As the CRO at a leading AI agent company, I have front-row seats to what forward-thinking companies are doing with AI – who’s getting it right, and why so many fail.
A few months ago, LG2 – Canada’s largest independent creative powerhouse – launched Fromeo, an AI-powered "cheese butler".
Fromeo was designed to help consumers match their taste to one of hundreds of artisanal and industrial Quebec cheeses.
Commissioned by the Quebec Federation of Milk Producers (QFMP), the goal was simple but powerful: create an innovative, engaging way to promote local cheese production and elevate the brand’s reach.
In a world overflowing with disappointing AI projects, Fromeo stands out as a masterclass in correctly deploying AI agents.
1. AI is a paradigm shift, not an add-on
Too many organizations apply AI as a bolt-on feature. Or worse, a "me too" experiment.
But that’s not what LG2 did. With Fromeo, they didn’t just “add AI” to their existing setup. They created a tool that replaced the prior cheese directory.
I often see companies skip the crucial step of really understanding their current process (or lack thereof) and then asking, "How could this be reimagined with AI?"
Because successful AI isn’t just about automating tasks — it’s about rethinking the entire approach. The LG2 team was ready to dramatically redesign their user experience and website from the ground up, allowing them to squeeze value from every layer of an AI investment.
2. Every choice should be fixed to business outcomes
Fromeo wasn’t about showing off AI. The team didn’t chase novelty for novelty’s sake.
They targeted a real business case: improving consumer engagement by helping people navigate the overwhelming choice of local cheeses.
Instead of asking whether it would appeal to stakeholders, the team asked if it made financial sense to use AI.
They aligned on an AI solution that would drive awareness by:
- Increasing engagement by providing personalized information to every website visitor
- Improving discoverability by delivering a manageable, user-first experience through an otherwise overwhelming amount of local cheese options
This focus on a clear, measurable outcome separates success stories from the graveyard of flashy but pointless pilots.
Too many companies forget this: AI is only valuable when it delivers business value. AI is an investment. Investments are about ROI.
3. The user journey is front and center
The Fromeo project wasn’t about slapping a chatbot on the bottom right of a page and calling it innovation.
LG2 carefully mapped how people want to explore, discover, and engage with cheese recommendations — and then built an experience that made sense, inviting site visitors to engage with the hard-to-miss digital butler.
The interaction felt elegant and effortless because the user journey was designed, not left to chance. Understanding human behavior, crafting intuitive flows, and ensuring every interaction contributes to the larger goal.
The result: an AI agent that felt natural, useful, and deeply aligned with how people want to make choices.
4. Knowledge Bases should be intentional
The intelligence behind Fromeo wasn’t generic. It was precisely shaped to the project's business goals.
The Knowledge Base that powers Fromeo is aligned with the business objective, the audience, and the user journey.
Notably, it includes rich, relevant details to serve its user — not just text descriptions but images, visual cues, dietary restrictions, allergy information, and pairing contexts that truly matter to the end user.
It was intelligently curated to deliver meaningful, accurate, and helpful recommendations that respected individual preferences and needs.
Too often, teams throw generic data at AI and hope it "figures things out." (One of the common mistakes companies make when deploying AI.)
Not here. Precision and alignment were baked in from the start.
5. The benefits of a Crawl-Walk-Run strategy
Fromeo deliberately launched in beta. The agency is gathering real user feedback and telemetry, refining the agent, proving its value before expanding its scope and – most importantly – getting a win for their account.
Yes, they could connect recommendations to local merchants, incorporate social signals, or even integrate loyalty. But they’re smart enough to wait, advancing only after the core value is validated.
That’s disciplined innovation.
A blueprint for getting AI right
ChatGPT has made interacting with AI feel deceptively simple — type a prompt, get an answer, magic! But while the surface experience seems effortless, sticking the landing on real, production-grade AI automation within the enterprise is challenging.
Fromeo is currently only available in French, so only francophiles can enjoy its magic. But I urge every business leader considering an AI initiative to study it closely.
This is what happens when you use AI with intention and forethought — when you prioritize user experience, assemble the right cross-functional team, and deploy technology in service of the process (not the other way around).
Strong teams lead to strong outcomes
What did our team do to make this successful? Botpress’ engineer-to-engineer support provided guidance and AI best practices, and we left the strategy, creativity and development to our partner. This is a testament to the strength of our partner-first go-to-market.
As for me, I’m hosting friends next weekend and can’t wait to try some new cheese recommendations, courtesy of Fromeo.
It’s a small but powerful reminder: when AI is done right, it doesn’t just solve problems — it delights.
.svg)
