Insights

Notes from the agent trenches.

How we build, what works, what doesn't. Practical guides, post-mortems, and case studies.

GUIDE

Choosing the right LLM for production agents

Why model selection is the most important decision in your agent architecture. We benchmark against Artificial Analysis for every build.

8 min read Read →
CASE STUDY

How we cut a fintech client's processing time by 40%

A multi-agent loan pre-screening system: document extraction, risk scoring, and routing — built in 3 sprints.

12 min read Read →
TECHNICAL

MCP servers: the protocol that makes agents useful

Why Model Context Protocol is the missing piece for tool-using agents, and how to build your first MCP server in an afternoon.

10 min read Read →
GUIDE

Running agents fully local with Ollama

When cloud LLMs aren't the right answer. Privacy, cost, and latency benefits of running Llama, Mistral, and Gemma locally.

15 min read Read →
SECURITY

Agent security: prompt injection in production

A real attack we caught in client production. What we learned, what we shipped, and how to defend against it.

9 min read Read →
METHODOLOGY

Why we ship in 1-week sprints (not 3-month projects)

How our methodology produces working production code every Friday — and why that matters for your team.

7 min read Read →
TUTORIAL

Building your first Hermes skill in 30 minutes

Step-by-step: from zero to a working skill that talks to your CRM. Includes the SKILL.md template we use internally.

20 min read Read →
INDUSTRY

The state of South African AI in 2026

Where SA businesses are winning with AI agents, where they're struggling, and what the next 12 months will bring.

11 min read Read →
ANNOUNCEMENT

OpenRouter inference credits now included in every course

Every student gets a starter budget to experiment with multi-model setups. Here's how to make it last.

3 min read Read →

Want these in your inbox?

One email per month. No spam. Practical AI agent content only.