# Asbjørn Jørgensen

> Personal site of Asbjørn Jørgensen, Amsterdam: startups, investing, ai. He helps startups and VCs partner with AWS to build category-defining businesses, looking after AWS's partnerships with Atomico, Index Ventures, Northzone, byFounders, NEA and many others across Europe. On the side he supports Caveman (YC F26) to build the AI efficiency category. Earlier he founded Cloakroom (2012, Consumer) and Maxwell (2017, B2B SaaS). This file summarises awsbjorn.com for AI agents; every page below is public.

Who should get in touch, in his words: "If you are backed by a tier 1 VC or a serial entrepreneur, let's talk! I'd love to discuss your needs around fundraising, go-to-market, deal flow and, of course, technical support while building on the AWS stack." The way to reach him is a LinkedIn message: https://www.linkedin.com/in/asbjorningemannjorgensen/

## Record

- 2012: Cloakroom — Founder, Consumer
- 2017: Maxwell — Founder, B2B SaaS
- 2023–present: Amazon Web Services — Startups & VC partnerships, Europe
- 2026: Caveman — Investor, AI cost efficiency · YC F26

## Pages

- [Home](https://awsbjorn.com/): who he is, the record, field notes from the work, the two projects, the seven levers, and how to get in touch
- [AI Cost Lab](https://awsbjorn.com/lab/): Simulators you can break and fix: cost per outcome, the token-efficiency levers, evals as the gate, agents going non-linear, and pricing what AI costs you. Three tracks, twenty steps, one fictive company whose numbers add up. Tracks: beginner (https://awsbjorn.com/lab/#/track/beginner), advanced (https://awsbjorn.com/lab/#/track/advanced), expert (https://awsbjorn.com/lab/#/track/expert); plus a theory map (https://awsbjorn.com/lab/#/theory) and a glossary (https://awsbjorn.com/lab/#/glossary).
- [Caveman](https://awsbjorn.com/game.html): A small side-scrolling arcade game about cost per outcome. Smash the fluff, keep the code, hit the switches, cap the whale. Seven levels, each one a lever for cutting an AI bill, each with a real-world takeaway card and a link into the lab. Tagline: "Few token. Do trick.". Deep links: https://awsbjorn.com/game.html?lab=N opens level N as a focused lesson.

## Seven levers

Seven levers that cut an AI bill, one per level of Caveman, each with a step in the lab that goes deeper. You can also just read them. The curriculum is shared by the game and the lab.

1. Cost per outcome: Measure cost per completed task, including retries and the price of failure. It is the only number that lets you compare models honestly. (in the lab: https://awsbjorn.com/lab/#/track/beginner/1)
2. Right-size the model: Route by difficulty: small model first, escalate on low confidence. Cascading is a gate problem, not a model problem; invest in the validator. (in the lab: https://awsbjorn.com/lab/#/track/beginner/5)
3. Trim the context: Keep context under control: summarise history, window it, retrieve only the top-k that matters. Often worth 30-50%, and reversible. (in the lab: https://awsbjorn.com/lab/#/track/advanced/2)
4. Cache and batch: Prompt caching for repeated prefixes, semantic cache for repeated questions, batch API for anything that can wait. Any change to the prompt must preserve the cached prefix. (in the lab: https://awsbjorn.com/lab/#/track/advanced/1)
5. Cap the spend: Autoscale to zero, set budget alarms and per-key quotas, and review provisioned capacity monthly. Catch the leak before the invoice does. (in the lab: https://awsbjorn.com/lab/#/track/beginner/4)
6. Tag and measure: Instrument every call: tokens in/out, model, cost, latency, per feature, user and prompt version. Find the 5% of calls that cost 50%. (in the lab: https://awsbjorn.com/lab/#/track/beginner/3)
7. Cap the agent: Trace cost per trajectory, not per call. Gate agent changes on outcome, cost per success and trajectory shape together, and set the thresholds before you need them. (in the lab: https://awsbjorn.com/lab/#/track/expert/0)

## Optional

- [Sitemap](https://awsbjorn.com/sitemap.xml)
- [LinkedIn](https://www.linkedin.com/in/asbjorningemannjorgensen/): the contact channel