Ask most Salesforce teams how customised their org is and you will get a shrug, or a guess pulled from whoever has been there longest. “Pretty custom, I think.” “We’ve got a lot of Apex.” Nobody actually knows, because nobody is measuring it. That is the gap the CEER methodology was built to close, and it deserves to be far better known than it is. What CEER actually is CEER — Config, Extend, Enhance, Replace — is a framework for classifying every piece of Salesforce customisation into one of four buckets, based on how far it moves away from the platform’s standard, out-of-the-box behaviour. It was proposed by Salesforce architect Steven Herod, presented at Dreamforce 2019, and later written up in detail by Lennart Svensson at Storebrand, who applied it to measure and track customisation across a real production org (“Using CEER to measure the level of customization in our Salesforce org”, Dreamforce 2019 slides). ...
The Tokenomics Trap: Why Cheaper AI Is Making Enterprise AI More Expensive
Every few weeks another headline lands about token prices falling off a cliff. A new model matches last year’s frontier performance at a tenth of the cost. Input tokens that cost dollars in 2021 now cost cents. On paper, AI keeps getting cheaper. So why do so many enterprise AI budgets keep blowing up mid-year? I’ve been thinking about this a lot lately, prompted by a BBC piece on the strange economics of AI tokens (worth a read in full), and by a string of vendor and analyst reports telling a version of the same story from different angles. The short version: the unit price of a token and the total cost of using AI have quietly come apart from each other, and most organisations are still budgeting as if they haven’t — a gap I’ve touched on before when writing about where token costs actually accumulate in production systems. ...
The Best AI Prompting Framework Isn't RISPO — Here's What Actually Works
If you’ve been searching for “the RISPO prompting framework,” you’re not alone — and you’re also not going to find much. There’s no canonical documentation for it anywhere: not in Anthropic’s, OpenAI’s, or Google’s official prompting guides, not in the academic prompt-engineering literature, and not on any of the sites that catalogue these things. It’s most likely a mishearing or a mash-up of RISEN (Role, Instructions, Steps, End goal, Narrowing) — a genuinely well-documented framework — with one of the dozen other acronyms floating around AI Twitter and LinkedIn. ...