Curl Corp., a Cambridge, Mass., startup spun out of MIT's computer lab, has released its initial Web-development software along with promises of "maximizing the user experience on the Web."

Heard such lofty promises from other tech hot shots before? Curl's pedigree and the way it changes the concept of Web page delivery appear to give it a good chance to make good on its word.

For starters, its founding team includes MIT computer researcher Tim Berners-Lee, who created the World Wide Web and remains a Curl advisor.

He began work on Curl in 1995, when it was a DARPA-funded scientific project to improve Web communications, along with other marquee names in Web circles: Michael Dertouzos, director of MIT's computer science lab, and Stephen Ward, a computer science and engineering professor. In all, about a dozen MIT researchers have a hand in Curl. Curl's chairman and CEO is Bob Young, formerly a managing director with Dillon, Read & Co. and an ex-IBM executive.

Spun off into a private company in 1998, Curl has raised about $50 million in venture capital and hired 150 people, mostly engineers, at its office in Cambridge's Technology Square.

Its first commercial releases came this week. The Curl Surge Lab is an open-source platform and programming language for Web developers. The Curl Surge browser plug-in resides on your PC or laptop so users can read "Curled" Web pages. Both are available for free online.

In plainest terms, Curl software speeds Web page downloads. It does so by allowing the Web page source code (the HTML, Java, Java Script, Flash, etc., that comprises text, photos, graphics and animation) to be compiled on a user's computer, not on far-flung servers.

Under current content delivery methods, when Web pages are compiled on servers the bulky files are carried on copper or fiber lines to users' Microsoft or Netscape browsers. Users are often left waiting 15-20 seconds waiting for media-rich pages to load.

Since Curl moves the work of turning code into viewable Web pages onto a client machine, smaller files get sent, faster downloads occur, and less bandwidth is used, according to Bob Batty, vice president of sales and marketing. It saves companies money, he added.

Though the software is free, Curl makes money by metering and charging fees to businesses based on the amount of "Curled" data their users download, Batty says. (The name Curl comes from an element of its source code, those curled brackets on a keyboard.)

Batty said, "A 100kb file in Java, Java Script and HTML could in Curl be as little as 10kb," and download time could be cut from 20 seconds to two seconds, he said. "Because Curl is a compiler we can put user interface for (Web) applications on the client device."

Batty added, "When the user interface is local, you're getting Web-deployed services that look and feel as if they're local on your machine," including rich animation and 3-D applications.

Curl is going after customers in several industries where consumers make heavy demands on their Web servers. They're also industries where better Web experiences can translate into more sales.

Among them: hospitality/ticket sales, financial services, entertainment and publishing. Batty said Curl expects to announce major customer deals in the coming weeks with "no less than two multinational companies."

Batty said one financial services firm, using Curl to power its financial wealth calculator, can run the software at a cost of about 30-40 cents per customer per year.

Noting that other companies including MIT spin-off, Akamai have worked to improve Web downloads by locating content servers near users, Batty said Curl thinks taking the concept the next step will permanently alter Web computing.

"We consider ourselves to be at the ultimate edge of the Internet -- at the device level," Batty said.