Olga Stokowiec

Wróć do strony „Piszę”

What Do Buyers Want

Who are all these AI features really for?... I asked during my GALA Soapbox session earlier this year. I argued that the localization buyer landscape has split into three speeds:

  1. AI Avant-Garde who have been experimenting, piloting, and deploying AI for internal operations long before localization SaaS caught up
  2. Localization Dinosaurs who still run on spreadsheets and need a ton of data governance and easily implementable off-the-shelf fundamentals before they let any AI in (I mean, they can let AI in earlier and see what happens)
  3. Middle Lane who want to innovate but need to handle change management and compliance constraints, and for whom robustness is a non-negotiable.

So who are localization SaaS providers building for if none of the above groups is a clear target audience?

I don’t know. But I can tell you what makes my eyes sparkle as a localization buyer when I integrate localization software into my daily operations. Mind you, this is a list as of September 2026, with the prerequisite of having access to coding agents on top of the TMS. And as the TMS and AI landscapes are evolving all the time, this might be obsolete in half a year.

  1. Solid API documentation.

I don’t need Byzantine nice-to-have’s. I don’t need features that get shipped to tick off a box and then rot forgotten under an ongoing backlog, never making it to the second, third, or tenth iteration.

I want well-structured, detailed API documentation that can help me and my coding agents solve my problems on my own, without waiting for answers from already overloaded customer support heroes.

  1. Detailed history log for projects, jobs, and hell, maybe even strings if we want to dream big.

This can probably also be solved by good API documentation.

If my translation job relies on the TMS-GitHub integration, I don’t want my history log to just finish on “status: completed”. To troubleshoot autonomously, I want to know when exactly it was exported to GitHub, so that I spend less time searching for the specific PRs and string values I need.

With continuous localization, TMS is part of the software development infrastructure. Releases move fast, issues need quick fixes. Anything that removes unnecessary clicks and copy-pastes from my user experience is gold. Those saved minutes compound over days and weeks.

  1. Extensive but flexible analytics.

Again, something that API documentation can solve to a large extent. [Me from a year ago wouldn’t believe how passionate the present me is about API documentation.]

Every business is different. Every process is different. Operational needs and requests from above vary org to org. There’s no way a fixed set of dashboards is a sufficient one-size-fits-all. Even if there’s a lot of dashboards with a lot of filters.

You can clutter the UI with more off-the-shelf reports that will be half-useful orrrrr you can give your users enough freedom to build their own.

  1. Integrations.

Well this goes without saying. No one’s running localization on manually uploading and downloading files. Not in tech in any case. Robust integrations are non-negotiable. Bare minimum: Git, CMS, CRM, a cloud drive for emergencies.

Yep, API documentation won’t save the day for everyone. But: if a business cannot make use of your API documentation, are they even at a level of knowledge and organizational maturity to truly benefit much from a shiny AI add-on?

The dream setup is: a fluff-free TMS with solid basics + flexibility + great API documentation.

Solid basics are all that the Dinosaurs need to move on from spreadsheets. Solid basics are what gives the Middle Lane the required stability to fit localization into a bigger business landscape. Great API documentation allows the AI Avant-Garde to maybe not give up on the classic TMS solutions completely. And great API documentation is also useful for the Middle Lane to tailor their language operations to compliance or regulatory needs and dampen the shock of change management.

And the cherry-on-top AI features? I doubt they will make or break the business. ChatGPT wrappers and proprietary metrics are not where the true innovation lurks.