---
title: "Computer Use at the Edge of the Statistical Precipice"
category: "talks"
date: "2026-06-30"
time: "11:10am-11:30am"
track: "Computer Use"
room: "Track 7"
speakers: ["Pierluca D'Oro"]
sourceLabels: ["Official conference schedule", "Public YouTube metadata"]
scheduleTrack: "Computer Use"
scheduleRoom: "Track 7"
scheduleLabels: ["Computer Use", "Track 7", "session", "confirmed"]
---
# Computer Use at the Edge of the Statistical Precipice

## Conference Context
- Date/time: 2026-06-30 · 11:10am-11:30am
- Track/room: Computer Use · Track 7
- Speaker(s): Pierluca D'Oro
- Session type/status: session · confirmed

- Track: Computer Use
- Room: Track 7
- Session type: session
- Status: confirmed

## Session Description
Evaluating Computer Use Agents (CUAs) on interactive environments is fraught with methodological pitfalls that the field has yet to systematically address. We show that a 1MB replay script that blindly executes a recorded action sequence without ever observing the screen outperforms frontier models on prominent static benchmarks, and prove that its expected success rate is exactly equal to the source agent's pass@k in deterministic environments. We trace this and other failures to two root causes: non-principled environment design (static, unsandboxed, or unreliably verified environments) and non-principled evaluation methodology (naive aggregation and misuse of pass@k for stateful UI interactions). To address the first, we propose PRISM, five design principles for CUA environments and instantiate them in DigiWorld, a benchmark of 15 realistic sandboxed mobile applications able to evaluate agents in over 3.2 million verified unique configurations. To address the second, we develop an aggregation framework that correctly accounts for the nested structure of CUA benchmarks. All together, we show that principled environment design and rigorous evaluation methodology are not optional refinements but prerequisites for meaningful CUA research.

## Media Evidence
No related AI Engineer channel video found yet.

## Evidence Graph
This evidence graph is generated from currently linked source material: official schedule text, related video pages, cached transcripts, visible slide text, dense/reconstructed slide pages, and AI slide-classification audits.

### Media Signals
No linked video, transcript, or slide source has been attached yet.

### Agent Reading Notes
Use these signals to refine the synopsis, topic links, people/company context, and method notes. If a source is a related external video rather than an exact official recording, keep it framed as supporting evidence.

## Transcript Status
No official session recording transcript was found by exact title match on the AI Engineer YouTube channel during this run.

## People
- [[pierluca-d-oro]]

## Notes
- Pending transcript synthesis when an official recording or confirmed matching video is available.

## Synthesis
### Synthesized Breakdown
# Computer Use at the Edge of the Statistical Precipice ## Conference Context - Date/time: 2026-06-30 · 11:10am-11:30am - Track/room: Computer Use · Track 7 - Speaker(s): Pierluca D'Oro - Session type/status: session · confirmed - Track: Computer Use - Room: Track 7 - Session type: session - Status: confirmed ## Session Description Evaluating Computer Use Agents (CUAs) on interactive environments is fraught with methodological pitfalls that the field has yet to systematically address. We show that a 1MB replay script that blindly executes a recorded action sequence without ever observing the screen outperforms frontier models on prominent static benchmarks, and prove that its expected success rate is exactly equal to the source agent's pass@k in deterministic environments. We trace this and other failures to two root causes: non-principled environment design (static, unsandboxed, or unreliably verified environments) and non-principled evaluation methodology (naive aggregation and misuse of pass@k for stateful UI interactions). To address the first, we propose PRISM, five design principles for CUA environments and instantiate them in DigiWorld, a benchmark of 15 realistic sandboxed mobile applications able to evaluate agents in over 3.2 million verified unique configurations.

### Speaker And Company Context
- [[pierluca-d-oro|Pierluca D'Oro]] — Founder at [[programma-labs|Programma Labs]].

### Topics Covered
- [[agentic-search]]
- [[ai-sandboxes]]

### Derived Links And Source Material

### Novel Concepts / Clever Methods
- No highlighted novel concept has been detected yet.

### Evidence Boundary
This synthesis is based on the official schedule and linked source pages. It should be revisited when exact session recordings or transcript-backed secondary sources are available.