Petteri Jalonen

Petteri JalonenNov 2, 2026

Looking for a freelance videographer (or creative agency) in the UK! Would love some recommendations 🙏 London-based is ideal, but I’m happy to hear about people elsewhere in the UK too. Details below ⬇️ I’m planning to shoot a couple of customer case studies at Focal before the end of the year. The job: ✍️ Script review and help with the creative direction 🎥 Filming interviews at the customer’s office, with good quality video and audio 🎥 Some B-roll of the team and the office 💻 Editing into a finished case study video (we’ll do the clipping for social ourselves) If you know someone who’s great at this, or it’s you, tag them in the comments or send me a DM. It’s a bonus if they’ve filmed customer stories or interviews for tech companies before. P.S. We’d need full rights to the material, including the raw footage.

Petteri JalonenOct 6, 2026

Working outside in Finland in October sounds like a questionable idea for a knowledge worker, but it's been one of my favorite fall rituals for years. Just grab a sweater, a coffee thermos, and your laptop, and it's surprising how enjoyable it can be. It also does wonders for thinking when you're not just stuck indoors for all of your working hours. Anyone else with a desk job doing the same?

Petteri JalonenAug 31, 2026

If I had to build an AI system for marketing from scratch right now, here’s how I’d do it. It’s kind of like going to the gym 🏋️💪: Getting sets and reps in matters much more than spending weeks or months researching the most perfect thing that maximizes your gains. It honestly doesn’t matter that much if you go with ChatGPT or Claude, or even something else. (People who really strongly push one way or the other are probably selling one of them one way or another). Yes, there probably are (mostly marginal) differences with one over the other in specific use cases. But what improves the quality of the work MUCH more is the context layers. These are the files and systems that give deeper understanding and explanations about your business to your LLM. All the way from basic things like your brand guidelines and ICP to advanced stuff like full understanding of all of your visual material. You can build these layers in different ways:  - .md files hosted in Github - Documentation in Google Drive - LLM skills - Specialized tools and plugins Start with the basics, you can build the fancy stuff later. One specific config over the other matters less than information being clear and accessible to LLMs and people (and C, that some safety guardrails exist so nobody can do anything catastrophic). Next, map out the most critical use cases: e.g. creative production, competitor analysis, SEO and GEO, ads management, etc. Start looking at specialized tooling that can add data and capabilities on top of the general purpose models to serve these needs. Such as: - Creative asset management  - Project management - Image and video creation and editing - Copy drafting.  - Notetaking and call analysis. Trying to have MCP/API capabilities as much as possible is crucial for reducing platform and context switching. It also more easily adds information to your context layer so the knowledge deepens. 📣 Lastly, get people sharing and talking! No matter the setup, individuals will use it in different ways and there’s so much people can learn from eachother. Set up regularly occurring knowledge shares, highlights in weekly meetings, etc. Here’s how my stack currently looks: - Claude Code as the main platform, with most of the context layer as a GitHub repository. - Specialized tooling on top: Circleback for call recording, Focal for creatives, project management and some analyses, Higgsfield AI for image editing/generation (mostly via MCP), Hyperframes for motion graphics, Peec AI for LLM visibility tracking,  Descript for videos, Meta Ads MCP.