Anthropic was valued at $380 billion in March 2026, yet its own case study and video about its growth marketing team tell the story of one person. Austin Lau, a growth marketer there, had never written a line of code. When the company launched its AI coding tool, Claude Code, he had to look up how to open a terminal, the text window where programmers type commands.
His job needs a constant supply of fresh ads. Each Google search ad calls for 15 different headlines, and the copy needs refreshing every few weeks. Lau used to brainstorm lines in a spreadsheet, count characters by hand, paste the copy into Google Ads and repeat. Image ads meant copying designs in Figma, a design program, and pasting in new headlines across ten or more versions in several shapes.
Within a week of trying Claude Code, he had built two tools. He described each problem in plain words, and the AI wrote and ran the code for him. Before tools like this, he says, a workflow like his would probably have needed a team of engineers to support it.
The two tools turn ad production into a repeatable process. The first, an add-on for Figma, turns one design and a list of headlines into dozens of ad versions with a single click. Building it took him about 45 minutes to an hour. The second asks for campaign data, existing copy and keywords, so its draft headlines and descriptions build on what has already worked. It checks those inputs against Agent Skills, sets of instructions he wrote on Anthropic’s brand voice, product accuracy and best practice for Google search ads. Once he has refined and reviewed the copy, it exports a file ready to upload.
Lau thinks most marketers use AI to speed up single tasks, such as writing copy or brainstorming. In his view, they have not thought through where it could be built into their workflow. Built into his, it cut the time to make an ad from 30 minutes to 30 seconds. The copy tool alone saves him hours a week, and the time saved lets him run more copy experiments to find the lines that perform best.
