Interview: Mike Lynch, founder of Autonomy

On Google, penguins and the future of search...

During the turn of the century technology boom years, Autonomy was something of a poster child for UK high-tech. A lot has changed since then in terms of the fluctuating health of the sector and the nature of the search industry - not least of all because of the rise of Google.

But Autonomy founder Mike Lynch - who remains CEO at the company - is convinced Google will not repeat its consumer success in the enterprise market and believes the search industry is on the verge of great things.

"Unstructured information" remains at the heart of Autonomy's approach. Lynch claims 80 per cent of information within the enterprise is unstructured and with the sheer volume of information ballooning, on the back of a move towards greater compliance and new data formats such as multimedia files, the need is greater than ever to be able to search this data quickly and effectively.

Applications serving this growing market are what Lynch refers to as "second generation" search technologies. Google, he admits rules the first generation but his sights are set on more ambitious challenges.

"A lot of people are getting the message of unstructured information very strongly. We're starting to see an appetite for second-generation technologies. There is no point in trying to out-Google Google."

"A lot of people think search is about looking for a word. It's not, it's about an idea," Lynch told silicon.com.

"Google has very little impact on our end of the market. The one thing it has done is dumb-down how people search, which is a shame," said Lynch.

"The important thing is to understand what Google does. Google is very successful at internet search. And when we talk about internet search we're talking about typing one or two words into a blank box.

"Say I'm interested in the effect of oil pollution on the penguin population of Alaska. Although that's the idea someone is looking for they will walk up to a search engine and just type 'penguin'."

"They would never walk up to a librarian and just say 'penguin'. And that's the Google effect. We've been trained to assume the search engine is dumb and that takes a little un-training in enterprise."

But Lynch believes businesses will all come around to the importance of search within the enterprise.

"The bigger you are, the more unstructured information you create," he explains, and the move towards the 'paperless office' is making a huge difference. "Email and the move away from dead trees are creating this data."

Autonomy already includes Astra Zeneca, British Aerospace, Nokia, Shell and Vodafone among a growing customer list.

"Now people are thinking about what the next generation of search will offer," he said. "You'll be able to search TV right down to every individual bit, not just the title or the meta data."

Lynch believes precise moments of speech and footage will be searchable within three-hour features, once number of hurdles have been overcome - not least of which is the "most boring" issue of digital rights management.

Lynch added: "The other issue is having technology which understands what is being said and shown in that bit of a TV show, because you can't have people watch it and meta-tag it every three seconds with what's going on."

"But once you can organise the technology then you can take these very big archives, like the BBC are doing at the moment, and make it all available," he said.

Perhaps of greater relevance to the enterprise is the growth in IP telephony and the application of such search technologies to archive and search all calls.

"A big growth area in making phone calls searchable is of course in the call centre," said Lynch. "So as VoIP really takes hold people will be searching phone calls like they do email. And in the call centre people will want to find all the calls where customers complained about the same thing.

"This is all just part of handling unstructured information and it's live in call centres right now. It's in the Vodafone call centre now and the most powerful way it works is that it listens in on the calls and suggests answers on the screen."

Such a system is fully integrated with CRM systems and searches similar enquiries and suggested responses and solutions in real time.

"There is a real return on investment here for mobile telephony and insurance call centres because there is a 30 per cent reduction in staff needs because the calls are 30 per cent shorter," said Lynch.

"Once you start talking like this the ROI becomes very clear. You're actually taking out jobs that would have been done by people," he added, though he stressed that this need not equate to redundancies but rather may be a case of freeing up staff from jobs which could be automated.

"There are all sorts of ROI figures you could produce which show how much time and money is wasted within the enterprise because of bad search."

And while time saving and ROI are major issues in driving enterprise level search adoption they are far from the most important driver Lynch identifies.

"Some of it will be very strategic, which is heartening. Big companies realising the importance of unstructured information is the dream conversation," he said.

"Another big driver at the moment will be a reactive need, and a key driver there will be corporate governance. Take for example a large US company. And the CEO gets called up before Congress because something has gone wrong. He hasn't done it but it's happened somewhere within his company. You'd be amazed how quickly they'll suddenly connect up all their unstructured information.

"In the US they have to personally sign off everything but how the hell do you know everything that's happening in your organisation? And what happens when the SEC knocks on the door and says we want everything to do with a certain subject?"

Similarly search can work the other way around - alerting somebody to something they really didn't know they wanted, or needed, to know about.

"Search is going to become a lot more than typing words into a box. It's going to become about alerting. This has just happened, or this has just happened in your Malaysia office or we're getting an awful lot of complaints coming into the call centre about this problem with the product," said Lynch.

"It's not traditional search because you're not typing something in a box but there's a crucial need."

And in flagging up problems, companies will also find they come across inefficiencies within their operation.

"A lot of it comes back to the fact that 80 per cent of our information is unstructured. The number of people who have worked out that the same problem is being worked on by 16 or 17 people across the organisation is massive.

"The classic we had was with British Aerospace and I remember watching this happen. They had two groups, each with £7m, with the second group working on exactly the same problem as they first group who were just 200 miles away," said Lynch.

"It's not uncommon to find the same thing worked on 20 times."

And the amount of data will only increase as electronic communications and the quantity of data being shared, stored and archived increases.

"If you want to back up on all of this, what's going on is really quite fascinating," said Lynch.

"If you were sitting in an English village 1,000 years ago then nothing happened. Occasionally every hundred years or so Black Death would come along, or the odd riot, and then what we've seen in the past century is an explosion in the amount of data being delivered to a person.

"In the 1920s you might have got one phone call to a village and some small boy would run out to get the person the call was for. And then everybody had a phone but the idea of having a mobile phone was a Star Trek thing - I think people forget that. I used to drive a couple of hours to a business meeting and have the radio on but now I'll be on the phone the whole way. And then there's email.

"The amount of data going to an individual is now immense," said Lynch. "The ability to slice, dice and route that information is going to become very fundamental."