Coca-Cola Optimises Orders, M&T Bank Grows AI, Google Enters Grid Weather

In partnership with

Free email without sacrificing your privacy

Gmail is free, but you pay with your data. Proton Mail is different.

We don’t scan your messages. We don’t sell your behavior. We don’t follow you across the internet.

Proton Mail gives you full-featured, private email without surveillance or creepy profiling. It’s email that respects your time, your attention, and your boundaries.

Email doesn’t have to cost your privacy.

Ditch the Gmail data grab

AI HUSTLE | September 10, 2026

Silos are the ultimate productivity killer in any growing business. When your product developers and your marketing team don't talk, you end up paying two different departments to build the exact same feature guide. Today, we are breaking down how to build an automated AI "Knowledge Linker" that scans your company's task managers and flags duplicate work before your teams waste hundreds of hours. Plus, we look at Coca-Cola’s massive AI-driven B2B ordering rollout, a regional bank scaling Copilot to 15,000 employees, and Google’s new weather AI targeting the energy grid. Let's dive in.

The Hustle: The Silo-Buster: How to Stop Duplicate Work Across Teams Automatically

The Goal: Stop departments from working in isolation and accidentally duplicating efforts by using AI to scan, cross-reference, and link redundant tasks in real-time.

The Tools:

* Make.com or Zapier (Workflow Automation)

* OpenAI API (GPT-4o)

* Your Team’s Task Managers (Jira, Asana, Notion, etc.)

Step 1: Consolidate the Inflow (The Input)

Set up trigger modules in Make.com or Zapier that listen for any new task created across your company's department tools. For example, set up triggers for "New Issue in Jira" (Dev team), "New Task in Asana" (Marketing), and "New Page in Database" (Notion/Ops). Configure the trigger to pull the Task Title, Task Description, Department, and Task URL.

Step 2: Centralize the Task Registry (The Trigger)

To compare new tasks against existing ones, you need a single source of truth. Direct your automation tool to write every newly created task into a centralized, running database—a lightweight Airtable base works perfectly for this. Store the task's title, description, department, date created, and original platform link.

Step 3: Run the Semantic Similarity Engine (The AI/Logic)

Whenever a new task is added to your central Airtable base, trigger a script that sends the new task's title and description, along with a list of all active tasks from the past 30 days, to the OpenAI API. Use a system prompt like this:

"You are an expert project manager. Compare the following New Task with the list of Active Tasks. Identify if there is any substantial overlap in objective, deliverables, or research (e.g., 'Create a guide for Feature X' vs. 'Document Feature X'). If the likelihood of duplicate effort is above 80%, return the ID of the matching Active Task and write a 2-sentence explanation of why they overlap. If no duplication exists, return 'No Match'."

Step 4: Post the Cross-Department Alert (The Output)

If the AI detects a match, parse the returned matching Task ID and original platform link. Use your automation tool to post an automated comment directly in both task threads (e.g., a comment on the Jira ticket and a comment on the Asana task). The comment should read:

"Automated Alert: Potential duplicate effort detected with [Department Name]'s task: '[Task Title]' ([Platform Link]). Consider collaborating to save time!"

Why This Hustle Works:

* Saves Hundreds of Hidden Hours: It flags redundant research, documentation, and asset creation before your team members spend days working on them in parallel.

* Bridges Software Divides: It allows developers to stay in Jira, marketers to stay in Asana, and ops to stay in Notion, while still maintaining 100% cross-organizational visibility.

🚀 The AI Pulse: 3 Signals to Watch This Week

Coca-Cola Scales "Perfect Basket" AI Recommendations Across Malaysia

Coca-Cola is leveraging its "Coke Buddy" B2B platform to recommend exact product types and quantities to 39,000 retail outlets in Malaysia. The "Perfect Basket" feature uses a central recommendation engine that analyzes past purchase history, seasonality, local weather patterns, and buying habits of similar businesses. During a recent campaign, 83% of participating retailers adopted the AI's recommendations, resulting in significantly higher sales revenue growth. Coca-Cola has already expanded similar AI-enabled B2B ordering capabilities to over 3 million outlets in Latin America.

The Hustle Take: Coca-Cola is showing businesses how to shift sales representatives away from basic order-taking and toward high-value account development. By blending internal customer data with external APIs (like weather and local trends), any B2B business can build a basic recommendation script to prompt their sales team with "smart suggestions" for clients, unlocking instant upsell opportunities.

M&T Bank Scales Copilot to 15,000+ Employees After Data Cleanup

US regional lender M&T Bank has successfully rolled out Microsoft Copilot and GitLab AI coding assistants to over 15,000 employees. The bank uses the tools to draft emails, summarize call-center conversations (saving 6 minutes per call), and generate code. However, the rollout didn't happen overnight; M&T spent years restructuring its data. They built a robust "data lineage" program using tools like Solidatus and Monte Carlo to track where information originates and how it moves, ensuring the AI only references governed, accurate company data.

The Hustle Take: You cannot build a great AI strategy on top of messy data. M&T Bank succeeded because they banned public LLMs and focused on data governance first. If you want to deploy AI assistants across your company, stop looking for shinier AI tools and start building a clean, structured internal knowledge base (like Notion or an enterprise wiki) so your AI actually yields accurate results.

Google Targets Energy Sector with Real-Time "WeatherNext 3" AI

Google DeepMind and Google Research have launched WeatherNext 3, a highly advanced AI weather forecasting model designed specifically for grid operators, energy traders, and wind/solar developers. The model updates every hour and predicts micro-variables like wind speed at turbine height (100 meters) and solar cloud cover. Because modern power grids are flooded with weather-dependent renewable energy—and facing massive load increases from AI data centers—accurate hourly forecasts are worth millions. Google is bypassing specialist weather vendors by delivering this data directly into enterprise cloud tools like BigQuery.

The Hustle Take: Weather is the ultimate macroeconomic variable. By packaging complex meteorological data into simple, queryable BigQuery tables, Google is making enterprise-grade predictive analytics accessible to non-scientists. If your business is heavily impacted by external variables (like logistics, delivery, construction, or outdoor events), you can now feed WeatherNext 3’s hourly data directly into your scheduling software to optimize operations automatically.