AI trims work hours—then expands them again

AI compresses – Six tech workers across Amazon, Google, and Apple-linked contracting, plus a logistics startup, describe how AI can compress tasks into minutes—documents, meeting notes, code review, and reporting pipelines. But they also say the saved time often gets redirect

For Priyanka Devi Ramesh. the change feels immediate: a document that used to take well over an hour now takes about 15 to 20 minutes. She’s a business intelligence engineer at Amazon. 30. living in Virginia—and her days have started to revolve around how quickly an AI tool can turn her thoughts into polished technical or customer-facing writing.

Her colleagues describe a similar speed bump in their own areas. At Google. a security engineer says what once took one to two hours to summarize six months of meetings now takes five to 10 minutes. A data scientist at Amazon says a monthly stakeholder report that used to consume 8 to 10 hours over a couple of days can collapse into a button click—after a heavier upfront stretch.

Yet in these interviews, the same pattern keeps showing up: faster output doesn’t always translate into less work. For some, the hours saved get poured into the next problem. For others, AI creates a sprint to build and validate automation systems, turning time-savings into short-term time costs.

Priyanka Devi Ramesh said AI has dramatically sped up document writing at Amazon using an AI tool called Pippin. She said she can spend hardly 15 to 20 minutes max to write and finalize a document that would have previously taken her well over an hour. On the technical side, she said she uses Kiro for brainstorming ideas and making logic updates in minutes. She also said she is building agents within Amazon Quick to automate common customer questions about dashboards and to surface insights from data.

Even with the speed, she said the time AI saves gets reinvested into the next problem. She said the team is constantly looking for ways to clean up messy data and finding opportunities to automate wherever possible—so time saved in one area gets reinvested into the next problem.

Prerit Pathak at Google offered a different kind of compression: meeting time translated into quick summaries. He is a security engineer at Google, 27, living in New York City. He said he uses Gemini for a variety of purposes, but recently it has bolstered his note-taking. He said he used to take shorthand notes during meetings to record interesting or important information. but now he lets Gemini take notes on his work calls. He said a summarization task—understanding what happened over the previous six months—once would have taken one to two hours and now takes five to 10 minutes.

At Amazon, Sarthak Gupta framed the transition as both promise and burden. He is a data scientist at Amazon, 29, living in Seattle. He said AI has been most helpful with building end-to-end automation pipelines for recurring workflows. He described a monthly stakeholder report that used to take 8 to 10 hours over a couple of days to create. That report, he said, involved pulling data, cleaning it, generating visualizations, and writing the summary.

Now, he said an AI pipeline handles the data pulls, transformations, and dashboard refreshes. He said he spends maybe 45 minutes reviewing the output and adding context before sending it out. But he also said his overall working hours are running longer than normal because his team is in the middle of an automation phase. He said building the pipelines. integrating the AI tooling. validating outputs. and onboarding all of this into existing workflows is front-loaded work. and that upfront investment is real. He added that the payoff comes later. when the same task that took a couple of days collapses into a button click. In the short term. he said. AI is actually adding hours to his week. not subtracting them—and he expects that to flip once the foundational pipelines are stable and the automation is doing the heavy lifting on its own.

Tanvi Pisal described speed as a lift in the earliest stages of design. She is a UX designer working as a contractor for Apple via Red Oak Technologies, 29, living in San Jose. She said one of the biggest ways AI saves her time is in early-stage product thinking and documentation. She described drafting product requirement documents. brainstorming user stories. mapping edge cases. outlining use scenarios. and refining ideation before reaching visual design.

Now, she said she can start with rough notes or a messy draft, and AI helps turn that into a much more structured document in minutes. She said what used to take her three to four hours can often be reduced to 30 minutes with feedback and refinements.

Udit Mehrotra said writing product documents is where he’s seen the biggest change at Amazon. He is head of product at Amazon, in his 30s, living in Seattle. He said every major initiative at Amazon starts with a written document. and for years the first hour or two of that process was building scaffolding—setting up the structure. filling in sections he knows by heart. and building something worth reacting to before getting to the actual thinking.

He said AI can now help him input the customer problem and constraints and get a solid first draft in minutes. He also said what surprises him is that it’s often more comprehensive than what he would have written on his own under time pressure. Still. he drew a boundary around the real work: he said getting from 80% to 100% is where the real work lives. and AI doesn’t change that. He said the

strategic judgment. the tradeoffs between what customers need and what’s technically feasible. and the decisions requiring years of accumulated context about a specific customer problem still take the same depth and care. What has changed. he said. is that he arrives at the starting line faster. with a more complete structure to react to and push against. He added that the quality of the final document is often better as a result—not because AI did

the hard thinking. but because he spends more of his time on it.

In a startup environment, Iren Azra Zou said the feeling is even more dramatic. She is a software engineer at the trucking logistics startup Double Nickel, in her 20s, living in New Jersey. She said she uses AI, mostly Claude Code, for the majority of her coding. She said it is hard to quantify time savings. but it feels like what used to take a week can now take a day.

She also said the team relies heavily on AI to review and provide feedback on code. unless a change is particularly risky. She said that saves a huge amount of time: instead of waiting days for human reviews. you get multiple rounds of feedback within hours. She added that it also means she spends less time reviewing others’ code. which she said probably saves her several hours each week. She acknowledged tradeoffs—she said less human review can have downsides—but said that right now the speed of iteration and innovation is incredibly valuable for the company.

Across these accounts. the same tension plays out in different roles and tools: AI can shrink the time required for specific tasks—writing. summarizing. building. coding. reviewing—but the total workweek doesn’t necessarily shrink with it. Priyanka Devi Ramesh said time saved is redirected to cleaning messy data and finding new automation opportunities. Sarthak Gupta said short-term hours can increase while automation pipelines are built. integrated. validated. and onboarded—before the longer-term payoff arrives.

The interviews also point to a corporate reality beyond individual workflows: the people using AI are often doing it at moments of high pressure—layoff. resignation. job search. or shifting workplace expectations—what the interviews describe as a “career crossroads” for these workers. Business Insider asked readers to share their story by filling out a form. It also provided a direct contact email—jzinkula@businessinsider.com—and a Signal handle, jzinkula.29.

AI at work Amazon Google Apple Red Oak Technologies Double Nickel Pippin Kiro Amazon Quick Gemini Claude Code automation pipelines productivity software engineering job search workplace expectations