aggregate intellect

Cohort 1

Cohort 2409

Cohort-2501

Cohort-2505

Cohort-2507

2510

2601

2603

2605

This document serves as a guide for developing your agentic product. It is designed to help you systematically define your product's goals, requirements, and implementation strategy. By working through each section, you will refine your product idea, identify potential challenges, and develop a comprehensive plan for building and deploying your solution.

Agents Playbook v2506

This document serves as a guide for developing your agentic product. It is designed to help you systematically define your product's goals, requirements, and implementation strategy. By working through each section, you will refine your product idea, identify potential challenges, and develop a comprehensive plan for building and deploying your solution.

Agents Playbook v2504

Are you interested in building LLM Agent Products, but unsure where to begin, or if you have the necessary background knowledge? We have curated this learning path using 70+ hours of (mostly) free resources that you can use to:<ol><li>Get started on your journey of building LLM Agent Products</li><li>Prepare to get the most of this course or similar ones, and</li><li>Evaluate areas where there's a gap in your knowledge</li></ol> After entering your email, you will be directed to a Miro Board. Contact maven@ai.science if you have any issues with this resource.

LLM AGENTS - Learning Path

This isn't just another course. You'll walk away with tangible assets that advance your career and concrete skills that set you apart.

Founder at Aggregate Intellect

Amir Feizpour, PhD

10+ years in AI and Product Development

Agentic Buildcamp - Build AI Agents Using AI Agents (Claude Code)

Photonics Researcher

Product Manager

Software Developer

Game Developer

Director, Risk Models & Data Science

Manager, Data Scientist

Dev Lead

Researcher

founder of my company

You're already trying to build something. Agentic Buildcamp is where pragmatic builders - developers, founders, and operators - ship a real AI agent using the KnowledgeOps methodology: 12+ guided workflows that take you from problem framing to production, delivered through Sherpa-B, an MCP server that connects to your CLI coding agent (eg. Claude Code). You learn by building, using the same agentic process you're learning to build with.Note 1: Join my next <a target="_blank" rel="noopener noreferrer nofollow" href="https://calendly.com/amirfzpr/agents-office-hour">Agents Office Hour</a> and ask about this program.Note 2: If you are interested in a mentorship version of this program, reach out to <a target="_blank" rel="noopener noreferrer nofollow" href="https://mailto:bootcamp@ai.science">bootcamp@ai.science</a>🟢 WHAT WILL I BE ABLE TO BUILD?You’ll learn to design, build, evaluate, and deploy agentic systems, and integrate required tools. By the end of the program, you'll have a working application ready to showcase in a public demo.▶️ <a target="_blank" rel="noopener noreferrer nofollow" href="https://tinyurl.com/llm-agent-demos">Playlist of previous public demos</a> 🟢 WHAT SHOULD I EXPECT TO HAPPEN DURING?This is a 3-sprint product development + 1 week of prep for the public demo.🖌️ SPRINT 1 - DESIGN (wk 1 &amp; 2)🔨 SPRINT 2 - DEVELOP (wk 3 &amp; 4)🚢 SPRINT 3 - DEPLOY (wk 5 &amp; 6)🎭 PUBLIC DEMO (wk 7)Sprint demos will be live and interactive.

Self-directed program where pragmatic builders ship real AI agents in 7 weeks.

Ship a Real Agentic AI Application

Experience Learning via Building with KnowledgeOps Methodology

Obtain Assets and Artifacts that Will Super-charge your Building

Most participants fail because they're not there to build. Outside of that we don't expect you to bring much prior experience 

Lots of enthusiasm to build

You will use a CLI tool (eg. Claude Code) as the primary way to learn and build (we provide resources for this)

Claude Code

Since we use coding agents heavily, you need good version control set up (we provide resources for this)

Biz or Tech Entrepreneurs - You want to build an agentic MVP to test the feasibility of your startup idea.

Hands-on Product or Operations Managers - You want to build an agentic MVP to explore a new feature in your existing product / process

Developers, Data Engineers, Data Scientists - You want to build an agentic MVP to demonstrate your skills for career progression purposes.

Building AI Agents: Bootcamp 2605 Public Demo

Amir Feizpour

Founder @ Aggregate Intellect

What agentic systems can be built in 7 weeks

Skills necessary to build agentic systems

Tool stack involved in building agentic systems

How our 7-week program helps you build an agentic system

How you can go from a vague idea to a working MVP

Build Agentic Apps Using Claude Code & Bootcamp Info Session

Assess if agentic approach is appropriate for your use case.

Design agentic workflow for your use case.

Implement your agentic workflow using Claude

Leave with some workflow files you can immediately use

Learn more about our 7-week program to help you build

Software Factory 101

Owen Zanzal

Principal Cloud and Agentic Ops Engineer @ Shoplogix

How time-series forecasting and optimization models anchor agentic decision-making in reality.

Integrating Agents with Classic ML

Lessons from deploying autonomous agents in live supply chain workflow; fail-safes and escalation logic

Operational Reliability in the Wild

Frameworks for determining when agents should act independently versus defer to human operators.

Balancing Autonomy and Oversight

Harness Engineering: How Claude Code/ Codex/ Gemini CLI work

Mohsin Iqbal

Sr. MLE @ Alethia AI

The difference between agent tools (e.g., Claude Code) and the model that powers them.

What a “Harness” Really Means

Inputs, execution loops, tool use, logging, and iteration cycles.

Core Harness Components

A minimal, notebook-based approach to structuring AI workflows.

Building Your Own Harness

Building a Gi-ralph Loop for Software Development

Joel Fernandes

Engineering Lead @ Nvidia

A clear mental model of iterative agent loops and how they differ from simpler workflows.

What a Ralph Loop Is (and Isn’t)

Key architectural patterns for building loops that don’t degrade over time.

Designing Stable Long-Running Loops

A new loop design (greatly improved ralph) aimed at improving robustness and outcome quality.

Introducing the “Giralph” Loop

How modern models can still destabilize loops—and how to detect and mitigate this.

Failure Modes in Practice

Identifying cases where simpler approaches outperform complex iterative systems.

When Not to Use Ralph Loops

Setting Up Claude Skills as an Evolving Library

Ying Wang

Fractional CTO

How to think about Claude skills as a growing system rather than isolated tools.

From One-Off Skill to Reusable Library

Practical ways to turn recurring workflows and pain points into useful skills.

Generating Skills from Real Needs

Approaches for updating, refining, and extending skills as needs evolve.

Keeping Skills Fresh Over Time

How to organize skills so they remain understandable, testable, and reusable.

Versioning and Maintenance

Techniques for shaping skills around your own workflows rather than generic use cases.

Learn directly from aggregate intellect