Day 10 of 28 · OpenClaw Challenge
Memory — It Remembers You
⏱ 7 min
📊 Beginner
Your agent can read email and manage your calendar. But right now, every conversation starts fresh — like meeting a stranger each time.
Today you'll learn about memory — the feature that turns a useful tool into a personal assistant.
The problem without memory
Imagine hiring an assistant who forgets everything about you every morning. Every day you'd have to re-explain: your name, your team, your preferences, your work style.
That's what an agent without memory is like.
Sure! A few questions:
— What team?
— What time do you prefer?
— How long should it be?
— Which calendar should I use?
The engineering team. Morning. 30 minutes. My work calendar. We've done this three times already...
Sorry about that! I don't have memory of our previous conversations. Let me set it up now.
↻ Replay conversation
That's frustrating. Now watch what happens with memory enabled:
Done ✓
📅 "Engineering Team Sync" — Thursday 9:30am, 30 minutes
👥 Invites sent to your usual 6 team members
📝 Standing agenda template added
I used your preferences: morning slot, work calendar, same team as last time. Anything you'd like to change?
↻ Replay conversation
Knowledge Check
What's the difference between these two interactions?
A
The second one used a better AI model
B
The second one had memory — it remembered the user's team, preferences, and patterns
C
The second one had more tools connected
D
The user gave a more detailed prompt the second time
Same request, same tools. The only difference was memory. The agent remembered who's on the engineering team, that the user prefers mornings, 30-minute meetings, and the work calendar. No re-explaining needed.
What the agent remembers
Memory in OpenClaw works on two levels:
Short-term memory — what happened in this conversation. The agent remembers what you discussed, what tasks it completed, and what's still pending. This resets when the conversation ends.
Long-term memory — what the agent learns across conversations. Your name, your team, your preferences, your work patterns. This persists forever (until you clear it).
When you say "I prefer morning meetings," the agent stores that in long-term memory. Next time you schedule, it remembers — even if it's weeks later.
Knowledge Check
If you tell your agent "I don't like meetings after 4pm" and then start a new conversation next week, will it remember?
A
Yes — it stores preferences in long-term memory that persists across conversations
B
Only if you write it in your custom instructions manually
C
No — everything resets when the conversation ends
D
Only if you remind it at the start of the new conversation
Preferences like "no meetings after 4pm" go into long-term memory. The agent will remember this in every future conversation until you tell it otherwise. That's the power of persistent memory.
Memory in action
Over time, your agent builds a detailed picture of how you work:
What do you know about me?
Here's what I've learned about you:
👤 **You**: Alex, Engineering Manager at TechCorp
👥 **Your team**: Sarah, Marcus, Priya, Jake, Liam, Nina
📅 **Scheduling**: Prefer mornings, no meetings after 4pm, Fridays are meeting-free
📧 **Email**: Professional but warm tone, always CC your manager on client emails
💻 **Work style**: Deep focus blocks in the afternoon, standup at 9am
🌍 **Location**: San Francisco, PST timezone
Anything you'd like to update?
Actually, Nina left the team last week. Remove her.
Got it — Nina removed from your team list. I'll stop including her in team meeting invites and group emails. ✓
↻ Replay conversation
Knowledge Check
When the user said "Nina left the team," what did the agent update?
A
The team list and all future behaviors — meeting invites, group emails, and team references
B
Just the next meeting invite
D
Nothing — it just acknowledged the message
Memory updates ripple through everything. Removing Nina from the team list means the agent won't include her in future invites, group emails, or references. One correction updates all downstream behaviors.
You're in control
Everything the agent remembers, you can see, edit, or delete. You can say "forget everything about my schedule preferences" or "show me what you remember about my team" at any time.
Memory makes the agent personal. But you decide what's personal.
Final Check
What makes memory the feature that turns a tool into a personal assistant?
A
It lets the agent learn your preferences, patterns, and context — so you never have to repeat yourself
B
It lets the agent store more files
C
It increases the agent's processing speed
D
It makes the AI model smarter
Memory is what makes the agent yours. Without it, you have a powerful but impersonal tool. With it, you have an assistant that knows how you work, what you prefer, and who matters to you.
🧠
Day 10 Complete
"Your agent remembers you now. It's not just a tool anymore — it's your assistant."
Tomorrow — Day 11
Web Browsing & Research
Your agent knows you now. Let's teach it to explore the internet on your behalf.