Good morning.

If you run a small business, trade, or technical team and use AI for quotes, emails, reports or coding help, the last few days delivered clearer costs and clearer rules. The big platforms are making everyday models cheaper and faster while governments tighten the fine print on energy use.

In today’s briefing:

⚡Google’s new Gemini 3.7 Flash cuts prices and improves coding agents

💰 OpenAI slashes the cost of its fast everyday model by up to 80%

🔋Australia’s AI minister doubles down on data-centre rules that protect household power bills

💡 Practical prompting tip for cleaner work documents

Assistant Minister Andrew Charlton is set to outline the government’s case for AI investment while defending the “triple-lock” rules for large data centres: they must bring their own renewable power, pay full network connection costs, and offer demand flexibility so household bills are not pushed up.

Why it matters for you: Data-centre growth is already lifting electricity demand in NSW and Victoria. These rules are the main federal lever trying to stop that cost landing on your power bill or local small-business rates.

South Australia has announced a royal commission into artificial intelligence, expected to start in October and report mid-2027. It will look at education, health and the state-level impacts of data-centre build-out.

Why it matters for you: State inquiries often shape local planning rules and skills programs that affect tradies and regional businesses first.

The national Office of AI (set up inside the Department of the Prime Minister and Cabinet) continues work on standards for data centres and copyright so Australian creatives keep control of their work. Legislation is aimed for early 2027.

The Daily Rundown: Google released Gemini 3.7 Flash, its latest everyday model aimed at coding and multi-step agent tasks. It improves on the previous version in software engineering and web development benchmarks and arrives with introductory API pricing of $0.75 per million input tokens and $3.75 per million output tokens (half the earlier Flash rate, available until the end of 2026).

Practical takeaway: If you use AI for code helpers, report generation or simple automation, this is worth testing against what you currently pay. Lower token costs mean longer agent runs or higher volume without the bill spiking as fast.

Source: Google blog

The Daily Rundown: OpenAI cut the API price of GPT-5.6 Luna by about 80%. It now costs $0.20 per million input tokens and $1.20 per million output tokens. The mid-tier Terra model also dropped 20%. These cuts come from efficiency improvements and apply to both API users and usage counting in ChatGPT Work and Codex.

Practical takeaway: High-volume tasks such as sorting emails, pulling data from invoices, first-draft writing or simple automation just got noticeably cheaper. If you or your tools run on OpenAI models, check your current usage, this can stretch the same budget further without changing how you work.

Source: OpenAI / Axios reporting

Gemini 3.7 Flash (via Google AI Studio or the Gemini app) A practical, lower-cost model tuned for coding tasks, document work and multi-step agents. Free tier available in AI Studio for testing; paid access through Google AI Pro/Ultra or the API. Good option if you already live in Google Workspace and want something that handles longer workflows without the premium price tag of the biggest models.

Try it: ai.google.dev or the Gemini app.

When you ask an AI to draft a quote, scope of works or client email, tell it the exact format and tone you need up front and force a final human-check step.

Ready-to-use example (for a tradie or small contractor):

“Write a plain-English quote email for a residential bathroom renovation in Sydney. Include: brief summary of the job, three line-item options (basic / mid / premium), total price range excluding GST, estimated start date, and a clear next-step question.”

Keep it under 180 words, friendly but professional, no sales hype.

End with: ‘Please reply if you’d like me to adjust anything before we lock in dates.’”

Both price cuts today land on the same tasks — sorting emails, pulling data from invoices, first-draft writing — which is exactly the territory Basic AI Workflows That Save You Hours covers: five repeatable workflows for the admin that eats your week, now cheaper to run at volume than ever. And the quote-email prompting tip above — exact format, tone, and a forced human-check step — is the setup discipline Why Your AI Output Looks Like Garbage (And the 3 Things That Fix It) teaches in one afternoon: vague tasks produce vague drafts, specific briefs don't. Both are in the AI Foundations track at Vortex Academy.