company information leaking out, to an extent that sometimes felt overblown. She required not just employees to sign nondisclosure agreements, but anyone else who entered Theranos’s offices or did business with it. Even within the company, she kept tight control over the flow of information.
O’Connell’s actions confirmed her worst suspicions. Within days, she was laying the groundwork for a lawsuit. Theranos filed its fourteen-page complaint in California Superior Court on August 27, 2007. It requested that the court issue a temporary restraining order against the three former employees, appoint a special master “to ensure that they do not use or disclose Plaintiff’s trade secrets,” and award Theranos five different types of monetary damages.
In the ensuing weeks and months, the atmosphere at the office became oppressive. Document retention emails landed in employees’ in-boxes with regularity and Theranos went into lockdown. The head of IT, a computer technician named Matt Bissel, deployed security features that made everyone feel under surveillance. You couldn’t put a USB drive into an office computer without Bissel knowing about it. One employee got caught doing just that and was fired.
* * *
—
AMID THE DRAMA, the competition between engineering teams intensified. The new group competing with Ed’s was headed by Tony Nugent. Tony was a gruff, no-nonsense Irishman who’d spent eleven years at Logitech, the maker of computer accessories, followed by a stint at a company called Cholestech that made a simpler version of what Theranos was trying to build. Its handheld product, the Cholestech LDX, could perform three cholesterol tests and a glucose test on small samples of blood drawn from a finger.
Tony had initially been brought to Theranos as a consultant by Gary Hewett, Cholestech’s founder. He’d had to step into Hewett’s shoes when Hewett was fired after just five months as Theranos’s vice president of research and development.
Hewett’s conviction when he’d arrived at Theranos was that microfluidics didn’t work in blood diagnostics because the volumes were too small to allow for accurate measurements. But he hadn’t had time to come up with much of an alternative. That job now fell to Tony.
Tony decided that part of the Theranos value proposition should be to automate all the steps that bench chemists followed when they tested blood in a laboratory. In order to automate, Tony needed a robot. But he didn’t want to waste time building one from scratch, so he ordered a three-thousand-dollar glue-dispensing robot from a company in New Jersey called Fisnar. It became the heart of the new Theranos system.
The Fisnar robot was a pretty rudimentary piece of machinery. It was a mechanical arm fixed to a gantry that had three degrees of motion: right and left; forward and back; and up and down. Tony fastened a pipette—a slender translucent tube used to transfer or measure out small quantities of liquid—to the robot and programmed it to make the movements that a chemist would make in the lab.
With the help of another recently hired engineer named Dave Nelson, he eventually built a smaller version of the glue robot that fit inside an aluminum box a little wider and a little shorter than a desktop computer tower. Tony and Dave borrowed some components from the 1.0, like the electronics and the software, and added them to their box, which became the new reader.
The new cartridge was a tray containing little plastic tubes and two pipette tips. Like its microfluidic predecessor, it could only be used once. You placed the blood sample in one of the tubes and pushed the cartridge into the reader through a little door that swung upward. The reader’s robotic arm then went to work, replicating the human chemist’s steps.
First, it grabbed one of the two pipette tips and used it to aspirate the blood and mix it with diluents contained in the cartridge’s other tubes. Then it grabbed the other pipette tip and aspirated the diluted blood with it. This second tip was coated with antibodies, which attached themselves to the molecule of interest, creating a microscopic sandwich.
The robot’s last step was to aspirate reagents from yet another tube in the cartridge. When the reagents came into contact with the microscopic sandwiches, a chemical reaction occurred that emitted a light signal. An instrument inside the reader called a photomultiplier tube then translated the light signal into an electrical current.
The molecule’s concentration in the blood—what the test sought to measure—could be inferred from the power of the electrical current, which was proportional to the intensity of the light.
This blood-testing technique