How to keep the level of your work the same across your team with simple automations. Delegating and the quality is slipping? Get clear on how the work should be done, then build an AI agent to guide your team through it.

Your business is growing, clients are coming through the door like freshly baked bagels, you can’t wait to hire more help and start delegating tasks.

But then you start handing off tasks, only to realise they’re not done to the standard you always offered your clients. The quality starts slipping, and you find yourself having to correct everything and constantly leave feedback for improvements. This was supposed to be easier on you, and yet things seem to be coming back for your final check.

What’s usually going on underneath

It’s natural to want things to be of a certain standard. A standard that you worked very hard to establish for a very long time. That’s how you’ve grown your reputation and it’s what you promise your clients.

The standard usually lives in your head, and you haven’t communicated that clearly to your team, or you haven’t set up any guardrails or outlined the quality of work that needs to happen. Quite often the team members are left to figure it out on their own: you give them the task, think they’ll be able to produce something, and they tend to deliver, but very often it’s not to the standard you had in mind.

Wanting to keep that standard is about making sure things transition smoothly for the client, and you can sleep soundly knowing work carries on as it did before. When you start delegating, there will always be a period of adjustment and teaching. But this initial process can be easier to execute, rather than causing additional stress and work for both you and your team.

What might help

Before you proceed to actually start handing tasks over, get very clear on the process of the task. Create a framework for yourself, getting very clear on how things should be done, in what way or format should the work be delivered, what are the considerations you take into account while executing that task.

Then use tools like ChatGPT or Claude to create an agent to guide someone else through this same process. When I say an agent, I mean a set of skills (so instructions for your AI tool) and an interactive artifact that you can share with the other team members.

The skill, which is written instructions for your AI partner, will contain your personal framework of how you’d like that specific task to be executed. The artifact is the interactive part that will guide the person on what they need to research, assess, what things they need to consider, and what deliverable they need to create with what branding guidelines.

What’s in a skill

A skill is a set of written instructions that you can give to any AI tool, such as Claude, ChatGPT or Manus, which is up and coming in the AI space as well. Most of the AI tools use the exact same infrastructure, so when you work in Claude and use a lot of skills, because they are essentially written files, you can easily migrate them across to ChatGPT, Gemini or Perplexity.

For example, if you want to set up an onboarding pack for your own client, you can work with your AI tool to guide it through the process that you currently do, by explaining the steps that you always take. In some cases, like with Claude, you can record a video of your actions and Claude will then translate that into written instructions for itself.

A rough idea of what an onboarding pack should involve:

→ A template for your contract

→ An outline of your payment terms

→ Your invoice template

→ A list of actions that you tend to take every time a new client comes through the door

→ Things that each client has to always provide you so you can start doing the work

A real example

Currently I’m working on an AI agent that will help my client and her team of five unify the reporting that they’re doing as a team and as a business for their clients.

At the moment, for example, each report that is created at the end of the month about a social media account’s performance is quite often different. It tends to follow the same overall arc but deliver the information focusing on different aspects of certain things, presenting it in different formats. The design of the reports is all over the place currently because everyone is doing what they think is best.

We are currently building templates that the AI agent can refer to and use every time it produces a specific report and that will deliver the same results across the entire team and the entire business.

The takeaway

What this creates is an AI agent that monitors the quality of work for you, without you having to control the output and workflow of someone else, freeing more time for you to focus on bigger things that move the whole business forward.

Of course, it’s sensible to build in a safety check right before the final product needs to be sent off to the client, but this way it will be much closer to what you would do yourself, and ensure the same standard across your business.

About the author

Jagoda Puczko is a business operations partner who helps growing founders close their laptop at 6pm. She has run operations for teams of 100+ at Apple and Little Dot Studios, directed award-winning films, and once had a headshot photography business with a very sophisticated pricing strategy.