I had two separate conversations this week that hint at how companies will more and more use computerized translation as a tool for global business.Translations tend to be either good (human) or fast (machines), but not both. My first conversation shows one way companies might improve on quality while meeting a real-time business need.

Idiom Technologies' WorldServer is software to manage the process of getting materials translated, primarily working with outside translators. Like other tools, it has a database for translated terms and phrases -- so a company doesn't, say, pay to have the company slogan translated with every batch of marketing materials.

Here's where it gets interesting: this latest version includes open APIs so companies can integrate their database of translated terms and phrases into other applications.

Symantec is using this technology to get security alerts out quickly in multiple languages, says Dave Rosenlund, Idiom VP of marketing and new business development. Researchers write an alert, the system scans the database of previous translations, and, for any unknown words, it turns to machine translation software.

Symantec and three other customers, including Google, pushed Idiom to create the APIs for this new version, says Rosenlund, all with the goal of connecting the database to machine translation software.

The other example this week came in discussing LexisNexis' new TotalPatent product, which will let people comb about 65 million patent records from 22 patent-issuing authorities. For some reason, not all inventing is done in English. So LexisNexis is working on a massive machine-translation project, and by the end of the year expects all new filings will get translated. By the middle of next year, it hopes to have the 65 million historical filings done.

The machine translations aren't perfect, not the sort of document a company will take to court, says Peter Vanderheyden, VP of global intellectual property for LexisNexis. But it will provide a good enough sense to narrow a search, and to create a shortlist of most relevant patent filings that may be worth spending money on a precisely technical, human translations.

There are other examples of innovators trying to build more real-time, accurate translation tools, including military efforts being tested in Iraq. They're all far from perfect. But as the world gets smaller, it's good to know the options are expanding.