Virtual Talk

Build vs Buy: the real cost of owning AI translation

Lokalise’s CTO and AI engineering lead on what it takes to build, run and trust AI translation in-house. Live panel and Q&A.

📅 November 24th, 2026

🕐 10am ET | 4pm CET

The code is the cheapest part of AI translation

Calling a model API and getting good German back takes an afternoon. Running AI translation in production, across thousands of strings and dozens of languages, takes glossaries, translation memory, context retrieval, evaluation, model routing, and the people to own all of it.

In this live panel, moderated by Customer Success Manager Victoria Milton, Lokalise CTO Magnus Slind-Naslund and Engineering Manager Sasho Savkov walk through what building AI translation in-house costs in people, money and process, and how Lokalise decides when a new AI model has earned the right to touch customer content.

What you’ll learn

Mistakes at scale

Why automation scales translation, and how it also makes errors harder to spot once you’re past a few hundred strings.

Sounding like you

Four systems AI needs to match your brand: glossary, style guides, translation memory, and context, plus the upkeep each one takes.

Trusting output

How to judge translation quality in languages nobody on your team reads, and why that evaluation system needs testing too.

How models earn their place

How Lokalise screens, rehearses, and routes every new AI model, one language pair at a time.

The verdict

When building in-house is the right call, and when owning it costs more than it’s worth.

Mistakes at scale

Why automation scales translation, and how it also makes errors harder to spot once you’re past a few hundred strings.

Sounding like you

Four systems AI needs to match your brand: glossary, style guides, translation memory, and context, plus the upkeep each one takes.

Trusting output

How to judge translation quality in languages nobody on your team reads, and why that evaluation system needs testing too.

How models earn their place

How Lokalise screens, rehearses, and routes every new AI model, one language pair at a time.

The verdict

When building in-house is the right call, and when owning it costs more than it’s worth.

Speaker profiles

Magnus Slind-Naslund

CTO, Lokalise

Magnus has spent his career building engineering teams that ship — at Teya, Spotify and Dailymotion — and now leads one working on a problem that touches every company trying to grow beyond its first market.

Sasho Savkov

Engineering Manager, Lokalise

Sasho is an Engineering Manager at Lokalise, where he leads the AI translation effort. He brings more than 15 years of experience in AI across academia, business sustainability and healthtech, and now applies it to localization.

Victoria Milton

Enterprise CS Manager, Lokalise

Victoria Milton is a Customer Success Manager at Lokalise. She helps product and engineering teams scale localization, and hears every day how they weigh up the build vs buy decision.

Magnus Slind-Naslund

CTO, Lokalise

Sasho Savkov

Engineering Manager, Lokalise

Victoria Milton

Enterprise CS Manager, Lokalise

Victoria Milton is a Customer Success Manager at Lokalise. She helps product and engineering teams scale localization, and hears every day how they weigh up the build vs buy decision.

Localization ROI: how to build the business case

How to build the business case that turns localization from a line item into your highest-return growth lever. Your board wants growth from global markets. Your competitors are already there. And your localization team has the tools to deliver. So why is localization still stuck in the "cost center" conversation? It's a familiar tension. The data overwhelmingly shows that localization drives revenue — through discoverability, conversion, and loyalty. Yet in too many organization

Read more Localization ROI: how to build the business case

AI Translation

AI translation quality evaluation: Looking beyond human review

The meeting is going well. Translation costs are down, turnaround times are shorter, and AI is taking on more of the work. Then your VP asks a question: “How do you know the quality of AI translations is good enough?” The expensive part usually starts once you try to answer that question inside a system you built: the real cost of buildi

Read more AI translation quality evaluation: Looking beyond human review

AI Translation·Product & News

Building an AI-powered translation flow using Lokalise API and webhooks

Managing translations in a growing product can quickly become repetitive and error-prone, especially when dealing with frequent content updates or multiple languages. Lokalise helps automate this process, and with the right setup you can build a full AI-powered translation pipeline that runs with minimal manual input. The pipeline is the cheap part. What it costs to keep is

Read more Building an AI-powered translation flow using Lokalise API and webhooks