I used AI to analyse 13 months of energy data, forecasts for the upcoming year, and make a decision in 15 minutes. The same analysis would have taken 3-5 hours manually.
Here's the exact process with copy-paste prompts.
What I did (and you can too)
When my Octopus Energy tracker tariff needed renewal, I had AI analyse 400+ days of pricing against my actual usage. It delivered:
- Data processing: 13 months of costs calculated, showing I'd saved £134 on tracker vs what fixed would have cost.
- Market research: Found April 2026 price cap predicted to drop 3-6%, Feb gas futures up 21%, tracker formula changed Dec 31st, and a 9-month lockout rule I never knew existed.
- Four scenarios: Both tracker (£2,037), gas fixed + elec tracker (£2,087), elec fixed + gas tracker (£2,120), both fixed (£2,171)
- Usage insights: 79% of my gas consumed in 5 winter months-critical for strategy
- Action plan: Stay tracker, switch if gas >7.5p for 14+ days, weekly checks Feb-Apr, monthly after
My time: 15 minutes. AI's work: 3-5 hours of professional analysis.
What you need
Gather three things:
- Monthly usage data (electricity and gas in kWh)
- Current vs. proposed tariff rates (unit rates and standing charges)
- Historical pricing (for variable tariffs)
For Octopus Energy Users
- Monthly usage: Copy from dashboard (Usage → Monthly breakdown).
- Tariff comparison: Request from support or check renewal emails.
- Tracker history: Download CSV from tracker.octopushome.net/historical-data
For Other Suppliers
- Check online account for usage breakdowns.
- Download statements or export usage data.
- Contact customer service for tariff details.
The Prompt
Copy this and fill in your data:
Re: [Name of your supplier] energy tariffs:
- the CSV has gas and electricity [Tracker/Variable] daily prices since [DATE] till [DATE]
- the [Tracker/Variable] standing charge for gas was XX.YY and for electricity was XX.YY
- last year's actual usage was as follows:
ELECTRICITY:
Month Total
January XXX.XXX kWh
February XXX.XXX kWh
[... continue for all months]
GAS:
Month Total
January X,XXX.XXX kWh
February X,XXX.XXX kWh
[... continue for all months]
- The new [Tracker/Variable] standing charges from [DATE] are:
GAS: XX.YY
ELECTRICITY: XX.YY
- [Your Supplier] states:
Here's a full breakdown of all our tariffs:
Before After
Tariff name [Current] [New Option]
Unit rate XX.YY p/kWh XX.YY p/kWh
Standing charge XX.YY p/day XX.YY p/day
Unit rate - Gas XX.YY p/kWh XX.YY p/kWh
Standing charge XX.YY p/day XX.YY p/day
- Today is [DATE].
- Research online if you have to.
- Advise on whether we should continue with new [Variable/Tracker] or [Fixed] Plan from [DATE].
Optional: Hybrid Strategy Analysis
How about keeping one (gas or electricity) fixed and one on tracker - would that make any difference?
For me, this revealed hybrid strategies cost £51-84 more. Your results will differ.
What you'll get
Professional analysis
- Precise cost calculations across all scenarios.
- Market forecasts from multiple sources synthesized.
- Hidden factors surfaced (formula changes, lockout periods, standing charge increases).
- Probability-weighted recommendations with specific numeric triggers.
- Custom monitoring framework based on your usage patterns.
My key results
- Recommendation: Stay tracker (£120-180 expected savings vs fixed over next year).
- Risk management: Specific triggers to switch if rates spike.
- Probability: 70% chance tracker performs better.
- Peace of mind cost: £15/month premium for fixed if I wanted certainty.
Time breakdown
Manual approach (3-5 hours)
- Data processing: 1 hours (parse 400+ prices, match to usage, calculate averages)
- Market research: 2.5 hours (find forecasts, regulations, hidden changes)
- Analysis: 1 hour (scenarios, breakeven, probability, frameworks)
- Checks: 30 minutes (verify calculations, catch errors)
AI Approach (15 minutes active)
- Download data: 2 minutes
- Copy usage: 3 minutes
- Write prompt: 3 minutes
- Read analysis: 5 minutes
- Follow-up question: 2 minutes
AI processes everything concurrently in 3-5 minutes while you do something else.
Beyond energy bills
Same technique works for:
- Personal: Mortgage refinancing, insurance comparisons, mobile contracts.
- Business: Software subscriptions, shipping carriers, payment processors, cloud hosting.
- Investment: Property rental yields, vehicle leasing vs purchasing, equipment financing.
Key tips
- Be specific: Exact numbers, not estimates. Include units and date ranges.
- Upload files: CSV and PDF uploads work better than typing data.
- Ask follow-ups: "What if X changes?" "What's the breakeven point?" "What am I missing?".
- Verify critical facts: Confirm AI understood your usage correctly, double-check quoted rates.
- Request action plans: Get specific triggers, monitoring schedules, contingencies.
Common Mistakes
- Insufficient data: Don't ask AI to estimate when you can provide actual numbers.
- Vague questions: Ask "Based on my usage and how much risk you want to tolerate, which maximizes expected value?" not "Which is better?".
- Missing context: Include your priorities, constraints, risk tolerance, time horizon.
- Treating AI as oracle: AI provides analysis, you make the decision.
Join the community
Learn practical AI techniques for business and life: join my Discord server.
Discuss AI, automation, and apps that save time and money. No fluff, just practical applications.
The bottom line
AI delivered:
- Higher-quality analysis than I could produce manually.
- Confidence to make a significant financial decision quickly.
- Monitoring framework I can use throughout the year.
- Found critical details I would have missed (9-month lockout, formula changes, standing charge increases)
Time saved: 3-5 hours. Value: Professional-grade analysis catching £134 in savings plus identifying risks and opportunities.
The future isn't AI replacing human judgment-it's AI amplifying it. Make better decisions faster, so you can focus on what matters.
Try this technique for your next complex decision. Share your results in the Discord community.