How to Use /Goal to Do More With AI

Key takeaways

Slash goals represent a new primitive in AI systems, enabling long-running, autonomous task management with feedback and self-evaluation.
Effective goals are defined by clear outcomes, success evidence, constraints, boundaries, and stop conditions—these are crucial for reliable AI automation.
While originally aimed at coding tasks, slash goals are applicable to a broad range of knowledge work such as audits, research, vendor assessments, and strategic reports.
Designing good goals includes calibrating scope, specifying output artifacts, and providing well-defined success rubrics to ensure meaningful results.
Slash goals are part of a spectrum of interaction—from prompts to fully autonomous agents—so choose the approach that fits your task’s complexity and certainty.
Experimentation and sharing insights will accelerate understanding of where slash goals work best and how to improve their design.

Summary

Introduction to Slash Goals and Context in AI Development

The episode begins with an overview of the concept of slash goals, a new primitive in AI systems like Codex and Claude Code that enhances long-term, autonomous task execution. It discusses recent innovations, interaction patterns, and the way slash goals enable more parallel, autonomous work compared to traditional turn-based prompts. The segment emphasizes how these features are transforming AI’s capabilities for complex, sustained tasks.

What is Slash Goal and Why It Matters

This section explains slash goal as a fundamental shift from simple prompts to a continuous, looping process where the AI works towards specific, auditable outcomes. It highlights the properties of effective goals—durable, with clear evidence of success, and flexible enough to adapt as work progresses. Slash goals facilitate autonomous, unsupervised work streams, particularly beneficial for tasks requiring persistent context, such as audits, research, or complex evaluations.

Characteristics of Good Goals and Use Cases beyond Coding

Key characteristics of well-structured goals are outlined, including the importance of defining success criteria, constraints, boundaries, iteration policies, and stop conditions. The discussion explores how slash goals can be applied to non-coding knowledge work like market analysis, literature reviews, vendor evaluations, and audits. Examples demonstrate how structured, evidence-backed outputs (like tables, reports, or citations) are vital for assessing completion and correctness.

Designing Effective Slash Goals and Their Practical Application

Guidance is shared on crafting effective goals—emphasizing scope scope calibration, specific artifacts, and success rubrics. The episode also delves into how knowledge work tasks, such as claims audits or market analyses, can benefit from slash goals by providing clear evidence trails and structured outputs. User-defined rubrics are often necessary, especially in subjective or complex evaluation scenarios. The segment advocates experimenting with slash goals to gauge their potential across various domains.

Limitations and Spectrum of Interaction Methods in AI Workflows

The episode notes that slash goals aren’t suitable for all tasks; sometimes traditional prompts or mixed interaction patterns are more appropriate. It stresses the importance of choosing the right level of autonomy, balancing user control with system capabilities. The conversation underscores that slash goals are part of a broader spectrum of AI interaction, from simple prompts to fully autonomous agents, and encourages ongoing experimentation to determine best practices.

Final Reflections and Future Outlook

The episode concludes by emphasizing the emerging nature of slash goals, the importance of continuous innovation, and the need to share experiences and learnings. It hints at future updates and deeper exploration of non-coding use cases, encouraging listeners to experiment and adapt new primitives in their workflows to unlock AI’s full potential.

Show notes

A practical primer on /goal, the new AI primitive showing up in Codex and Claude Code. NLW explains how /goal differs from a normal prompt, why it matters for longer-running agent tasks, what makes a good goal, and how to think about using it beyond coding for audits, research, vendor reviews, market landscapes, and other knowledge work where the AI needs a clear finish line and evidence of completion.

Sign up for AI Executive Catchup: ⁠https://aiexecutivecatchup.com/⁠

Brought to you by:

KPMG – Research from KPMG and the University of Texas at Austin shows the highest-impact AI users treat AI like a reasoning partner — and those skills can be taught at scale. Learn more at ⁠⁠⁠kpmg.com/us/Sophisticated⁠⁠⁠

Section - Section turns AI investment into workforce transformation and ROI - ⁠⁠https://www.sectionai.com/⁠

Outsystems - Stop wondering how AI will change your business and start building the agents that will lead it - http://outsystems.com/

Scrunch - The AI customer experience platform - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://scrunch.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Zenflow Work - Agents for knowledge work - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://zenflow.free/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Interested in sponsoring the show? sponsors@aidailybrief.ai