Stop guessing: How to find out where AI will actually pay off in your business
Jovana Lazarevska
The question we get asked most about AI is which tools to use. Should everyone get a Copilot licence? Is Claude better than ChatGPT? Is that new AI-powered CRM worth switching to?
For our 11th webinar, Phil was joined by Anish, our AI and Automation Lead, to make the case that this is the wrong place to start. The right first question is much simpler: where are we leaking time, and what is it costing us? Get that answer and the tool choice tends to sort itself out.
Three flavours of automation, not one
Phil opened by separating “AI” into three distinct tools, because most businesses default to the same one regardless of the job.
Traditional automation (think Power Automate or Zapier) is fast and rigid. It follows exact steps and breaks the moment something unexpected shows up. AI automation keeps that same rigid process but lets a large language model handle the messy bits in between, like reading a lease document or a passport and pulling out the right fields even when the format varies. Agentic AI is the newest layer: give it an outcome and enough context about your business, and it works out the steps itself. It is far more flexible, but also far less predictable, so it needs guardrails.
Phil’s analogy stuck: a robot trained only to wash plates fails the moment you hand it a mug. Teach it to recognise categories and it copes with a mug too, but a teaspoon might still trip it up. Tell it to simply “keep the kitchen clean” and give it the full context, and it should handle anything, just not always perfectly.
RIPE: how to spot the right candidates
Anish then walked through the framework Blue Llama uses to find good opportunities before touching any AI at all: RIPE. Is the process Repetitive, Irritating, Predictable, and Entry-friendly (does it already sit inside easy-to-integrate tools like Outlook or SharePoint, rather than a 20-year-old legacy system)?
Score enough of these highly and you likely have a strong candidate. In a live example from a Jersey business, engagement letters scored 19 out of 20 on RIPE. Client onboarding scored high too, but not as predictable. Approval minutes were repetitive but scored low on irritation and entry-friendliness.
TRIM: turning that into a number
Here is the twist. What scores highest on RIPE is not always what delivers the most value, which is where TRIM comes in: Time drain, Risk exposure, Impact on scale, and Money.
In the same business, engagement letters, the RIPE winner, actually scored low on TRIM. Client onboarding, ranked second on RIPE, came out on top once the real cost of the labour, the risk of getting it wrong, and the impact on growth were factored in. That is the point of running both frameworks: RIPE tells you what is annoying, TRIM tells you what is actually worth fixing.
What the numbers look like in practice
Anish shared results from a three-hour workshop Blue Llama ran with local businesses back in May. For one organisation, automating client onboarding alone was worth close to £100,000 a year in savings. Across their top three processes, the total annual saving came to just over £100,000 against a £35,000 build cost, a payback period of roughly 17 weeks and a three-year net saving of around £283,000.
For another client, a chatbot built around five workflows for a specialist inspection team delivered a 6x return on investment and roughly £122,000 in savings over three years, without compromising the quality of the reports the team produced.
What to do next
Start by listing the processes in your business that feel repetitive, irritating, or both. Score them against RIPE, then push the strongest ones through TRIM before committing to build anything. It usually surfaces a different answer than gut feel would.
If you want to run through this properly, Blue Llama is hosting a free “AI for Profit” workshop on Tuesday 22 September, working through RIPE and TRIM live with your own opportunities in the room. Places are limited to five or six businesses.