At SlatorCon Silicon Valley 2024, three seasoned technology experts offered their collective 75+ years’ of expertise in localization and software to conference attendees.
The panel featured Pavel Soukenik, Head of Global Solutions at Acolad, Nitin Singhal, VP of Engineering (Data, AI, and Integrations) at SnapLogic, and Agustín Da Fieno Delucchi, Director of International Data & Applied Science at Microsoft.
Slator’s Senior Research Analyst, Alex Edwards, moderated the discussion and kicked off by asking the panelists for their perspective on the “build or buy” question for localization software and applications.
Microsoft’s Da Fieno Delucchi emphasized the importance of market analysis to leverage existing technologies before building new ones from scratch. “Unless you are willing to enter the market competing with another solution, don’t bother trying to build those pieces,” he stated.
“Some things we might decide to build ourselves, but there’s a beautiful area where it’s possible to partner with other companies.” — Pavel Soukenik, Head of Global Solutions, Acolad
Acolad’s Soukenik added that enterprise buyers may also consider partnering with other companies to achieve a successful localization program, as part of the build or buy needs analysis.
He told the SlatorCon audience, “some things we might decide to build ourselves, but there’s a beautiful area where it’s possible to partner with other companies.”
On the topic of connectivity, SnapLogic’s Singhal emphasized the benefits of interoperability between software solutions: “a system is not a sum of its parts, but is actually a product of its interactions”.
Singhal highlighted how plug-and-play solutions do not require heavy code, and how leveraging APIs enables companies to easily integrate translation to existing processes with tools like SnapLogic.
“A system is not a sum of its parts, but is actually a product of its interactions” — Nitin Singhal, VP of Engineering (Data, AI, and Integrations), SnapLogic
The discussion on software and integrations led the panelists to discuss the challenges of data and legal compliance. Singhal said, “the key thing is to understand your data lifecycle, see what are the integration and the leakage points, make sure that those are minimized from access and data collection by default, and that your architecture is such that you can configure metadata for future [changes in] policies.”
Microsoft’s Da Fieno Delucchi agreed: “you need to try to anticipate [regulatory changes in data]. […] Being intentional about designing your [data] strategies […] is fundamental for all players in the industry.”
To conclude, Da Fieno Delucchi, who dedicates a large proportion of time to assessing the quality, cost, and performance of large language models, told the SlatorCon audience, “we need to be very intentional about including the humans in the loop.”
“We are using [large language models] to become more efficient, to handle more volume, but it doesn’t mean that we are going to remove humans. I think we need to be very, very intentional about it,” he concluded.