Your AI Agent Is a Master Programmer With Serious Amnesia (Here’s How to Work With That)

My memory of that same period was patchy at best. Not because of drink or drugs being involved (although some of that may have contributed) but because my memory for details over time just isn’t that great.

Memory is far from straight forward. This is probably why generative AI is having such troubles with it. But why is is so problematic? Why are we inventing all these “solutions” such as “Skills”, “memory.md”, “Context sharing” etc to help our models know what is going on over a long period of time?

Lets take a look.

Your AI Model Has Serious Amnesia

I often see an analogy online referring to their coding AI Agent as a “Junior Developer”. I think that the junior developer analogy isn’t the best. I remember being a junior and I wouldn’t have been able to do even 10% of what generative AI can currently do.

Lets take the Opus 4.6 model. It has an amazing depth of knowledge.

• You need to ask about how to shape some new piece of data you wish to store? Opus will give you good advice.
• You have a bug in a Go program and you can’t work it out? There’s a good chance that Opus can help.
• You need to write some integration tests in a php CRUD app? Opus can assist.

I wouldn’t have been good at any of this as a junior.

Why is this important? Why does it matter? It matters greatly because your mental model of generative AI will severely influence how you interact with it and hence how good its output will be.

The Memory Problem

Generative AI has a big problem with memory and I think anyone who has worked with AI enough has experienced this.

• I would ask an Agent to do a task it has done 10 times before — Create a PR — suddenly it can’t remember where my remote repo is and starts using find and grep everywhere. Even though I actually have a “skill” for this specific task.
• I ask the Agent about something I spoke about just yesterday. It replies as if we’re speaking about it for the first time.

When you have a conversation with Claude.ai, it won’t remember things across different conversations.

This is quite weird if you think about it.

The UI and UX is the same across conversations, the model is the same across conversations, yet it can’t remember what you spoke about in another conversation just 5 minutes ago.

It’s very easy to get sucked into believing Claude is your own personal AI Agent and there will be some natural familiarity after working with it over a long period. Just as you would if you worked with a human over a long period of time. You get used to each others’ ways of working. You naturally fall into a groove together. However, this is not the case with your AI model.

Every new session, it knows nothing. It’s brand new on every convesation. Let that sink in.

How To Work With Your AI Model

Your mental model of your AI Agent has to be accurate so you understand how to utilise it’s strengths and avoid it’s weaknesses. If you think of your AI Agent as a junior then you will treat it like a junior and only assign and ask it specific things.

That’s why I like to think of my AI Agent as a “Master of Programming” — it’s knowledge of syntax, design patterns, algorithms is far superior to what any one human can store in their brain. The sheer scope of what it “knows” about is extremely vast.

As a Software Developer, if you realise this, then you can start to tap into that.

• You aren’t sure if the current implementation of a certain feature is optimal or not. Ask your AI Agent for its opinion.
• You come across some syntax you’ve never seen before. Ask your AI Agent to break it down until you understand it.
• You need to implement a large new feature across multiple existing services and there is a whole bunch of complexity involved. Ask your AI to help you plan it out, write up RFCs, make sure all angles are covered etc.

Tips To Consider When Interacting With An AI Agent

With a good mental model in place you can start utilising this when interacting with your AI Agent. It will aid you when framing your prompts.

Keep Context files in the repo

Keep a CLAUDE.mdfile in your code repo. This can contain things such as architecture decisions, coding conventions, domain terminology, key design patterns and project history.

The balance here is knowing what to include and what not to. If too many things are included in this file then the context gets easily bloated and it doesn’t end up being that useful. If not enough context is included then it won’t prove that useful.

CLAUDE.md files get auto-loaded into the context. If you keep separate files to split out these things: ARCHITECTURAL_CONTEXT.md , DOMAIN_KNOWLEDGE.md etc then these have to be manually loaded into the chat. Just to bear this in mind.

Include specific files if possible when asking for changes

Starting a new session with your AI Agent and then asking something vague — “Please write tests for this new UI Card we just implemented” — will probably work but you are risking the AI agent doing something you didn’t intend.

“We just implemented @UICard.tsx and I would like you to write some unit tests to determine the right UI is showing up depending on various props being passed down. Please use the “unit-testing” skill”. This is likely to work much better, the Agent can find the relevent context much quicker and the test outcome is much more likely to be what you were envisioning.

Yes, this approach takes a bit longer, but over time your codebase will thank you for making that extra effort.

Conclusion

Working with AI seems almost required now in Software. I’m sure there are still a tiny subset that don’t use AI but the large majority are. It’s generally pretty easy to use. Type in a vague prompt in English and sit back and watch the code being generated. However, the way it’s being used varies massively from person to person.

To use generative AI well does take adjustment and your mental model of AI is very important. Your AI Agent is not a Junior. They have a vast amount of knowledge at their “fingertips”.

Your mental model will shape your interactions which in turn shape the output. Don’t be deceived by the familiarity you feel with your Agent. This can lead to assumptions that just aren’t true. Just remember, every new conversation is a blank slate. A completely blank slate. Your Agent has amnesia and it’s memory is never coming back.

Subscribe to My Weekly Updates on Medium!

Enjoyed This Post?

If you found this blog post helpful, why not stay updated with my latest content? Subscribe to receive email notifications every time I publish.

If you're feeling really generous you can buy me a coffee. (Btw, I really like coffee…)

What You'll Get

• Exciting Discoveries: Be the first to know about the latest tools and libraries.
• How-To Guides: Step-by-step articles to enhance your development skills.
• Opinion Pieces: Thought-provoking insights into the world of frontend development.

Join Our Community

I live in the vibrant city of Prague, Czech Republic, with my family. My blog is more than just articles; it's a community of like-minded developers who share a love for innovation and learning.

About me

I'm a passionate Frontend Developer specialising in React and TypeScript. My professional journey revolves around exploring and mastering new tools and libraries within the JavaScript ecosystem.

Check out my LinkedIn and Github if you are interested.