The Model 3 is a supremely important product for the future of Tesla. It’s what CEO Elon Musk says will help the company get out of the red, and it could also be the first widely adopted all-electric car. And on the eve of one of the most important deadlines in the company’s 15-year history, the company is still figuring out the best way to build it.

Today marks the end of the second financial quarter of 2018, and it’s also the deadline that Musk set for Tesla to make Model 3s at a rate of 5,000 per week. Tesla needs to reach this rate of production so that it stops losing money on every Model 3 it makes. Once it reaches this rate, the company can climb toward the eventual goal of bringing in a 25 percent margin on each of the sedans.

Speculation flourished as the deadline approached — some traders believe the company won’t make it and have bet heavily against its stock, analysts are largely split, and many fervent fans don’t particularly care. Musk has remained confident that the company will reach the pace he demands, and over the past two weeks, Tesla granted factory tours and interviews with the CEO to the The New York Times and the The Wall Street Journal.

The tours could be seen as a small olive branch to the media, which Musk often criticizes, or simply an effort to show the world (including analysts and investors) “everything’s under control, situation normal” ahead of the end of the quarter. Whatever the reason for the access, though, details in both pieces show that — even after Tesla manufactured and shipped tens of thousands of Model 3s to customers since production began almost a year ago — the company is still changing the way it makes its most important car.

Those changes can be big, like the new tent-covered assembly line that Tesla recently constructed in one of its parking lots at the Fremont, California factory. But they’re often much smaller, like the spot welds that hold the car’s frame together, for example. According to the Times, Tesla executives approved an idea from the company’s engineers to have robots make about 300 fewer “unnecessary” welds in order to save time on production.

Robots still play a big role in production of the Model 3, even though Musk has admitted to an over-reliance on automation. But the company is apparently still trying to figure out how to best use those robots. In some cases, that involves tweaking very small things, like which bolts the robots are using:

For months, Tesla engineers struggled to get a robot to guide a bolt through a hole accurately to secure part of the rear brake. They found a maddeningly simple solution: Instead of using a bolt with a flat tip on its threaded end, engineers switched to a bolt with a tapered point, known as a “lead-in,” that can be guided through the hole even if the robot is a millimeter off dead center.

In other cases, though, Tesla is pushing the robots past their prescribed limits:

 Tesla at times pulls robots off the line and tests them operating at speeds greater than specified by the supplier, said Charles Mwangi, Tesla’s director of body engineering.

“We are actually breaking them to see what the maximum limit is,” Mr. Mwangi said. The idea is to find ways of accelerating production without spending capital on new machinery. In the future, rather than adding more machines to increase output, “we can just dial up our equipment,” he said.

Meanwhile, the WSJ says in its factory tour report that Tesla spent between $80 million and $90 million in an “automated warehouse system” that was supposed to send parts to specific workstations around the factory. When the gambit ultimately failed, parts of that conveyor system became some of the “scrap” Tesla used to create the assembly line under the tent outdoors.

That Tesla is still making changes to the way the Model 3 is built after some 30,000 have rolled off the line is emblematic of how the company operates differently compared to other automakers. At Ford, or GM, or Mercedes-Benz, decisions like these are often made during the prototyping phase, and the results of those choices are sussed out via rigorous testing long before cars wind up in the hands of customers.

It’s a daring choice for a company that has had issues with quality control. And results from changes made now might not surface until thousands more Model 3s are on the road, well past the moment where we know if the company hit Musk’s marks. It’s also likely to be another dividing line. Experts may say it’s a sign that Tesla is still learning on the fly, while supporters of the company will say it proves a willingness to embrace new ideas.

A spokesperson for Tesla declined to comment on the changes to the production process of the Model 3. But Tesla is certainly aware of the risks associated with the quality and reliability of its cars. One of the many risk factors included in the company’s most recent quarterly filing with the SEC addresses this quandary head-on. “While we have performed extensive internal testing on the products we manufacture, we currently have a limited frame of reference by which to evaluate detailed long-term quality, reliability, durability and performance characteristics of our battery packs, powertrains, vehicles and energy storage products,” the company wrote. “There can be no assurance that we will be able to detect and fix any defects in our products prior to their sale to or installation for consumers.”