The Good, the Bad, and the Ugly of the AI Pivot: A C-Suite Audit for the Localization Industry

For the past two years, localization has been stuck on what Dave Ruane, Global Director of Client Solutions at Lion People Global, calls the “LOC sine wave of panic.” One month it’s existential dread – will AI replace us, will pricing collapse? The next, relief, because a big order just landed. Then a client asks how fast you can adapt to whatever’s next, and the cycle starts over.

If you’ve been to an industry conference in the last year, you’ve watched this play out live. First came the panic of feeling left behind. Then the buying spree – AI tools, connectors, orchestration platforms, co-pilots, consultants – all under the banner of “AI-enabled localization.” 

That phase is largely over. The industry isn’t experimenting anymore; it’s pivoting for real. Some LSPs got there early, made sharp calls, and are still solving their customers’ evolving problems. Others have a polished tech demo out front and not much behind it. 

To cut through that “polished LinkedIn version” of events, we hosted an Elevate Accelerate session with Rodrigo Corradi, an AI Transformation and Deployment Consultant, and Olga Blasco, Fractional C-Suite Executive, “Accelerating Into the AI Pivot: The Good, Bad, and Ugly.” This article pulls out the sharpest takeaways from that conversation: why the LSPs actually winning knew their own numbers before AI arrived, why Rodrigo thinks most leadership teams are practising anatomy when the moment calls for medicine, and what separates a real pivot from an expensive tech demo. 

Watch the full recording here:

Technology stopped being the differentiator

The biggest misconception during the experimentation phase was that having access to a large language model, or licensing an orchestration platform, counted as an edge. It doesn’t, and hasn’t for a while.

As Olga put it:

“Disruptive technology is first kind of led or adopted by a few, but then everybody else follows. So, having access to that disruptive technology is no longer your competitive advantage, because you might be able to use the exact same underlying tech as your competitor.”

Whether you built the tech, licensed it, or acquired it, everyone’s working with roughly the same toolkit. The real pivot is about execution: how quickly you can integrate new capability without breaking the services clients already rely on, and whether that integration actually strengthens your position in the market or just adds noise. 

The good: companies that were already disciplined

Rodrigo asked what separates the companies genuinely succeeding, and answered simply: they were the ones already operationally strong before generative AI showed up.

These were LSPs that had adopted machine translation and post-editing early, built solid KPIs, and kept a real handle on their metrics. Most importantly, they knew their own baseline while a lot of the market never got past sharing a quality score or two with clients, and had no way to prove what generative AI was actually doing for them once it arrived.

“Those organizations that were used to baselining their operational efficiency, exposing that to their customers, were the ones that later, when it came to a new technology, knew where their starting point was. For a lot of companies, that starting point just wasn’t there; they couldn’t baseline, and therefore they couldn’t measure the efficiency gains or any other gain when it came to adopting AI and getting an ROI.”

It’s like measuring the speed of a racing car with a broken timer, on a track you never measured. No baseline, no way to prove ROI – just guessing.

The companies that got this right also treated linguists and subject matter experts as more than production capacity, bringing them straight into QBRs and RFPs rather than keeping them buried just in delivery. Clients who were just as lost navigating AI as their suppliers found real reassurance in working with a partner who could show up with genuine depth, and that open dialogue is what built the stickiness now paying off.

The bad: enthusiasm without a foundation

Not every company that jumped on AI did so carelessly. Rodrigo frames “the bad” as well-meaning rather than reckless. The problem was sequencing: organisations ran straight at pilots without any structure underneath them, producing what he calls “careless churn” – experiments that were never statistically valid, never predictive, and impossible to draw conclusions from. Teams burned time on results nobody could interpret.

Much of this enthusiasm was also pointed inward, chasing margin gains and cost savings rather than asking what customers actually needed – missing the more valuable conversation of being honest with a client about shared uncertainty, and figuring it out together.

There was also a quieter failure of competitive intelligence: Many LSPs kept measuring themselves against the same rivals they’d always tracked, without noticing that language is now, for many buyers, just another data problem opening the door to better-funded tech-sector competitors most weren’t even watching.

On top of that, plenty of organisations assumed their existing teams could simply absorb AI deployment without any real training, and found out that they couldn’t. Closing this kind of skills gap organically – learning by doing – can take eighteen months to two years. In a market moving as fast as this one, that’s not a luxury companies have. Bringing in fractional expertise to bridge that gap is often the difference between winning an RFP now and losing the account for good.

The ugly: when the problem is leadership, not technology

This is where the conversation got most pointed. When LSPs come under real pressure, the instinctive response is the old playbook: cut costs, cut headcount, cut rates. It buys time, but strips out institutional knowledge without fixing anything structural.

