If it were a Hollywood movie, the pitch would be: Mission Impossible meets Silicon Valley. But it's real life, and it's one of the oddest -- and most innovative -- things the Central Intelligence Agency has done in years.

The CIA has decided to create its own venture capital firm, called "In-Q-It," to help the agency connect better with the Internet revolution. The fear at Langley is that in a world of start-ups and instant millionaires, the CIA isn't getting technology's best and brightest anymore. So the spymasters have opted to create their own start-up, with plans for an office on Sand Hill Road in Palo Alto, where the leading venture capitalists hang out, and a $150 million kitty.

The idea is for In-Q-It to fund promising technologies that can help the CIA keep pace with the information explosion. It will be small, with a staff of 20 to 25 people, and will operate much like a normal venture fund -- partnering with other companies and funds. In theory, its activities will be entirely unclassified. And though it will be a nonprofit organization, the goal is to invest wisely enough that after five years, it will be self-financing.

As examples of the kinds of problems In-Q-It will work on, agency officials cite the need for smarter CIA search engines, better ways to visualize data, and better security for CIA web surfers. But the broader goal is to link an agency that can't give stock options with the cleverest minds in the tech world.

The CIA plans to make a formal announcement of In-Q-It in a few weeks, but they've already chosen a board of directors -- including such tech luminaries as John Seely Brown of Xerox PARC, Jeong Kim of Lucent, Alex Mandl of Teligent and Norm Augustine of Lockheed Martin. And the board just hired the first CEO, an energetic 39-year-old named Gilman Louie.

Louie's background gives a sense of what an unusual operation this is likely to be. A fourth-generation Chinese American (and son of a World War II veteran of the Army Air Corps), he made his name creating video games, including the Falcon air-combat simulator, beloved by computer buffs around the world. He's a genuine Silicon Valley entrepreneur, who started his first company in his mother's garage in San Francisco when he was 22 and hangs out with other wealthy geeks. His favorite recent movie is "The Matrix," natch, which he has seen four times.

Louie appears to know relatively little about the CIA and its venerable traditions -- which is probably a plus. He decided to take the job (at what's likely to be a huge pay cut) because it was a way "of doing something for my country."

"I was amazed the the agency was willing to do this," says Louie. "It's very `out of the box.' "

Out of the box, it is. CIA officials explain that In-Q-It evolved out of growing frustration that the agency was losing its once-vaunted technical edge, which helped create the U-2 spy plane, overhead satellite reconnaissance and dozens of other collection technologies.

The idea for the venture-capital fund was hatched in conversations between the new CIA director, George Tenet, and a former investment banker named A. B. "Buzzy" Krongard, who joined the agency in February 1998 as counselor to the director. Krongard had been CEO of Alex Brown, the investment bank that helped launch Microsoft, Sun Microsystems, AOL and dozens of other high-tech IPOs.

Tenet floated the venture-fund idea to his CIA colleagues in May 1998, referring to it simply as "The Enterprise." The project was turned over to a group that included an energetic young woman named Sue Gordon, a former Duke basketball player who was working in the agency's science and technology directorate. She began making the rounds in Silicon Valley to flesh out the idea and drum up support. Gordon also helped come up with the name: "In" for intelligence; "It" for information technology, and "Q" in the middle because it was the code name for James Bond's technology wizard, and it just sounded cool.

Within six months, Gordon and her colleagues had created a framework for the new organization -- a rate of speed that's typical for a Valley start-up, but almost unheard of in government. Her message to skeptical CIA colleagues was: "In order to move forward, you're going to have to let go" -- again, a sentiment rarely heard in government.

The In-Q-It team members know they'll face resistance from techies, for whom suspicion of the intelligence world is almost a cultural requisite. Indeed, Louie himself was lobbying the government a few years ago to abandon its efforts to control export of encryption software.

What the intelligence community didn't understand back then, says Louie, is that technology is moving too fast now for anyone to try to control it. Nowadays, a 16-year-old can download encryption software from the Net that's so sophisticated it will confound codebreakers at the National Security Agency.

"The cat is already out of the bag," says Louie. The government's biggest job is just to keep up with the pace of change.

The best thing about In-Q-It, to me, is that it's risky. The CIA and the rest of the government need to catch the entrepreneurial, risk-taking spirit that's driving the Silicon Valley technology revolution.

The CIA's new venture fund may fall flat, but so what. Washington has been a zero-defect culture for too long. If we want a CIA that performs better, we'll need to take more risks -- and give our government the freedom to fail.