Which side of this do you want to be on?
When I'm speaking to a CEO or a CMO, that's the question I ask. Do you want to ignore the fact that AI is present, or shall we learn to execute with it in a way that is safe, reliable, and enhances your team's abilities? Particularly when it comes to content build, the interactive work, and empowering your people to build small connectors and MCPs that speed up the work of their entire team.
This type of thing is already happening at your competitors. The moat, if you like, is for you to see what's possible, take that away, and think about what can be done with it to improve the speed at which your business operates. That's the main principle behind a lot of what I do.
Software as a service can be disrupted. A website can be disrupted. But if you have first-party data that you're collecting and analysing yourself, that adds a level of value you can put towards better features for your customers: features that help them make decisions, or get them through the purchase journey quicker.
Even in my Builtvisible days I was writing about data scraping for content research and analytics, capturing things like pricing data and stock levels. It's been a feature of my work for the last decade.
The AI model fetches the data through an MCP, builds the dataset for you or interprets it, and monitors, through routines and whatever automation you need, to make sure the changes have had a positive effect. It's almost a built-in testing and follow-up engine, where the AI will tell you: maybe we should go and look at that.
The pushback is usually: hang on a minute, don't MCPs hallucinate? The MCP is only collecting data from API sources and writing it to a SQL database, so it's deterministic data. You ask a question about the performance of your site; it formulates the right query, gives you the raw data if you want it, and interprets it for you. It's fine to go and check those answers are correct, but most of the time they are.
Not everybody is using AI for a hundred percent of their work, and some people may never need to or want to. In a company whose processes still need some refinement, we can automate and rebuild a process, but in the meantime I work with the individuals to train them up and give them those abilities, so they know what they're doing with AI. A fleet of AI-enabled people in any company is a very positive thing. And when someone builds their first idea, the penny drops, and I love how excited people get about it.
Assure was about helping a compliance team not get so bogged down in the data collection. The price monitor is about having the information to hand so the people in the business can make decisions quickly. And naturally, scraping a hundred thousand prices from quite high-end web technology behind Akamai bot protection is no easy task.
That's where I thrive. But I'm always keen to share and hand over, in a three to six month engagement, and be your source of knowledge while your team upskills.
What is the hard problem you can't solve for a reasonable budget?
We've come out of a world where you needed to invest in quite specific SaaS tools to do things that were fundamental to the operation of your business. This new world lets us build that intellectual property into the core of your company.