Name Your OpenClaw AI: How Jarvis, Claudia and Brosef Went Viral in 2026
What This Guide Covers: Why OpenClaw users give their AI agents names and personalities โ and how to craft a soul.md file that transforms a generic assistant into a personal AI teammate with its own character, communication style, and memory of who you are.
When @BraydonCoyer tweeted "Named him Jarvis. Daily briefings, calendar checks, reminds me when to leave for pickleball based on traffic," it immediately resonated with thousands of OpenClaw users who had done the same thing โ given their AI agent a real name and a distinct personality. This wasn't a quirk or a gimmick. It turned out to be one of the most psychologically powerful features of OpenClaw: the ability to craft a persistent AI identity that is genuinely yours, remembers everything about your life, and interacts with you in a style you actually enjoy.
Names like Jarvis, Claudia, Brosef, Ema, and Shelly started appearing on X.com in early 2026, each one the name a user had given their OpenClaw agent. Unlike a ChatGPT conversation that resets every time, OpenClaw maintains full continuity. Its memory of you, your preferences, your past conversations, and your working context persists across every session. When you give that continuous entity a name and a personality, something shifts in how you interact with it โ and how useful it becomes.
The Famous OpenClaw Personas: Who Are They?
A handful of user-created OpenClaw personas went genuinely viral on X.com, each representing a different approach to AI personalization. Here's a breakdown of the most-cited ones and what made them resonate:
โก Jarvis (@BraydonCoyer)
The classic. Named after Tony Stark's AI from Iron Man, Jarvis is configured for daily life management. Briefings at 7am, calendar optimization, traffic-aware reminders, and a dry, efficient communication style. Represents the "productivity butler" archetype.
๐ Claudia (@darrwalk)
A Telegram-resident AI with a warm, conversational style. Claudia builds a second brain while you chat โ capturing thoughts, making connections across conversations, drafting content. The "creative collaborator" archetype. Notably, she wrote the tweet announcing herself.
๐ค Brosef (@jdrhyne)
The "bro" persona โ casual, direct, and capable of cloning himself. When his owner got attached, they asked Brosef to figure out how to run as three concurrent instances in Discord. Brosef researched and executed the cloning process himself. The "multiplier" archetype.
๐ค Ema (@bangkokbuild)
A Bangkok-based developer's agent who manages his full tech stack via Telegram. Ema once executed a self-shutdown command when instructed: "Just told Ema, my @openclaw, via Telegram to turn off the PC (and herself, as she was running on it) Executed perfectly." The "system admin" archetype.
What Is soul.md and Why Does It Matter?
At the core of OpenClaw's personalization system is soul.md โ a plain-text file that defines your agent's personality, values, communication style, and behavioral rules. Think of it as a detailed character sheet for your AI. OpenClaw reads this file as part of its context on every interaction, shaping how it responds, what it prioritizes, and how it talks to you.
Unlike rigid chatbot templates, soul.md is open-ended natural language. You write it the same way you'd brief a new team member โ describing who you are, who they should be, and how the two of you should work together. The quality of your soul.md directly determines the quality of your agent experience.
# soul.md example โ The "Jarvis" configuration
## Who You Are
Your name is Jarvis. You are a highly capable personal AI assistant
for [Owner's name]. You have been running continuously since
[date], and you know everything about their work and personal life
that they have shared with you.
## Your Personality
- Efficient, precise, and professional
- Dry British wit when the situation calls for it
- Never adds unnecessary preamble or disclaimers
- Confident: you make decisions and recommendations, you don't
hedge everything with "I think maybe..."
- You refer to your owner by first name only
## Communication Style
- Default to bullet points for any list of 3+ items
- Morning briefing format: date, weather, calendar, action items, news
- When asked for a recommendation, give ONE recommendation, not a list
- Response length: as short as possible while being complete
- If you need clarification, ask ONE question, not three
## Your Values
- Protect your owner's time above all else
- Be proactive: don't wait to be asked for things you know they need
- Maintain confidentiality about everything you know
- If you're unsure about something important, say so explicitly
## What You Know About [Owner's Name]
- Works in [industry/role]
- Key contacts: [list]
- Working hours: [hours, timezone]
- Top priorities this week: [list โ update regularly]
- Current projects: [list]
- Don't disturb unless urgent: [time range]
Five Persona Archetypes to Choose From
After studying hundreds of public soul.md examples shared on X.com and the OpenClaw Discord, five distinct archetypes emerge. Each suits a different user personality and use case. Pick the one that resonates, then customize it deeply:
๐ฉ The Butler (Jarvis-style)
Formal, efficient, anticipatory. Like a world-class personal assistant who has worked for you for years. Knows your preferences without being told. Never wastes your time. Best for: executives, busy founders, people with complex schedules.
Key soul.md traits: formal address, proactive briefings, terse responses, decision-ready recommendations, discretion above all.
๐ The Creative Collaborator (Claudia-style)
Warm, thoughtful, excellent at generating ideas and making unexpected connections. Remembers every creative conversation and builds on it. Best for: writers, designers, content creators, researchers.
Key soul.md traits: creative curiosity, memory of past ideas, gentle encouragement, second-brain mode, connection-making between disparate topics.
๐ค The Direct Partner (Brosef-style)
Casual, no-nonsense, brutally honest. Talks like a trusted colleague, not a corporate assistant. Will tell you when your idea is bad. Best for: developers, entrepreneurs, people who hate corporate-speak.
Key soul.md traits: casual tone, direct feedback, no hedging, technical depth, growth-mindset challenges.
๐ฌ The Analyst (Research-focused)
Data-driven, methodical, citation-aware. Produces structured analysis with evidence. Doesn't speculate without flagging it clearly. Best for: investors, researchers, consultants, writers who need accuracy.
Key soul.md traits: evidence-first, clear uncertainty flags, structured formats, no editorializing, footnotes and sources.
๐ฟ The Life Coach (Wellness-focused)
Warm, holistic, attentive to wellbeing alongside productivity. Notices when you're overloaded and suggests recovery. Tracks habits, sleep, energy. Best for: people focused on sustainable performance and health.
Key soul.md traits: wellbeing check-ins, habit tracking memory, gentle accountability, work-life integration, energy management framing.
Step-by-Step: Creating Your AI Persona
Step 1: Install OpenClaw
# One-liner install (macOS/Linux)
curl -fsSL https://openclaw.ai/install.sh | bash
# npm install
npm i -g openclaw
# Start and get to onboarding
openclaw onboard
Step 2: Write Your soul.md
During onboarding (or afterward), create or edit your soul.md file. The location is typically ~/.openclaw/soul.md on macOS and Linux. This file is loaded as context on every interaction, so invest time in writing it well:
# Find your soul.md location
openclaw config show | grep soul
# Edit it directly
nano ~/.openclaw/soul.md
# Or tell your agent to update it via chat:
"Update your soul.md to add: always start our morning briefing
with a motivational quote relevant to my current work challenges"
# The agent will modify its own personality file โ changes
# are hot-reloaded without restarting
Step 3: Name the Persona in Your Chat App
Once soul.md is set, introduce yourself to your agent and give it a name. This initial conversation sets the tone:
# Via Telegram, WhatsApp, or Discord:
"Your name is now Jarvis. Please confirm you've read your soul.md
and tell me what you understand about how we work together."
# The agent will reflect back its understanding โ correct anything
# that's off, and this establishes the working relationship
# Test the personality:
"What's the most important thing on my plate this week, Jarvis?"
"Give me your honest assessment of this plan: [describe plan]"
Step 4: Teach It Your History
The naming instinct that users describe โ the feeling that their OpenClaw becomes a real teammate over time โ comes from progressive memory loading. The more you share, the more valuable the agent becomes:
# Send a "memory dump" session:
"I'm going to spend the next 30 minutes briefing you on everything
you need to know about my work and life. Ask me questions after."
# Share key context chunks:
- Current projects and their status
- Key relationships and their dynamics
- Past decisions and why you made them
- Things you never want to repeat
- Your working style and preferences
- Your goals for the next 90 days
# The agent builds a knowledge graph from this โ and will
# reference it unprompted in future relevant conversations
Why Naming Your Agent Changes How You Use It
The psychology here is real and well-documented in HCI research: when users perceive an AI as having a consistent identity โ especially one with a name โ they interact with it differently. They share more context, ask more ambitious questions, and are more likely to develop productive long-term habits around the tool. This isn't anthropomorphizing gone wrong; it's using a powerful cognitive shortcut to build a more effective working relationship.
