Analyzing Quality with a Click

In my last blog I discussed the newly patented Microsoft Food Logging technology which will soon allow me to photograph the Thai curry on my dinner plate and help me log my caloric intake. Today I would like to introduce a similar concept but this time to capture the food at the beginning of the supply chain well before it reaches my plate. Instead of calculating calories, this new technology uses the images to calculate quality metrics.

Before the steamed rice hit my dinner plate, the rice had been grown, harvested, processed, and packaged. During the processing at the rice mill the rice was continuously checked for various quality parameters including grain size and damage. Consumers today expect large, beautiful whole grains of rice, not broken discolored pieces.

Currently, quality control is cumbersome, as it requires samples to be sent to a lab and manually analyzed using calipers to measure length, scales to measure weight as well as visual inspection to discern color and cracks.  These tests are often performed by several quality engineers in succession and can take hours to complete. This often results in delays as the grains cannot be further processed until the results are verified.

Building on the power of artificial intelligence, this quality process can be drastically simplified. Bühler, a Swiss grain processor announced at Microsoft Ignite last week that they are employing Artificial Intelligence to analyze grains in a fraction of the time it currently takes a quality engineer. By using a smartphone app, the engineer can take a photo of the rice and upload it to the Microsoft cloud for instant analysis.

As demonstrated in the above image, the painstaking manual analysis is replaced with a simple click. The photo is then analyzed using Microsoft intelligent cloud technologies and presented to the user in an easy-to-read graphic.

With 480 million metric tons of rice being consumed annually worldwide, a low cost and objective solution built on the power of AI increases efficiencies and maximizes yield.

For more information on the Bühler and Microsoft solutions, click here.

A Picture Worth a Thousand Calories

I will confess now: I am one of the 300 million people who have downloaded weight loss apps. I am currently using and having success with Weight Watchers. My success in part is attributed to keeping track of what I eat, and being aware of the calories and fat that are in the every day foods I am consuming. In other words, monitoring my intake has made me aware and thus able to make better choices. As I approach my goal weight, I must admit that long term, I do not see myself searching for every food, determining the portion of each item on my plate and keying it in – truthfully, it is very tedious. I have grown weary.

Fortunately for me, Microsoft has just received a new patent for “Food Logging”. According to the patent, “the Food Logger provides various approaches for learning or training one or more image-based models of nutritional content of meals. …Given the trained models, the Food Logger automatically provides estimates of nutritional information based on automated recognition...This nutritional information is then used to enable a wide range of user-centric interactions relating to food consumed by individual users“.

Microsoft intends to develop this for use at a restaurant as well as with home-cooked meals. With this new technology – which combines computer vision and machine learning, I will soon be able to maintain my ideal weight by simply taking a photo of my entire plate and letting the app do the logging for me.

In addition to being used for weight loss, identifying allergens and sugars are other obvious advantages of this technology. Microsoft plans integration into medical devices and exercise devices as well.

---

 

Dear Future, I am Ready…

It is an exciting time in technology. The explosion of Artificial Intelligence has everyone using the power of AI in everyday tasks –

• asking our phones for directions, or to set a reminder
• filtering our Inbox to help us focus on the important emails
• websites suggesting products based on past buying and predictive analytics
• receiving fraud detection texts for an outlier credit card transaction

This reliance on technology and AI will continue to grow and soon we will see it transform the way we buy household items and food. Soon everyone will ask our phones to not only call Mom but to order grocery goods such as paper towels and milk, and a driver-less car or shared-ride service car will deliver it.

In store experiences will probably focus on improving the customer experience. While in our favorite store, we might scan a bag of blueberries and see which farm in Washington state they were grown and the date picked.  We may also be able to verify that the farm is certified organic.  Next, an online coupon suggests organic raspberries and directs us to their location in the store. Perhaps we opt-in to recipe ideas and our app directs us to the flour and sugar aisle to grab the remainder of our pie ingredients. And instead of standing in line to buy our groceries, we simply walk out and our account is automatically charged. As we wait for our shared ride, our phone pings and we glance at our app to confirm our purchase receipt.

In under a few minutes we have researched the product we intend to purchase, received directions to our next purchase, used a coupon and made our purchase. The customer experience was simple, fast and hassle-free. No lines, no mess no fuss.  Soon, this will be just a normal trip to the grocery store.

For information on how Microsoft is investing in AI’s future in retail, click here.

 

 

Inheritance

This past holiday season DNA test kits broke sales records from black Friday through New Years. Millions of us wanted to know from where we inherited our defining characteristics such as our big noses, curly hair or green eyes.

Just like us, manufactured products have characteristics that are inherited.