At this point, it’s a leadership problem, not a market or tooling one. Rodrigo put it in an interesting way:

“It’s a difference between somebody who studies anatomy and somebody who studies medicine. So there are leaders that can describe pretty much everything that goes on in localization. But if localization is going through a difficult patch, you have to be a doctor, not just know the anatomy. And you need to be able to diagnose the issues and move forward with that diagnosis.”

A lot of leaders brought in from outside the industry can describe the business perfectly well. What they often can’t do is diagnose why it’s failing and act on it – because localization is genuinely complicated, and recognising that the business model itself is the problem takes courage most boards are reluctant to back. So pilots get funded with enthusiasm, and the moment things get busy, they quietly go on hold. That’s how ground gets lost.

Rodrigo shared a useful analogy: the shift from silent film to talkies in the 1920s. Plenty of the biggest stars of that era never made the jump, because their style of acting was built for a medium without sound, and they simply couldn’t or wouldn´t adapt. New disciplines had to be invented from scratch: sound engineering, dialogue writing, audio editing. It wasn’t a smooth transition, and it didn’t happen overnight. 

Localization has been going through something similar for decades; the industry has talked about moving words, when what it’s actually been doing is moving knowledge. That institutional knowledge is exactly what will feed the next wave of client-side AI roadmaps, and the leaders who make it through, in Rodrigo’s view, are the ones willing to admit they don’t have it all figured out: “Humility is your superpower when it comes to AI.”

What clients actually want now: governance, not magic

As the initial AI excitement fades, what sophisticated buyers care about has shifted. Olga referenced the idea that the most valuable AI-era roles aren’t the people building the models, but the people deciding how AI gets used, who it affects, and what it’s not allowed to touch – governance, privacy, ethics. As she put it: “The least glamorous corner of tech is suddenly the most important one.”

Early on, LSPs could win business just by showing off what AI could do. That’s not enough now; clients are increasingly aware of the legal exposure that comes with ungoverned AI use, and the partners who stand out can protect a client from a future lawsuit as well as impact real business objectives.

That points to a bigger truth: transformation is a people problem before it’s a technology problem. Previous tech shifts played out over roughly a decade. AI is moving two to three times that fast, fast enough to overwhelm teams and leaders alike.

Olga pointed to something Google’s CEO, Sundar Pichai, said:

“We are now in the part of the AI cycle where people want to see the value in the products they use every day.”

No grand claims about revolutions – just making the technology useful for what Olga called “ordinary Tuesday afternoons.” That’s the tone leadership needs: less about big statements, more about making AI genuinely usable for people. Transformation is driven by people, not technology. If leadership doesn’t bring people along, the whole thing stalls.

Key takeaways for executives

Rodrigo shared a set of takeaways aimed at CEOs and senior leaders – whether they’re already deep into their AI pivot or just starting to ask whether they should be.

• Stop treating your tech stack as your edge. Most competitors can access something similar. Your advantage is in execution, trust, and how well the technology fits your existing business.
• Get your baseline right before you chase ROI. You can’t measure a gain you never defined a starting point for.
• Don’t wait for the perfect strategy. The companies making real progress picked a well-defined problem, solved it, measured it, and moved to the next one. In many cases, the problems start small and then build. 
• Every pilot needs a commercial reason to exist. Experiments that don’t tie back to strategy just create noise.
• Audit your team’s actual skills, not the ones you assume they have. Training gaps caught early are cheaper than RFPs lost later.
• Watch a wider set of competitors. Language is increasingly a data problem, and the data industry has a lot of players your usual competitive tracking probably misses.
• Lead with humility. Ask customers what’s actually hurting them instead of guessing and building in isolation.
• Take governance seriously. Privacy, ethics, and risk management are now a genuine differentiator, not a compliance checkbox.

Where this leaves the industry

What comes through most clearly is that initial panic and experimentation phases have largely passed, and what’s left is a more honest, “do the work” phase. Nobody on the call was selling a silver bullet. Technology is easy to buy and will keep changing faster than anyone can plan for. What’s harder to buy is clear thinking, the discipline to measure what’s working, and leadership that can bring people through change without losing them along the way.

A couple of years ago, the question in most boardrooms was whether AI would change the industry. Nobody’s asking that anymore. The question now is what’s changing this quarter because of it.

If you’re trying to work out where your organisation sits – good, bad, or somewhere in between – Lion People Global’s Consultancy Solutions team works directly with localization leaders on baselining operations, closing the AI skills gap, and building the leadership case for the pivot ahead.

Explore our Consultancy Solution