What do you do if you get more than three billion queries a day?
If you are Ben Gomes, Google's Tanzania-born, India-bred, US-educated vice-president of search, you are responsible for helping to answer all of them - in the shortest time possible, on all devices: desktops, tablets, phones. And now, in the spoken word too.
Search is Google's raison d'etre and cash cow, bringing in a bulk of its $50bn (£33bn) revenues last year. It is also, says Mr Gomes, "about having a continuous conversation with the user to find out what he wants".
For a change, we are having a conversation with Mr Gomes, the boy-like 45-year-old guru of search in Googleplex, the funky low-slung company headquarters set in the manicured greens of Mountain View, California.
From his modest lair, Mr Gomes and his team work relentlessly on their fine-tooth comb search of the worldwide web to serve up the popular search engine, which is now a part of our everyday lives.
"When I joined Google in 1999, search was about basically finding the words that you search for in a document. Then we took this view that we were going to understand what you want and give you what you need," he says.
Today, crawling through more than 20 billion pages a day on the continuously expanding world wide web, Mr Gomes and his army of search - a substantial number of the company's 44,000 employees - use algorithms in an attempt to make search intuitive, multimedia and super smart.
The "maths that computers use to decide stuff" - as algorithm expert Kevin Slavin called it once - helps rank pages in order of their importance, identifies spelling errors, provides alternatives to words, predicting auto-complete queries, and does unified searches using images, audio and video and voice.
And then, with a hint of understated pride, Mr Gomes talks animatedly about Knowledge Graph, a new function launched last year to make the site's algorithms "act more human" in an attempt to offer instant answers to search questions. Time magazine called it the "next frontier for search".
"It's a database of all things in the world. It pulls together different databases and unifies them into a single coherent one that has about 500 to 600 million people, places and things in them and about 18 billion attributes and connections between those things," he says.
