New Page 1

Making money out of complex mathematics

By Maija Palmer, IT Correspondent

FINANCIAL TIMES
Published: January 26 2007 02:00 | Last updated: January 26 2007 02:00

Autonomy is a case of clever UK technology tapping into two rapidly growing global trends: the voracious use of e-mail, video and pictures and the increasing regulatory scrutiny of companies.

Cambridge-based Autonomy makes search software that allows companies to catalogue and access unstructured information such as e-mails, pictures, instant messages, voice calls and video clips - spotting common themes and pulling together the data. Autonomy helps MTV, for example, catalogue its library of music videos.

Unstructured information is essentially everything that does not sit neatly in a computer database. It is estimated to account for about 80 per cent of all the data held by a company.

Heavy use of e-mail and instant messaging by businesses is in itself increasing interest in Autonomy software, but new regulatory requirements are adding a sense of urgency.

Last month, for example, the US enacted new laws on the information companies must provide during lawsuits.

As well as traditional documents, companies must now supply e-mails, instant messages and call records when requested, all within 99 days. Failure to do so is a criminal offence.

"It means the argument 'the computer ate my homework' will just not work with the courts anymore," says Mike Lynch, chief executive of Autonomy.

At the same time, Autonomy's software increasingly appeals to companies whose businesses rely on sorting through large amounts of video and picture content. Autonomy sorts video, ringtone and picture content for mobile operators such as 3 and Vodafone.

Its search software is relatively unique in that it does not rely on key words. Instead it uses pattern-recognition algorithms based on work by Thomas Baynes, the 18th century mathematician. That means it can make sense of a set of pictures without anyone having to manually catalogue and assign tags to them, saving content companies considerable time.

Pattern-recognition also makes Autonomy software better suited to searching the growing repositories of Chinese-language data.

Autonomy is not a search company such as Google, whose keyword search technology dominates the internet. It tends to be found in corporate back offices. However, the company is experimenting with consumer applications in China, where it has a small joint venture with one of the country's biggest internet service providers to provide online TV and video clips.

There have been talks with US content companies on providing similar services elsewhere.

Mr Lynch sees the next two years as a crucial time for Autonomy.

The scrappy UK challenger already took itself into the big league at end of 2005 when it bought Verity, its larger and more established US rival. Much of the revenue lift in yesterday's figures came from Verity.

Autonomy has so farmanaged the integration of Verity in an exemplary manner. Mergers of software companies are notoriously difficult to pull off, howeverMr Lynch says not only have all the old Verity customers migrated on to Autonomy software, these customers have also been persuaded to pay more for the new kit.

Now, Mr Lynch says, the challenge will be to stay ahead of the game. The biggest threat may come from enterprise software companies such as Oracle IBM and SAP.

Autonomy has a head start with more than 16,000 corporate clients.

It is also building up an ecosystem of more than 350 technology companies - including HP, Cisco Systems and IBM - which have integrated Autonomy technology into their own software. With enough of these relationships, Mr Lynch hopes to establish Autonomy as the industry standard before rivals muscle in.

While Autonomy's core algorithms are protectedby 111 patents, Mr Lynch notes wryly, the mathematics are so complex that "you could write it down the side of a building and only a handful of people with advanced mathematics degrees could make any sense of it."