Nishant Gupta, Meta Superintelligence Labs

Source Video

Deterministic Infra for Non-Deterministic AI Agents - Nishant Gupta, Meta Superintelligence Labs

Method

This deck is slide-only. The existing captured video frame set supplies candidate frames, then local OpenCV rejects sponsor/title/speaker-only frames, crops visible slide surfaces, deduplicates, and saves the cropped slide images.

Cropped Visible Slides

Recreated text/layout view: open HTML recreation
AI slide classifier: title_card confidence 0.99
Text source: agent_vision.

Slide text:

BUILDING DETERMINISTIC INFRASTRUCTURE FOR NON-DETERMINISTIC AI AGENTS

THE EMERGING CONTROL PLANE FOR AUTONOMOUS AI SYSTEMS

Nishant Gupta

Tech Lead @ Meta

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Dense multi-column comparison slide with small text labels.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.98
Text source: none.
OCR decision: ready — Architecture slide with layered diagram and small labels.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Dense failure-tree diagram with many labels.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Process diagram with step chain and chart labels.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — System architecture slide with labeled control boundary and filters.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Dense control-plane architecture slide with multiple labeled subsystems.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Timeline and observability slide with dense track labels.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.98
Text source: none.
OCR decision: ready — Diagram slide with multiple labels and small state text; OCR will likely read the embedded annotations more reliably than a quick visual pass.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Routing diagram with several labels and smaller annotations; OCR is appropriate for the node labels and path text.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Chart slide with axis text, labels, and small annotations; OCR is the right capture method for the graph text.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: agent_vision.
OCR decision: ready — Table-like mapping slide with multiple rows and small relational labels; OCR will capture the pairings more reliably than manual transcription here.

Slide text:

The Reliability Rosetta Stone.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: none.
OCR decision: ready — Chart/trajectory slide with several labels, dates, and axis annotations; OCR is better suited than manual reading for the full figure text.

Recreated text/layout view: open HTML recreation
AI slide classifier: content_slide confidence 0.99
Text source: agent_vision.
OCR decision: ready — Dense slide with a quote panel and multiple takeaway bullets; OCR is appropriate for the smaller body text and sidebar copy.

Slide text:

AI agents are distributed systems. Treat them accordingly.

Classification audit: raw/sources/slide-ai-classification/dense/APh1Vx0oLmQ/audit.json