Responsible AI Handbook: Part 2 – Green Standard AI
In the digital age, AI has become a familiar tool for businesses in planning, customer care, market research, and content creation.However, behind every AI command is a data center that consumes electricity, water, and emits CO₂. Without mindful usage, the environmental cost can quickly exceed expectations.
KisStartup – with experience supporting thousands of businesses on their innovation and digital transformation journeys – has compiled this guide to help companies use AI responsibly, efficiently, and in an environmentally friendly way.We call it Green Standard Prompting: boosting productivity while reducing emissions.
Why do we need "Green Standard Prompting"?
Every AI command consumes energy and water:
- Gemini (Google): approx. 0.24 Wh, emits 0.03 gCO₂, and uses 0.26 ml of water per average text prompt.
- ChatGPT (GPT-4o): approx. 0.3 Wh per prompt.
1 million prompts can consume around 300 KWh, equal to a household’s electricity use in one month.
So, every time you revise a prompt repeatedly, you're multiplying the power and water usage. That’s why carefully crafting your prompt is not only time- and cost-effective, but also a clear ESG (Environmental, Social, Governance) action.
Principles of Green Standard Prompting:
- Be clear about your goal: what you want, for whom, and in what format.
- Provide enough context: product, data, constraints.
- Limit output length: specify word count or number of bullets.
- Choose the right model: simple tasks → lightweight models.
- Ask AI to request more info if needed, instead of guessing.
- Save and reuse good prompts to avoid repetition.
Example Green Standard Prompts
1. Planning Content Marketing
System (Role): You are a Sustainable Content Marketing expert.
User Prompt:
- Goal: Plan 2 weeks of content for a {industry} fanpage.
- Audience: {target customers}
- Context: product {...}, USP {...}, budget {...}
- Output (≤200 words):
- 14-day content calendar
- Captions ≤30 words
- Hashtags ≤5 per post - Constraints: prioritize repurposing existing content, ask up to 3 follow-up questions if data is missing.
2. Developing a Green Export Plan
System (Role): You are a Green Export & ESG expert.
User Prompt:
- Goal: Create a 6-month export plan for {product} to {market}.
- Context: certifications, production capacity, current partners
- Output (≤250 words):
1. 5 green requirements/VSS (Voluntary Sustainability Standards) for the marke.
2. 3 current capability gaps
3. 3 priority actions for the first 90 days
4. 2 long-term opportunities - Constraints: include a checklist for executives, ask up to 5 follow-up questions if data is missing, cite sources.
Steps to Build a “Green Standard” AI Assistant for Your Business
- Define the assistant's role (e.g., Content Coach, Export Advisor).
- Standardize the system prompt (role + green principles).
- Create a sample prompt library (like examples above).
- Train with real data (products, certifications, customer info).
- Test & refine to minimize prompt iterations.
- Integrate into workflows (chatbot, CRM, Notion/Slack).
- Monitor & report on resource savings (tokens, kWh, CO₂, water).
Checklist Green AI Prompting
Before you type a command:
- Is the goal, target audience, and output format clear?
- Is the data sufficient so AI doesn’t have to guess?
- Have you set an output length limit?
When choosing a model:
- Do you really need a large model?
- Are you asking for images/slides unnecessarily when text suffices?
During execution:
- Does the AI ask follow-up questions when data is missing?
- Can this prompt be reused?
After completion:
- Is the output immediately usable, or does it require re-running?
- Can the prompt be shared with teammates?
