My ~€226/month colleague

April 7, 2026

It's the first time since Claude Code in November that I've now found AI again to be extremely helpful.

I work as a data & AI consultant. I know people barf when they hear the word consultant, but it's the best description of what I do. A business has a data problem, or a slow process, and I help them solve that using technology. On top of giving fancy advice I also build stuff.

The problem: as a solo consultant, I run multiple projects at the same time. On average the past couple of months, I have about 5-8 active projects. Active could mean waiting for a day to meet for a discovery session, or spending a couple of days building a tool. Maybe half the time of a project is spent on what I call administrative tasks. Things like sales calls, feedback meetings, or writing a proposal. These things are part of my work, but preferably I spend more time on doing instead of talking about doing. Also, during projects there can be days (maybe even weeks) where a client takes time to respond. It's hard to keep track of what we've talked about and what the current status of all these projects is.

The solution: what I've found to be extremely helpful with these administrative tasks is a mix of Obsidian, Granola, Wispr Flow, and Claude Code. I'll untangle these one by one.

Let me explain how these all work together and how it's helped me cut probably 70% of the administrative stuff.

Claude Code is the assistant I talk or chat with. I can start the week and say, pick up all the projects and give me the last action on each project. It'll tell me what the latest action on each project is and I'll talk into Wispr Flow to share what's the latest. Claude updates the project.

Let's say I have a meeting where I get feedback from a client on how a tool is working for them. I get all the feedback in a Granola transcript, paste that into Claude and it'll update the project status. Even better is when I tell Claude to go into the code repo for that project and actually process the feedback.

From these project statusses I can do numerous things. I can ask it to write an e-mail with a rundown of the latest actions. I can ask it to write a proposal for an expansion of a project. I also have a document that contains my 'north star', the goals I have for this year and the next. I can see how we're doing and what I'm spending my time on.

Additionally, I can now ask Claude what projects I've been working on, and what subjects might be nice to share on social media, for example on Linkedin.

Now I've been adding more tools to this. For instance a read-only email cli. I see I've got some new e-mails from a client. I can ask it to fetch those emails, process the feedback, and maybe even deploy new code. All from one interface.

This is the closest I've come to an 'openclaw like'-assistant that actually provides value for me. It doesn't have access to everything and also isn't proactive, but it's perfect for what I'm working with.

Return to blog