Portal by Spotify cut my Claude Code token usage by 90%
What happened
Spotify engineers built an internal tool called Portal that streamlines how they send prompts to Anthropic’s Claude Code model. By restructuring prompts, caching reusable parts, and compressing code snippets, Portal slashes the number of tokens sent to the model by about 90%, cutting costs and speeding up response times. The author shares the design patterns and open‑source snippets they used to achieve this efficiency.
Why it matters
Portal shows that clever prompt engineering and tooling can dramatically reduce LLM operating expenses, encouraging other companies to invest in similar token‑optimisation layers. This could accelerate broader adoption of code‑focused LLMs by making them more cost‑effective at scale.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from Anthropic
Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-FreeAnthropic · 14 Nov 2026Claude designed FPGA scaler board for 6th-gen consolesAnthropic · 13 Nov 2026I asked Claude Fable 5.2 to turn my site into a 90s Silicon Graphics workstationAnthropic · 13 Nov 2026Anthropic is endorsing SB 53Anthropic · 13 Nov 2026Understanding and addressing AI harmsAnthropic · 13 Nov 2026This page was collected from engineering.atspotify.com and summarised automatically by AIPulsee's pipeline. See how it runs.