Very old speedlights were manual and fired their full charge each time. As flashes became smaller, they could be run by AA batteries or rechargeable NiCd's hardwired into the flash unit.
The earliest automation featured a built-in sensor in the flash unit itself which would sense sufficient exposure and then "dump" the rest of the charge like KN4SMF said. It wouldn't need the camera body to tell it when light was sufficient as AgX says.
Relatively speaking they chewed through batteries / charges. As a consumer it was a pain to discover how short your battery life and you felt it in terms of how long you had to wait for the "charge" indicator to tell you you were ready for the next shot.
The thrysistor was actually a good, real benefit to users of small portable flash units. It's really a magic bullet! I'm very glad the technology was either generic or widely licensed because it really works!
If you turned down the power (used a larger f/stop) or worked closely (needed less flash for the shot), you were rewarded with an orange glow just a few seconds after your last shot.
At first, you would have to mount the flash in a shoe and aim it straight forward. To solve that issue, and make it easier to use bounce flash, there was a trend towards making the sensor face forward even if the flash was tilted (like the Vivitar 283).
Pentax made the Spotmatic IIa with a "strobo-eye" on the camera, Vivitar gave a cable you could put in the camera's hotshoe and mount the sensor onto, so you could connect the flash to the camera and aim it anywhere while the light returning to camera was measured through the remote sensor.
TTL flash came on the scene much later, like with the Olympus OM-2. The Olympus flash also had a tiny sensor that would work if you didn't put it on an Olympus hotshoe.