Andrea Tabacchi, Chief Customer Officer, hits ATA hard with a solid, tech-centric schooling on all things Phrase
By Bridget Hylak, Administrator, ATA Language Technology Division
Members of ATA’s Language Technology Division, in collaboration with the Professional Development Committee, welcomed Andrea Tabacchi, Chief Customer Officer for the newly-branded Phrase language technology platform during ATA TEKTalks 2.1 on February 28, 2023. Approximately 200 registrants attended the online event.
The first “episode” as ATA TEKTalks moves into its second year, ATA TT 2.1 hit the tech button hard and left some participants reeling. Nonetheless, the information presented was thorough, timely and essential, as Phrase’s newly-branded and robust platform has mindfully crafted a masterpiece of inclusive, forward-thinking language industry technology.
While the high-tech register of some of Tabacchi’s answers may have overwhelmed some audience participants, others were no doubt there specifically for the juicy details. In short, some measure of lang tech literacy was required to drink from the tap of Tabacchi’s beer barrel that day.
As with most ATA TEKTalks, LTD Admin (yours truly) and Assistant Admin Daniel Šebesta, co-hosts, started with a standard opening question: With so many CAT and LOC tools and/or translation environments available on the market, the landscape can be pretty confusing to newcomers. What would you say are your tool’s top 3 features that are one-of-a-kind?
Tabacchi cut to the chase (and fast!), mentioning that he’d rather focus on overall capabilities as opposed to specific features, an early indicator and solid opener on how Phrase is currently posturing itself industry-wide. He specifically mentioned AI, analytics, and Phrase’s “Orchestrator,” which makes content flow between systems “at 100 miles an hour.”
With a simple smile, Tabacchi said humbly, “In Phrase you can automate a lot,” which, by the end of our time together, clearly proved to be an understatement.
“Another MT problem we identified was engine customization, and our proprietary MT engine, NextMT, is closing the gap between a full blown customization and generic engines, making it seamless for users to leverage content from translation memories and glossaries.”
“We have a team dedicated to building and improving integrations, and we are continuously releasing new ones. Building connectors and plugins means building bridges with platforms where content is created, allowing our users a lot of low-code or no-code flexibility. We’re kind of like the Zapier of the localization market… we really want to connect the whole industry as much as possible.
“Orchestrator in particular has managed to shift custom project flow tweaking, which our industry seems to require a lot of – from manual to autopilot.” Specific actions, workflows and even QA can be automated using it, he said, and business logic developed on top of the platform’s API will eventually allow a no-code interface users can use to bring automation to every area of the Phrase Suite.
If business analytics are important to your bottom line – and certainly, the wider that line is, the more important they are! – then Phrase has a lot to offer. “We were the first to build an integration with data warehousing platforms like Snowflake (dedicated to large enterprises), and with the latest release of our Advanced Analytics, we are bringing advanced reporting to a whole new level – especially in combination with other data (e.g. Google Analytics). This allows our users to be on top of their game backed by stats, not just gut instinct.”
This detail alone may be why 10-15 of the world’s top 20 LSPs, and large enterprises like Uber, Zendesk and Shopify, are among the platform’s clients.
Tabacchi had a lot to say (and a few things that he could only hint at) about Phrase’s hard work in the area of MT.
“A few years ago we started looking at the overall MT space, and we asked ourselves, how can we make it easier for our users to implement MT?
“We realized one of the obstacles was unpredictability, and to solve that we released the quality estimation model and MT autoselect, which leverages metrics to estimate the best MT model for given content types.
“Another MT problem we identified was engine customization, and our proprietary MT engine, NextMT, is closing the gap between a full blown customization and generic engines, making it seamless for users to leverage content from translation memories and glossaries – TM and MT on the fly – kind of unique!
“There’s actually a lot more cooking on the MT front, but I can’t tell you now…” he concluded with a smile.
(To which one or two of our co-hosts whispered, “Dang!”)
We were curious about how the Phrase/Memsource merger came about, and Tabacchi answered simply, “It was really a matter of timing and synergy. Handling software localization is hard, and so is building an enterprise-grade TMS. Each of us had excellent features and capabilities in place, and the marriage resulted in a comprehensive, holistic environment we call ‘Phrase Suite’ that brings together industry-leading parts of both companies. Things are moving too fast to build everything out, so instead of recreating the wheel separately, we joined forces to respond to what our customer base was asking for.
“We’re a language software company, not a marketplace – and that’s what we want to be,” he continued. “We are trying to empower our users, all of whom maintain control of their own operations, integrations and IP,” he explained.
“We are committed to offering one solution for the whole supply chain from small and large development teams who want to go global fast, to organizations with mature localization teams and processes, large and medium size LSPs, and of course, freelance translators.”
Apparently, he wasn’t joking. A quick glance across the landscape suggests that Phrase seems to be taking an intentional, leading role in localization, including language variants (and dialects) as a most normal part of the process. Moreover, as Tabacchi pointed out, they are currently working on a full range of certification tools, customized to the role a user plays within the platform (e.g., translator, PM, etc.), and they will also offer a “system integrator” certification for those looking to build out additional components themselves.
So with all that umph, what is the lowdown on cost? This question occasionally elicits some stammering or embellishment from ATA TEKTalks guests, but Tabacchi answered with a few specifics.
“We offer flexible entry-level subscriptions that are quite feature-rich, including our MT offering. Cost very much depends on the type of user and size of the organization, but specific pricing is available on our website. Integrations, MT and other add-ons like Orchestrator and Advanced Analytics comes at an extra cost, also outlined there.”
As for compatibility, like many newer lang tech tools and platforms, “Phrase is a 100% cloud-based system and can be used on Mac or PC. We also do provide a desktop editor for iOS.” Tabacchi also highlighted the platform’s robust support system, which includes regular webinars and user-friendly trainings. “Ultimately, our clients are the architects of their own ecosystem, and we do all we can to enable and support their efforts.”
Please join us on Thursday, April 27th, as we host ATA TEKTalks 2.2, featuring memoQ’s co-CEO Peter Reynolds (register here: https://www.atanet.org/event/ata-tektalks-is-memoq-the-right-tool-for-you/ ). Future installments this year will feature LSP.expert (Q3) and Lokalise (Q4). We may even feature a “bonus” TEKTalks 2.5 showcasing an exciting new tool that has just emerged from the RWS universe…
The Language Technology Division appreciates your input and suggestions on any tools, questions or conundrums you’d like to see featured in an upcoming ATA TT seminar – please reach out to us with your suggestions, and we hope to see you soon!
NOTE: LTD TEKTalks are not lang tool trainings (there are enough of those, just Google away!), but rather, showcase language industry tools and technologies, and especially the people and philosophies behind them.