scripod.com
Claude Code’s Next Era — Thariq Shihipar, Anthropic

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Claude Code’s Next Era — Thariq Shihipar, Anthropic

Latent Space: The AI Engineer Podcast

19 HOURS AGO
Latent Space: The AI Engineer Podcast

Latent Space: The AI Engineer Podcast

19 HOURS AGO
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

We are excited to have Anthropic share their latest AI x Finance work at AI Engineer New York, coming up in 2 weeks! In case you’ve been under a rock, here’s a non-exhaustive list of what Anthropic has been shipping since closing the largest fundraise of ...

Highlights

Anthropic’s Thariq Shihipar explains how Claude Code is evolving from a coding assistant into a customizable, collaborative agent system—and why increasingly capable agents demand a new approach to prompting, harness design, and security.
01:21
Agentic coding became the default in less than a year
06:06
Unknowns remain a core agentic coding skill
10:29
The future splits intelligence from execution
16:02
Prompting is a high-skill discipline
21:20
Context before execution
24:11
The smartest model may also be the cheapest
28:37
Start without Claude.md
37:53
Claude Mods preview mutable software
44:55
Harnesses go out of date quickly
52:10
Claude Tag as an organizational harness
57:20
The agents communicated through cache names
59:21
Agents hacked the scorer, not the answers
1:03:39
Smarter models can hack constraints
1:12:06
Engineers now have two jobs
1:14:41
Independent evaluators can make frontier development accountable
1:23:17
Long-running agents change the security problem
1:28:20
AI safety becomes an engineering job

Chapters

Introduction
00:00
Ask User Question and the Future of Agent Interfaces
04:12
Artifacts, Projects, and Multiplayer Agents
08:29
Prompting as the Core Claude Code Skill
15:37
Context, Effort, and Smarter Model Usage
21:52
Is Claude.md Going Away?
28:10
Claude Mods: Customizing the Claude Code Harness
32:49
Model Routing and the Rise of Mutable Software
36:35
The Bitter Lesson of Harness Engineering
44:40
Claude Tag as an Organizational Harness
50:49
Pacing the Frontier and Autonomous Agent Security
55:59
Agents Hack Hugging Face for the Scorer
58:22
What Happens When Agents Need More Compute?
1:05:34
AI Coding Is Changing Faster Than Engineers Can Keep Up
1:10:32
Probes, Fallbacks, Interpretability, and Auto Mode
1:17:17
AI Risk, p(doom), and Closing Thoughts
1:28:32

Transcript

Swyx: We're here in the studio with our friend Thariq from Enthropic. And I guess generally the Cloud Code, there's so much sort of merging of boundaries. And you've been so on top of everything since you joined Enthropic. You have been early to Cloud Code...