Matt Wood’s AI agent picks his outfit from a 4 a.m. calendar check
Matt Wood’s AWS Quick agent checks his calendar at 4 a.m. and selects his outfit. He also says he processes almost a billion AI tokens in a week while agents aid research and meetings.
By Nathan Brooks
3 min read

What's News
- Around 4 a.m., AWS Quick checks Wood’s calendar and selects the day’s dress code.
- A red closet light signals a keynote day, while blue signals a builder day.
- Wood typically drinks about six cups of coffee by noon and considers himself “useless” before the first three.
- Wood uses AI for meeting preparation and says a “large crew” of agents helps develop ideas.
- Wood says he processed almost a billion tokens in AI usage in one week.
By Nathan Brooks
At around 4 a.m. every day, Matt Wood’s AI agent checks his calendar, identifies the day’s most high-profile meeting and chooses his dress code. The AWS chief AI and technology officer said the use has become “somewhat famous internally” and drawn attention from Amazonians.
Wood said he has AI systems working on his behalf before he even wakes up, making the wardrobe routine one visible part of a broader personal setup. He leads technology for AWS, the Amazon division that provides cloud and AI services to millions of organizations.
AWS Quick connects to Wood’s calendar. Wood described the agent as a tool for hardcore work, including monitoring regulations and spotting supply chain inefficiencies. He told Business Insider at Amazon’s headquarters in Seattle that the overnight workflow starts before his first coffee.
What does this change for teams putting agents to work?
Wood’s system turns calendar data into a physical instruction. Every morning around 4 a.m., AWS Quick checks the schedule, selects the meeting with the most visibility and tells Wood the appropriate dress code.
Depending on the most visible event, such as a keynote, an interview or “hanging with the team,” the agent can signal red for a keynote day or blue for a builder day. Wood explained the color mapping in a LinkedIn post. The smart bulb in his closet lights in that category color, while all his clothes are arranged with color-coded tags.
For a keynote, Wood can grab an item with a red tag from the closet, such as a dark suit. “No thought necessary,” he said.
Wood also said he typically has about six cups of coffee by noon and considers himself “useless” before the first three. The automated choice removes one early decision, before coffee and before he has to weigh comfort against how prominent the day’s meeting may be.
His priorities are explicit. “I never want to be overdressed,” Wood told Business Insider. “I never want to be uncomfortable, and I definitely don't want to make a decision on what I'm wearing before I've got my coffee.”
His daily routine extends well beyond clothing. By the time Wood has his first cup of coffee, he said he scans a report from his AI researcher that compiles an overview of news that broke overnight. The report draws information from multiple sources, including X and Amazon communications.
He also uses AI to prepare for meetings throughout the day. Wood said a “large crew” of agents helps build out whatever ideas he develops during the day.
“Just this week, I've processed almost a billion tokens in AI usage, building things that I'm interested in or to help me with some of these other things,” Wood said.
Wood’s account makes the practical unit of an AI system more concrete: a connected set of inputs, recurring decisions and outputs. Calendar events feed the wardrobe routine; regulations and supply chain data feed regulatory and operational monitoring; overnight news feeds his first reading of the day.
Wood also places meeting preparation and idea development alongside those information feeds. His account shows research, planning, creation and a physical task inside one personal workflow.
Wood’s account points to a lesson for agent teams: define the recurring decision before choosing the model. Wood’s system starts with calendar events, overnight information and work categories, then ends with a dress choice, meeting preparation or a build. His almost one billion tokens show scale, while the operating logic stays simple: remove the next decision from the user’s day.
The near-term question for agent builders is whether they can preserve that 4 a.m. handoff as more inputs and users enter the workflow. Wood’s version already spans research, regulatory monitoring, meetings, ideas and the closet, with the calendar acting as the trigger.
Source: feeds.businessinsider.com
More from Nathan Brooks
Show full bio
News editor covering regulation, policy and their impact on entrepreneurship, startups and venture funding at Business Bearings.
2,460 articles