Manus AI is a tool a contact introduced me to a few weeks ago, and after several days of testing I want to record what I have found so far. The experience has shifted how I think about running my businesses with AI agents.

For years my main large language model has been ChatGPT. I turn to Gemini when I am not convinced by an answer, and I use Claude for specific tasks. Inside ChatGPT I built a set of custom GPTs and projects to automate parts of my property business and Operations Director. More recently I added Genspark to the mix, a tool that pulls together several models and agents to handle more complex work. The honest truth is that stitching these tools into a full workflow has never been simple. I use Zapier and Make to connect one step to the next, and those connections break often enough that confidence in the whole system erodes fast.

Manus works differently. Rather than asking me to link a language model to a separate automation tool, the platform holds both together. When I set a task, a coding window opens in the background and builds the workflow needed to complete the job, the same kind of window many people glimpse for a moment when ChatGPT drafts a PDF. Manus runs this step as standard, and once built, the workflow stays available as a small AI agent I call on again later.

I have tested this on a narrow set of tasks so far. Creating a task inside our Airtable operating system is one example. Normally this step takes manual entry or a voice note passed to my assistants, and even with automation in place the output still needs a person to check. With Manus, the task gets built, logged into the right apps, and pushed through to Slack to notify the relevant person, with no manual handling from me at each stage.

The plan I am on costs $40 US a month and includes 8,000 credits. A single complex workflow uses a thousand or two of those credits, so cost management matters. I am not yet sure how the economics settle over a full month of heavier use. What I am confident about is the cost sitting well below hiring an employee, and I expect the cost to sit below a virtual assistant too, depending on where that person is based and what they are paid.

This distinction matters to me because my goal has always been to run 90 percent of my businesses through AI agents rather than people, not to replace my team, but to free them to work at a leadership level rather than repeating administrative tasks daily. Manus looks like a genuine step toward this goal, not because the tool works without flaws, but because the gap between what a person needs to know to use the tool and what the tool produces is smaller than anything else I have tested.

None of this replaces the discipline that got me to day 2075 of this streak. A new tool is only as good as the habit of using the tool consistently, checking output, and correcting course when something goes wrong, and something goes wrong now and then. I go back and tweak instructions, review notifications, and adjust the plan when output is not right. The tool saves time. The tool does not save judgment.

I bring this same discipline mindset back to why I run every day. This mission started as a way to save children's lives through the ultimate ultra marathon, running barefoot-style, on consecutive days, covering the distance of a lap around the world. As of today I have covered 16,009.57 kilometres of the 40,075 kilometre target, with 24,065.43 kilometres remaining. The number matters less to me than the £1 million goal for Great Ormond Street and BBC Children in Need, because every day I show up on this streak and share what I am learning, in running or in business, more people see the mission and more money reaches children who need support.

I do not run to build an audience for its own sake. I run because consistency, shown daily without exception, tends to earn attention that a single big moment rarely does. The same logic applies to how I build my businesses. A tool like Manus AI will not change my mission. The tool changes how efficiently I run the operational side of my life so more of my time and energy goes toward the run, the fundraising, and the family time I protect as carefully as the streak itself.

I will keep testing Manus over the coming weeks, on tasks like reconciling financial transactions and managing tenancy paperwork, and I will report back honestly on where the tool holds up and where cost outweighs benefit. For now, day 2075 stands, the distance grows one day at a time, and the mission behind the distance stays the same as day one.

If you want to follow progress, in running or in the AI experiments alongside the running, subscribe and watch the next episode. If you believe in saving children's lives through a mission built on daily consistency rather than a single grand gesture, share this with someone who has not seen the mission yet.

Watch the full episode: https://youtu.be/1WdR932ntsc