Food products specifically can inherit important characteristics from their ingredients. For example, the shelf life of a prepared airline meal is based on the earliest expiration date of any individual ingredient. When creating the formula or recipe for the meal, a best practice is to set a parameter on the least sustainable ingredient to pass its shelf life to the end product, in our example the meal. So if our airline meal contains dairy or seafood those ingredients will drive the overall shelf life of the entire meal.

Taking this one step further, with Dynamics 365 for Finance and Operations you can not only set the meal to inherit the ingredient’s shelf life but you can configure acceptable ranges of shelf life days for each customer. So before you ship that meal to Airline ABC, Dynamics will make certain the shrimp is fresh. This upstream inheritance logic reduces returns and streamlines operations by guaranteeing the product being shipped meets your customer’s shelf life needs.

Attribute that to the attribute

Say what? As I was researching for this blog post I realized that the word “attribute” is spelled the same but pronounced differently depending on whether you are using it a noun or a verb.

---

Noun – an abstraction belonging to or characteristic of an entity

Ex: Attributes of cheese are fat content, age and moisture.

Verb –  give credit to

Ex: I attribute my success as a blogger to my manager, Danny.

---

Well, enough word fun for the day, let’s expand on last blog’s topic and talk about the noun attribute and how many raw materials have attributes that affect how they are used in a formula. In addition to tracking these attributes we need to go one step further and consider them in our formulas, specifically – how they may affect the formulation.

A common flavor enhancer and shelf-life enhancer is potassium chloride, which can have varying potency that must be neutralized with a solvent such as water when used in a formula.  The act of neutralizing the potassium chloride is called balancing. Depending on the actual potency of the potassium chloride, we want our formula management system to balance our formula and calculate the correct amount of water needed to achieve the desired balance of potassium chloride and water.

The importance of this balancing extends beyond the formula to affecting costs, operational efficiencies and ultimately revenues. Best practices are to automate this process to ensure data integrity.

To watch a video on how to automate batch balancing in Dynamics, click here.

 

 

Achieving the perfect balance

You may thought this was another article on balancing your work and life or perhaps tips on dieting, but it is neither. Instead of life hacks we are going to talk about how to perfectly manage product attributes when producing and selling food products.

Attributes are any characteristic of a raw material or finished product that affects how it is used or sold. A great example of this are dairy products. Attributes of cheese are pH level, moisture, hardness, acidity and shelf life, to name a few.

Processing by Attribute

A cheese processor will strive to select the best wheel of cheese for shredding based on characteristics of the wheel such as pH level and moisture. These attributes effect how easily the cheese will shred. It is important to select the correct wheel for shredding to decrease scrap and reduce machine downtime. How that wheel is selected is also important – it should be automated, i.e. the production order pick list should only display the correct wheels, not relying on the operator to decide.

Upon receipt of the cheese wheel, quality control will perform a test to determine the values of each attribute and track those values to each wheel or batch of wheels.  By controlling which cheese is available for shredding, errors and setup times are reduced.

Fulfilling by Attribute

Just as production control used attributes to select the best wheel to shred, customer service will use the shelf life attribute to select the best finished goods to ship to each customer. For example, a high end grocery retailer might have a requirement of 21 days shelf life for milk while a discount store has only 7 days shelf life requirement.  Think of the time saved by only showing the cartons of milk to ship that meet our customer’s shelf life criteria.

For more information about how you can perfectly manage your attributes, click here.

Grow Your Own?

The unemployment rate in the US is the lowest in 17 years but manufacturing jobs in general are still on the decline.  Food manufacturing, however, is a bright spot as it has contributed to our country’s employment with about 150,000 new jobs being added in the last 10 years.

According to the Bureau of Labor from 2008 to 2018:

Unfortunately, projections for the next ten years do not look as bright with an anticipated decline in specialty food manufacturing jobs of 4.6%.

According to the Bureau of Labor, clerical, bookkeeping and purchasing agent position will lose the most ground over the next ten years. These positions will see double digit losses in percent change by the year 2026.

The good news is there are jobs gaining positions within the food sector and many of these jobs do not require four year or higher level degrees. For example, maintenance, machinery workers and machine operators will all experience solid growth by 2026.

Instead of spending the time and money to recruit, hire and relocate new talent, wouldn’t it make more sense to grow the trusted talent within your organization? Knowing that the workforce is changing, now is the time to put a plan in place to retrain and reskill your workforce.

Suggested steps to retrain:

• Identify what skills your workplace needs
• Align your workforce with these skills
• Create goals: courses, certifications, performance
• Perform reviews and assign measurements
• Continually monitor skill gaps and perform skill profile analysis

These tasks fall into the human resource or human capital management process within your organization. A solid framework for managing these processes can be obtained through Microsoft Dynamics 365 for Talent.

For more information how to bring out the best in your people, click here.

