Is your prompt library already out of date?

Hello,

This is Simon with the latest edition of The Weekly. In these updates, I share key AI related stories from this week's news, list upcoming events, and share any longer form articles posted on the website.

P.S. — sharing this issue with a couple of people gets you a free guide. More on that partway down.

I was talking to someone about their AI use this week, and one thing they mentioned was that they had created a prompt library for their most frequent tasks. When you first hear this, you might think it sounds like a good and efficient idea. Instead of constantly repeating a similar question, you can paste it from a selection of prompts you have saved. However, it actually feels like quite a dated workflow these days. If this conversation had happened a year ago, I think it would have been an interesting idea, and something many of us were doing. But there are now far more sophisticated solutions for the tasks we find ourselves repeating, and if you're an active follower of AI, you might already be thinking that creating a Skill would be a better approach.

And I think you would be right. These days, we can give our AI tool a package that bundles up the prompt we use most often, along with further information such as specific instructions and context. At its most basic level, all the details you want a Skill to consider go inside a markdown file called SKILL.md. This holds the core instructions, plus any required steps and details of your desired output. You can then optionally add scripts, references and assets. Once you have created your Skill, you can use it at any time with a simple prompt and it will work through your instructions every time. This is a huge time saver, and it also gives you a certain level of repeatability, so that you get more consistency in the results.

So as you can see, Skills are far more sophisticated than a simple prompt library. They give you a lot of control, and they are very time-efficient too.

To make this more concrete, here is an example of a first Skill that almost anyone could build: a meeting follow-up. Most of us write some version of this email several times a week, and it usually has the same shape each time. In your SKILL.md, you would start by describing the job in a sentence or two, for example, that you want a short, friendly follow-up email after a meeting, written from the notes or transcript you paste in.

You would then add the steps you want it to follow, which might be to pull out the decisions that were made, list each action with an owner and a date, and flag anything that was left unresolved. Finally, you would describe the output you want, perhaps a subject line, a two-sentence opening, the actions as a short list, and a sign-off in your usual tone. Once that is saved, you only need to say something like "use my follow-up Skill on these notes" at the end of a meeting, and you get the same structure every time without rewriting the instructions from scratch.

If you've not yet explored creating your own Skill, it's definitely something I would encourage.

AI’s growth story is much bigger than chatbots.

The market was already valued at an estimated $800 billion by late 2025, and continued adoption could push it beyond $1 trillion in the years ahead. That growth will require more data centers, computing power, software, networking and intelligent devices.

But the most obvious AI names may not be the only companies positioned to benefit.

MarketBeat’s free 10 Best AI Stocks to Own in 2026 report identifies ten publicly traded companies helping power the next phase of the AI buildout. Some are established leaders. Others occupy less obvious corners of the expanding AI ecosystem.

This free report names 10 stocks positioned to ride that wave before the rest of the market catches on.

Gartner's survey of more than 2,000 CIOs and technology executives found 71% of organisations expect to deploy agentic AI within the next 12 months, but only 21% have governance capabilities for it in place. The firm presented the findings at its IT Symposium in Barcelona on 9 November, as reported by CDO Magazine. A separate Gartner survey of more than 1,000 functional leaders found 58% see some positive value from AI, but many are struggling to scale projects beyond pilots. Gartner's advice is to treat controls such as audit trails, access limits and human review as "good drag" rather than red tape.

Why it matters: Agents act rather than just answer: they send the email, change the record, approve the refund. Deploying them without clear ownership, logging and a way to switch them off is how a pilot becomes an incident. If your organisation is in the 71%, the governance work needs to start now, not after launch.

US employers attributed 1,265 job cuts to AI in October, less than 3% of all announced cuts and the lowest monthly figure of 2026, according to outplacement firm Challenger, Gray & Christmas on 5 November. AI is still behind about 121,000 announced cuts so far this year, around one in five, but "market and economic conditions" has overtaken it as the main stated reason. Separately, Fortune reports a survey by assessment firm SHL in which about 19% of US respondents who had been replaced by AI said they were later asked to come back. Half of those returning wanted a substantial pay rise, and 42% wanted written job guarantees.

Why it matters: The idea that you can swap staff for AI and bank the savings is running into reality. Some organisations cut too early and are now paying more to rebuild the same capability. Before restructuring around AI, test whether the tools can actually do the work at scale, and remember that people you let go now know their value.

The Nasdaq fell 5.5% and the S&P 500 4.4% last Thursday, their worst sessions since April 2025, in a sell-off that wiped about $2.5 trillion off the Nasdaq 100, according to AP and Reuters. Against that backdrop, four people familiar with the talks told Reuters on 6 November that some potential investors doubt Anthropic can start its roadshow this week and list in November as planned. Some want a valuation below the recently discussed $1.5 trillion, down from earlier talk of up to $2 trillion. Bloomberg separately reported that Microsoft is winding down a 2023 deal to rent AI cloud capacity from Oracle and has tightened oversight of some data-centre projects with smaller providers.

Why it matters: The investment cycle that has subsidised cheap, fast-improving AI tools is now being questioned by the people paying for it. That doesn't mean your vendors are in trouble, but it does mean more pressure on prices and margins. Check the financial health and contract terms of the AI suppliers your organisation is becoming dependent on.

Thanks for reading, and see you next Thursday.

Simon,

Was this email shared with you? If so subscribe here to get your own edition every Thursday.

You can also follow me on social media: