Ai Jarvis mark 1 -- 2
Budget: $750 – $1,500 AUD
Project Overview
Hello, and thank you for taking the time to read this.
I’m working on Jarvis (Mark 1) — a personal AI assistant designed for practical, everyday use, not a demo chatbot or a flashy concept.
The goal of Mark 1 is reliability, clarity, and good system design.
Advanced visuals, autonomy, and cinematic features will come later.
⸻
Preferred Tech Stack
For this phase, I’m planning to use:
• Voiceflow – conversation logic and orchestration
• Airtable – memory, preferences, and rules
• ChatGPT API – reasoning, drafting, summaries
• ElevenLabs – British butler–style voice
If you’d like to suggest an alternative approach, that’s completely fine — please just explain why you think it would be better for Mark 1.
⸻
Core Design Principles (Very Important)
Please read these carefully, as they define the project’s direction:
• Jarvis should always ask for confirmation before taking actions
• Jarvis should never auto-send messages
• Preferences and memory should be saved only with explicit approval
• Privacy, restraint, and predictability are more important than automation
This project focuses on control and trust rather than doing everything automatically.
⸻
Mark 1 Feature Scope
Voice & System Control
• Open and close applications
• Open websites
• Basic browser tab control
• Window switching
• Timers and simple system commands
⸻
Daily Routines
• Qur’an study routine
• Flipping / selling routine
• Night routine
• Timers, notes, and simple checklists
⸻
Qur’an Support (Mark 1 level)
• Open Quran.com at the last saved ayah
• Open translation and tafsir (e.g. Ibn Kathir, if available online)
• Save resume point
• Create short reflection notes
This is meant to support consistency, not provide advanced religious interpretation.
⸻
Flipping / Selling Assistant
• Generate listing titles and descriptions
• Suggest pricing (based on manual inputs)
• Provide negotiation message templates
• Alert me when a buyer replies
When a buyer replies, Jarvis should:
1. Send a notification (phone/watch)
2. Show what the buyer said
3. Suggest reply options
4. Ask for confirmation before sending anything
⸻
Identity, Tone & Voice
• Jarvis addresses me as: “Sir”
• When speaking about me to others: “Mr. Khodr”
• British butler–style voice (calm, professional, not exaggerated)
• Neutral, discreet responses when I’m not present
⸻
Devices
• Desktop / Monitor: main control center
• Phone: alerts, confirmations, reminders
• Watch: silent notifications and quick decisions
⸻
Out of Scope for Mark 1
To keep this phase focused, the following are intentionally excluded:
• Full custom OS or UI shell
• Autonomous decision-making
• Auto-sending messages
• AR, holograms, or cinematic interfaces
• Advanced “Mark 2” features
⸻
A Small Request When Applying
To help me better understand how you think, please include a short answer to one of the questions below (a few sentences is enough):
1. How would you design the confirmation flow to ensure Jarvis never acts without permission?
or
2. What do you see as the biggest technical risk in this build, and how would you reduce it in Mark 1?
Thank you — this really helps me choose the right collaborator.
⸻
What I’m Looking For
• Someone who enjoys building clean, well-scoped systems
• Comfortable keeping Mark 1 simple and stable
• Clear communication and thoughtful decision-making
• Interest in a longer-term project if Mark 1 goes well
Hello, and thank you for taking the time to read this.
I’m working on Jarvis (Mark 1) — a personal AI assistant designed for practical, everyday use, not a demo chatbot or a flashy concept.
The goal of Mark 1 is reliability, clarity, and good system design.
Advanced visuals, autonomy, and cinematic features will come later.
⸻
Preferred Tech Stack
For this phase, I’m planning to use:
• Voiceflow – conversation logic and orchestration
• Airtable – memory, preferences, and rules
• ChatGPT API – reasoning, drafting, summaries
• ElevenLabs – British butler–style voice
If you’d like to suggest an alternative approach, that’s completely fine — please just explain why you think it would be better for Mark 1.
⸻
Core Design Principles (Very Important)
Please read these carefully, as they define the project’s direction:
• Jarvis should always ask for confirmation before taking actions
• Jarvis should never auto-send messages
• Preferences and memory should be saved only with explicit approval
• Privacy, restraint, and predictability are more important than automation
This project focuses on control and trust rather than doing everything automatically.
⸻
Mark 1 Feature Scope
Voice & System Control
• Open and close applications
• Open websites
• Basic browser tab control
• Window switching
• Timers and simple system commands
⸻
Daily Routines
• Qur’an study routine
• Flipping / selling routine
• Night routine
• Timers, notes, and simple checklists
⸻
Qur’an Support (Mark 1 level)
• Open Quran.com at the last saved ayah
• Open translation and tafsir (e.g. Ibn Kathir, if available online)
• Save resume point
• Create short reflection notes
This is meant to support consistency, not provide advanced religious interpretation.
⸻
Flipping / Selling Assistant
• Generate listing titles and descriptions
• Suggest pricing (based on manual inputs)
• Provide negotiation message templates
• Alert me when a buyer replies
When a buyer replies, Jarvis should:
1. Send a notification (phone/watch)
2. Show what the buyer said
3. Suggest reply options
4. Ask for confirmation before sending anything
⸻
Identity, Tone & Voice
• Jarvis addresses me as: “Sir”
• When speaking about me to others: “Mr. Khodr”
• British butler–style voice (calm, professional, not exaggerated)
• Neutral, discreet responses when I’m not present
⸻
Devices
• Desktop / Monitor: main control center
• Phone: alerts, confirmations, reminders
• Watch: silent notifications and quick decisions
⸻
Out of Scope for Mark 1
To keep this phase focused, the following are intentionally excluded:
• Full custom OS or UI shell
• Autonomous decision-making
• Auto-sending messages
• AR, holograms, or cinematic interfaces
• Advanced “Mark 2” features
⸻
A Small Request When Applying
To help me better understand how you think, please include a short answer to one of the questions below (a few sentences is enough):
1. How would you design the confirmation flow to ensure Jarvis never acts without permission?
or
2. What do you see as the biggest technical risk in this build, and how would you reduce it in Mark 1?
Thank you — this really helps me choose the right collaborator.
⸻
What I’m Looking For
• Someone who enjoys building clean, well-scoped systems
• Comfortable keeping Mark 1 simple and stable
• Clear communication and thoughtful decision-making
• Interest in a longer-term project if Mark 1 goes well