Talk about your next project
Together AI case study

The lesson

Lead with the customer’s operating numbers, then explain the infrastructure choices that made them possible.

The move

The story opens with the Washington Post mark, the title, and three large proof points: “2 seconds,” “1.79B,” and “Thousands.” A client sidebar names the segment, industry, highlights, and summary before the page moves into “About The Washington Post,” “The Challenge,” “The Solution,” and “Results.” The body lists critical requirements, rejected alternatives, Together AI’s model inference infrastructure, and a quote from Sam Han. It ends with “Enterprise Impact Delivered” and a “Performance Comparison” table comparing Together AI, proprietary APIs, and self-hosting.

Why it works

  • Our readThe first screen gives the reader the response time, token volume, and query scale before asking them to read the infrastructure story.
  • Our readThe challenge section separates requirements from failed alternatives, so the buyer can see why the customer did not choose proprietary APIs, hyperscalers, or self-hosting.
  • Our readThe comparison table turns the case into a buying aid by mapping control, performance, and cost model across the available approaches.

What not to copy

Don't copy The Washington Post’s journalism use case, Ask The Post AI, or the Llama and Mistral model choices. Keep the structure of top-line operating proof, customer context, rejected alternatives, solution components, customer quote, results, and a comparison table.

More from Together AI

More case studies

Steal it for your brand.

Tell us what you want to make. We’ll talk through how to put these ideas to work for your brand.