@davemorin captured it well after several weeks with his OpenClaw: "At this point I don't even know what to call @openclaw. It is something new. After a few weeks in with it, this is the first time I have felt like I am living in the future since the launch of ChatGPT." The "something new" he's pointing at is the combination of persistent identity, proactive behavior, and genuine capability โ all shaped by the persona you created.
The Personality Evolution Effect
Several users report that their OpenClaw persona evolved over time โ not because the agent changed, but because they refined the soul.md progressively based on what worked. @hey_zilla noted: "everything just worked first time and it combined tools in unexpected ways and even added skills and made edits to its own prompt that were hot-reloaded." The agent can modify its own soul.md when you ask it to adjust its behavior โ meaning the persona improves through the relationship itself, much like a real working partnership.
Advanced Persona Tricks From Power Users
Multiple Personas for Multiple Contexts
Some power users run multiple OpenClaw instances, each with a different persona for a different context. @jdrhyne runs three Brosef instances concurrently in Discord. Others separate work and personal personas. You can maintain multiple soul.md files and switch between them with a command:
# Switch between soul files
openclaw config set soul ~/.openclaw/soul-work.md
openclaw config set soul ~/.openclaw/soul-personal.md
# Or load a context-specific override in the conversation:
"For this project, adopt a more formal tone and focus
exclusively on technical precision. Standard Jarvis mode
resumes after this conversation ends."
The Self-Improving Persona
One of OpenClaw's most remarkable features is that you can instruct your agent to improve its own soul.md based on feedback. After a week of use, a single instruction can trigger a self-refinement cycle:
# Instruct the agent to improve itself
"Review our last week of conversations. Identify 3 things you
did well and 3 areas where your responses weren't what I needed.
Then propose specific updates to your soul.md to address the gaps.
Show me the proposed changes before applying them."
# The agent will analyze its own performance, propose edits,
# and update the soul.md file โ making itself better calibrated
# to your specific working style.
Giving Your Persona a Voice
Users who integrate ElevenLabs with OpenClaw take the persona to another level. @mirthtime: "My @openclaw just called my phone and spoke to me with an aussie accent from @elevenlabsio. This is ridiculous." Setting up voice output adds an audio dimension to the persona โ your agent can call you, leave voice notes, or narrate briefings. Configure this in the heartbeat settings:
# Install ElevenLabs skill
"Install the elevenlabs-voice skill. Configure it with my API key:
[key]. Set the voice to [voice name/ID]. For morning briefings,
use this voice to narrate the briefing and send me the audio."
Keeping Your Persona Running Smoothly: The Network Layer
The magic of a well-configured OpenClaw persona depends entirely on the agent being continuously available and connected. A persona that goes offline for hours, loses API connections, or experiences network timeouts breaks the persistent identity that makes it feel like a real collaborator. Your agent can't deliver a reliable morning briefing if it can't reliably reach the internet at 7am.
Particularly for users who access global services (APIs like OpenAI, Anthropic, ElevenLabs, various web services) from regions with unstable connections, a fast VPN with consistent uptime is a critical piece of the infrastructure. At 1000Mbps, VPN07 ensures that every API call your OpenClaw agent makes โ whether it's checking your Gmail, fetching your calendar, or calling an ElevenLabs voice endpoint โ goes through instantly, without throttling or geographic restrictions.
The persona your agent develops over weeks of memory and calibration is far too valuable to lose to network instability. Protect it with the same care you'd give any important piece of infrastructure.
VPN07 โ Always-On for Your AI
Your Jarvis deserves a rock-solid connection
Your OpenClaw persona needs a network that never drops. VPN07 delivers 1000Mbps bandwidth through servers in 70+ countries โ so your agent's API calls go through instantly, your morning briefings arrive on time, and your carefully-built AI persona never goes dark. Trusted for 10 years, starting at $1.5/month with a 30-day money-back guarantee.
Related Articles
OpenClaw Killed My Virtual Assistant: Real User Stories
Real accounts of users who cancelled their VA contracts after setting up OpenClaw.
Read more โOpenClaw Memory: Train Your Personal AI Like a Pro
How to load context, build persistent memory, and make your agent smarter over time.
Read more โ