Using Blockchain to Save Green(s)

In total 172 people were reported to be infected with E. coli from romaine lettuce in 32 states in March and April of this year. The CDC recommended that everyone throw out their romaine lettuce unless they were certain it did not come from farms in Yuma. It is the last sentence that caught my attention –  as consumers we have no idea where our lettuce (or any other veggie) is grown. The bag will tell us who distributes it but not where it was grown or who is responsible for growing it.

So, everyone – households, restaurants, retail – is being advised to throw away food that is most likely perfectly good, but who wants to take the chance of getting sick or infecting their family? Certainly not me. But I hate the idea of throwing out perfectly good food.

According to the USDA, in the US we throw out more than $30 billion a year in vegetables alone. In total, we throw out over 30% of our available food supply, which is a whopping 133 billion pounds of food a year. While there are many reasons food is thrown out, for more info read here, it certainly caught my attention that poor supply chain optics are one of them. In fact, a report in 2017 found that 65% of companies do not have full visibility into their supply chains.

I wondered why the CDC would recommend such a drastic action and not just perform a recall to determine the source of the E. coli. From what I can surmise the distributor is unable to perform a recall.  They are among the 65% of companies who do not have visibility into their own supply chain – in other words, they don’t know which farm or field the infected lettuce was grown.

How can distributors see into their supply chain? The answer lies in block chain technologies, not the crypto currency blockchain but the data transaction blockchain. Using a centralized ledger for the entire lettuce supply chain to record temperatures, quality tests results, movement and lot numbers. Everyone who handles the lettuce records the data: the farmer, the carriers, processors, wholesaler, and retailers.

With the entire supply chain recording the data, it does not fall on one entity such as the distributor to keep track of every movement and every data point.

Recently Walmart piloted blockchain for recalling mangoes. Normally a recall would take 6 days and 18 hours to trace a bag of mangoes from the store back to the farm. With blockchain, they performed the exact same recall in 2.2 seconds. You read the right – from a week to 2 seconds. Wow, that is a disruptive technology that will save some serious green(s).

To learn more, visit azure.com/blockchain

“Clean Up, Clean Up!”

You probably recall the children’s song that Barney made so famous, “Clean up, clean up, Everybody, let’s clean up”.  Those words are being said, if not sung in food manufacturers and processors today as they are being pushed by consumers to clean up their ingredients. Food processors are reducing added sugars, eliminating or reducing genetically engineered ingredients (GMO), replacing saturated fats with “good” fats and eliminating synthetic colors and flavors. But these changes cannot happen overnight, the new or “clean” ingredients must be tested and phased in.

How do you go about this transition? With strong formula management.

Comprehensive formula management includes the following characteristics:

• Version control
• Batch sizes
• Ingredient types
• Potency management
• Catch weight
• Co and by-products
• Scalability

For example, by creating multiple versions of a formula differentiated by varying effectivity dates and different batch sizes, you can slowly phase in ingredients starting with smaller test batches and eventually phasing up to production size batches.

You will also want to closely manage the ingredients themselves by indicating whether each ingredient is active, compensating or a filler. A compensating ingredient is one whose quantity is calculated based on an attribute of another ingredient. In other words, the proportion of this ingredient is not fixed. As you phase out sucrose and replace it with stevia, you can manage the amount of sugar for a specific batch based on the concentration of the stevia. In this example, the sucrose is a compensating ingredient and the stevia is a filler ingredient with potency. You can use potency management to define the concentration of an ingredient as well as adjust prices for purchase orders and sales orders.

90% accuracy in formula management is not enough.

Clean up your recipes and increase your formula accuracy with Microsoft Dynamics 365 for Finance and Operations formula management.

 

Best thing since sliced bread

We’ve all heard that familiar adage describing something new and inventive that is all the rage. Something we didn’t know we couldn’t live without. Anticipating what the consumer will perceive as the next food trend she can’t live without is the perfect task for machine learning. Combining disparate data sources such as sales history and internet search data such as Google Trends arms food manufactures with the insight needed to have the right products on the shelf.

Cricket flour scones, anyone?

One of the easiest ways to work with the data is using Power BI. With Power BI, we can choose from many data sources to easily build graphical reports and dashboards.

Notice in the image to the right the data sources available in Power BI. For data coming from Azure Machine Learning, we would select the fourth choice, to Get Data from Databases. It is important to point out that we can import data from multiple sources, building a Dataset in which to perform our analysis.

In the image above, we see how data can be imported from multiple sources to build out tile visualizations such as charts, tree maps, tables, scatter and bubble graphs just to list a few. You will group your tiles into dashboards and you can view your Power BI dashboards from any device.

Imagine pulling out your smart phone in the parking lot of your biggest distributor and reviewing the analytics immediately prior to your meeting.

 

 

Now you know for sure – no cricket flour scones (yet).