RSSAmplifier

Blog

Chris Parsons

I help leaders cut through the hype and leverage AI to transform their teams and dodge the chaotic hype that wastes time and money.

chrismdp.comRSS feed ↗10 posts

Latest posts

The Harness Is The Bottleneck

DeepSeek V4-Flash 0731 is doing better than I expected. It does not beat the best premium coding models: it is slower, stops early more often, and needs a decent harness built around it. But at about twenty cents a run(!) it is close enough to useful that the model cost is no longer the thing holding the work up. The harness is. Over the past few months I have learned a lot about running Ralph…

Intelligence Cost Is Collapsing

I am just back from a family holiday, and the model releases that happened while I was away show that AI models are becoming interchangeable faster than AI strategies can keep up. If your AI strategy is a bet on one vendor’s model staying ahead, think again: whatever intelligence you are paying for today, every competitor you have will be able to rent next quarter. And if you run a normal business…

Webinar: What CEOs Must Do Next About AI

On 9 July 2026, I gave a webinar about the AI decisions that only a chief executive can make. It is the first technology shift I have seen where the CEO’s own fluency sets the ceiling for the whole company, and that makes for an uncomfortable briefing: you can delegate the rollout, but you cannot delegate understanding it. The single decision that matters most this summer is how much to lean into…

Don't Just Give Everyone an AI

We built our companies to make software, because software used to be expensive to make. The teams, the roles, the handoffs, the sign-offs all exist for that reason. Now AI is making software cheap, and that raises a question nobody at the event was asking. If the thing you built the company to produce has suddenly become cheap, is the company still the right shape? That was the one idea I took to…

Slice Your Job Into Skills

Ask most people how they feel about AI and the answer is worry, not excitement. A recent Pew poll found that only 16 percent of Americans think AI will have a positive impact on society, while almost a third expect it to harm them personally, even as more people use AI chatbots than ever before. 1 The more people use it, the less they seem to trust it. Underneath a lot of that worry is the fear of…

Skills Are Meant For Sharing

One of the engineers on a team I train built something genuinely clever: a tool that turned plain questions into the gnarly database queries his team ran by hand a dozen times a day. It saved hours. For weeks, nobody else knew it existed. Then he showed it off in a sprint review, and within a week half the team was using it. The gap between a) someone building something useful with AI and b)…

Stop Saving Tokens. Start Writing Scripts.

The most powerful thing about a Large Language Model is what sets it apart from any other regular task a computer might perform. Computers normally do something exactly the same way every time (the correct term for this is “deterministic”): give it the same input and you get the same output, with no judgement involved. A Large Language Model is the opposite (it is “non-deterministic”). You can…

Webinar: Escape the Great AI Lock-In

On 4 June 2026, I gave a webinar about the moment cheap AI stops being cheap. One team I spoke to recently was handed a $10,000 monthly token budget by their management and blew through the whole thing in a day and a half. For two years AI has felt magical and effectively free, and we have quietly built our workflows on that feeling. The meter is coming back on, and the workflows we built are more…

I Don't Want an AI God

If you have been anywhere near the AI engineering crowd for the past year, you will have noticed a lot of us are quietly running our own personal agent orchestrations: n8n workflows that almost work, Ralph loops on cron, and bespoke harnesses vibe coded for one. We have given them access to everything , so they can look up anything about us they want. For myself, I have taken Ralph loops pretty…

Open Models Are Ready

I run a custom Kanban board as markdown files in an Obsidian vault, with multi-step projects for everything and background worker agents doing the work using systemd, cron and claude -p . Useful but expensive: the whole thing fits within a Claude Max subscription at about $200 per month alongside my normal interactive use, and I was hitting the limits before the cycle ended. So I spent this week…