Notes from production.
Posts on AI agents, outbound and prompt engineering.
- AI Automation
- Lead Generation
- Strategy
- Prompt Engineering
- Context Engineering
- Case Study
- Growth
- AI AutomationSeptember 17, 20265 min read
Your AI agent has a security hole nobody scoped for
Prompt injection attacks are up 340% and hit a third of deployed agents. After 4 years shipping these systems, here is what actually contains it.
- Lead GenerationSeptember 16, 20265 min read
Why most teams kill their AI SDR within three months
Between 50 and 70% of teams that deploy AI SDRs churn inside a quarter. After 4 years building outbound systems, the reason is not the technology.
- AI AutomationSeptember 15, 20265 min read
Your AI agent works in the demo and dies in production
Most AI agents pass the demo and fail the first real week. After 4 years building these systems, the problem is almost never the model.
- StrategySeptember 14, 20264 min read
AI search is eating your B2B traffic. Your site needs a new job.
Nearly 60% of searches now end without a click and Gartner expects a 25% drop in search volume. Here is how I think B2B sites have to change.
- AI AutomationSeptember 12, 20265 min read
Voice AI demos beautifully and falls apart on real calls
Most voice agents ship with 2 to 3 seconds of round trip delay and drop calls under load. After 4 years deploying these, here is what separates the ones that work.
- Lead GenerationSeptember 10, 20265 min read
Cold email reply rates dropped 15%. Better copy will not fix it.
Reply rates fell to 5.8% this year and 17% of cold emails never reach the inbox. After 4 years running outbound, the fix is infrastructure, not writing.
- StrategySeptember 8, 20265 min read
Your AI bill tripled in a week. Here is where the money goes.
Audits find 40 to 60% of enterprise LLM spend is waste. After 4 years running these systems, here are the six places the money actually leaks.
- StrategySeptember 4, 20265 min read
Your automation is not broken. It is unowned.
Most business automation fails months after launch, when an API changes and nobody is watching. After 4 years maintaining these systems, that is the real cost.
- Prompt EngineeringAugust 30, 20266 min read
The four questions that stop AI writing 500 identical emails
Batch personalization collapses into sameness because each LLM call is stateless. Here is the reasoning framework I built to fix it in production.
- Prompt EngineeringAugust 26, 20265 min read
Prompt engineering is not dead. It stopped being about wording.
Gartner says prompt engineering is out. After 4 years writing prompts that run in production, here is what actually changed and what still matters.
- Context EngineeringAugust 22, 20265 min read
Context rot: why your AI agent gets worse the longer it runs
Model accuracy degrades measurably as the context window fills. After 4 years building agents, here is how I budget context so they do not decay.
- Context EngineeringAugust 19, 20265 min read
Context engineering: what actually belongs in the window
Better context beat better prompts 22 points to 3 in Shopify's testing. Here is the framework I use to decide what an AI system sees on every call.
- Case StudyJanuary 15, 20266 min read
How I generated $74K for a marketing agency in 45 days with AI
Most agencies waste months on proposals and cold calls that go nowhere. Here's the exact automated outbound system I built for Galax Media — and why it worked when everything else hadn't.
- StrategyNovember 11, 20245 min read
Why most AI automation fails (and what actually works)
Businesses spend thousands on AI tools that deliver nothing. It's not the tools that fail — it's the order of operations. Here's the mistake I see every single time.
- GrowthAugust 3, 20247 min read
From 200 to 20K followers in 3 months: the system behind the growth
It wasn't luck. It wasn't going viral once. It was a repeatable system for content production, distribution, and lead capture. Here's exactly how it worked.
Want this built for you, and kept running?
Free, yours either way. Reply within 24 hours.