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

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

outline.mdskills/talk-braun-context-graphs/references/

Source outline

Source: Alan Braun, Models Alone Aren't Enough: Engineering Context as a Graph for Human + Agent Teams, AI Native DevCon New York 2026. Recording. All anchors refer to original numbered cues in the transcript. Short quotations below preserve the supplied transcription, including spoken wording.

TopicSource anchorExact short quoteApplication and scope
Outcomes framing00:01:49–00:04:03, cues 28–59“AI outcomes equals context times intelligence.”The equation is a conceptual framing, not a quantitative law.
Workflow productivity00:04:06–00:05:20, cues 60–77“planning, with building or with reviewing and with shipping?”The scope extends beyond coding.
Permissions00:05:45–00:07:25, cues 84–109“the same permissions that you do as a human”Agent access must respect caller context.
Downstream pressure00:07:30–00:09:17, cues 110–140“a good layer optimizes for every stage”Faster execution can increase downstream burden.
Connected relationships00:09:35–00:10:35, cues 145–160“Connected is the key word.”People, knowledge, and project goals inform delivery.
Three questions00:12:32–00:13:49, cues 191–209“why is this code like this?”Trace rationale, reviewers, and impacts across sources.
Seven layers00:14:22–00:15:31, cues 219–238“There are seven layers of context”The source includes knowledge, communication, work, code, people, assets, and data.
Tool bloat00:16:22–00:18:05, cues 250–278“working backwards from the jobs to be”Task-oriented retrieval helps avoid tool and context bloat.
Reference then detail00:18:09–00:18:42, cues 279–288“the right references for what we need for this task.”Use connected references to target detail retrieval.
Planning demo00:19:26–00:24:57, cues 300–377“just the technical plan.”The demonstration compares planning, not shipped implementation.
Reported results00:25:01–00:25:28, cues 378–384“up to 48% fewer tokens”Improvement numbers are reported test outcomes with limited methodology.
Evaluation00:26:41–00:27:25, cues 402–414“Take five prompts”The speaker suggests expected outcomes and blinded comparison.
Small starting scope00:27:28–00:29:20, cues 415–444“Don't do 15. Connect 3 to 5. Start small.”Start with linked work, ownership, decisions, and a small source set.

Source limitations

  • The State of Teams figures and performance improvements are reported results; the transcript does not supply full methodology or establish universal gains.
  • The demo introduction describes a stream-hub feature, then narration describes a Jira Slack digest. The baseline is also characterized differently in the recap; preserve these ambiguities.
  • Teamwork Graph, MCP, CLI counts, and product capabilities are talk-time descriptions, not current documentation or executable integration guidance.
  • The context-times-intelligence equation is a framing, not a measured mathematical model.
  • Promotional interlude cues 289–299 are omitted from the bundled transcript; the remaining cue numbers retain their source positions.
  • This is a spoken transcript; unseen slides, linked repositories, and third-party research were not independently inspected. No outcome beyond the described talk is established.

skills

tile.json