CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

ainativedev/aidevcon-2026-nyc

AI Native DevCon 2026 New York — practical agent skills grounded in conference talks

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Overview
Quality
Evals
Security
Files

outline.mdskills/talk-chanezon-gordon-agent-experience-evaluation/references/

Talk navigation

Talk: AI Agents Are Your New Users
Speakers: Patrick Chanezon and Jay Gordon
Event: AI Native DevCon New York 2026
Recording: https://www.youtube.com/watch?v=I81RrZcQaT8

This index is a paraphrase of the supplied corrected subtitles. The accompanying transcript retains the original cue numbers and timestamps, so a reference remains stable even where an interlude has been removed. There are no reliable per-cue speaker labels; introductions and handovers provide context, not a complete diarization.

TopicTimestampTranscript cues
Agents as product users; learning from traces00:23–02:187–31
Old SharePoint examples, literal documentation use, IDE diagnostics02:22–04:1832–57
Propensity versus efficacy05:34–06:5973–90
Evaluation methodology: scenario, criteria, profile, gates08:02–10:36105–139
Baseline, added context, lift or drag, cost10:48–11:47142–156
Environment effects, repeated runs, trajectories11:54–12:52157–170
Promotional interlude omitted from skill source12:59–13:35171–181
Scope implementation; automatic improvement is a future goal13:40–15:07182–203
GitHub bug-fix comparison setup and controlled simulator15:11–17:33204–235
GitHub results: 15 runs per arm, success, time and tokens18:29–19:59247–263
Cosmos DB propensity and efficacy framing21:03–22:49280–304
Agent kit, scale, comparison profiles, judge and runtime checks22:51–25:42305–347
Repeated code snapshots; compare generated code, kit and docs25:46–28:01348–382
Python leaderboard, C# examples, criterion-level improvements28:04–29:45383–405
Recurrent issues and improvements; unresolved count scopes29:52–31:08407–425
Open-source tools and applicability to product teams31:46–32:29435–443

Useful short quotations

Verify surrounding cues before using a quotation to support a claim.

  • Cue 39, 02:48: "They, they take your docs very, very literally."
  • Cue 137, 10:27: "But then it wouldn't compile or wouldn't deploy."
  • Cue 146, 11:06: "And then you measure the lift or drag."
  • Cue 165, 12:34: "Then you need to make many runs."
  • Cue 202, 15:00, excerpt: "We're not there yet."
  • Cue 393, 28:43: "Matter of fact it put C# in."

Source limits

The subtitles do not show the demo's exact commands, repository addresses, slides, or underlying run data. Product versions, prices, release plans, and capabilities mentioned in the talk describe that presentation, not necessarily current behavior. References to a language count around 26:14 are inconsistent with the spoken list; do not repeat a precise count. The offset result and issue counts need the qualifications in SKILL.md.

skills

talk-chanezon-gordon-agent-experience-evaluation

tile.json