You think the shopper is smart?

With the rise in consumer preferences towards natural, organic and non-GMO food, retailers are faced with the challenge of supplying fruits, vegetables, and protein with a shorter shelf life, and adjusting to these trends of a dynamic marketplace. 86% of shoppers are confident the food they buy is safe from germs and toxins, down from 91% in 2014. Retailers must become more operationally efficient or increase their stock to overcompensate for higher rates of spoilage in order to counteract shorter shelf life challenges. Planning for fresh produce is more complicated than for non-perishable goods. According to a BlueYonder study, 68% of shoppers feel disappointed with the freshness of their purchases, and 14% of shoppers seek organic certification.

By using machine learning solutions, retailers will be able to optimize the environmental conditions affecting spoilage. In addition, there are risks of being out of compliance on food, health and environmental safety regulations with very high penalty, like Walmart paid $81M in environmental compliance.

How can you keep up?

Grocery retailers generally have low profit margins, so slight improvements to efficiency are important. Our machine learning solution is aimed at helping retailers improve their management of shorter shelf life products, and ultimately their profitability through optimization of their energy cost and prediction of temperature control equipment failure.

• Energy Savings: In some cases, utilities can amount to up to 50% of profit margin for a store, and energy savings driven by machine learning translate immediately to profit margins. For example, within the perishable seafood or meat sections, overcooling is a significant cost that can automatically be optimized by sensors that measure temperature in a cooler or refrigerator.
• Effectivity and Efficiency: Better allocation of resources like people and machines is very useful for top and bottom line. E.g. out of stock inventory can lead to $24M lost sales per $1B retail sales. Automatic tracking of inventory levels can help increase productivity and also revenues.
• Predictive Maintenance: Because refrigeration equipment has to run 24 / 7, there are high breakdown rates of equipment. Sensing equipment can be applied to HVAC and Nitrogen equipment to predict failure ahead of time. Even just small freeze / thaw cycles can quickly damage product and lead to waste for retailers.
• Compliance: FSMA and EPA includes multiple guidelines for retailers and grocery stores to follow, with high penalties for out of compliance.
• Consumer behavior: Consumer preferences and potential trends can be identified and acted upon if predicted. The Amazon store could even track which products you are interested in, but had not purchased.
• Risk mitigation: We could observe financial transactions, customer behavior etc. to predict risks, fraud, shoplifting etc. automatically and proactively.

Organizations are already moving to smarter technology for help.

What if the store was also smart?

Grocery retailers could use advanced analytics through IOT and other technology to revamp the way they monitor their stores.

1. Video feeds
2. Point Of Sale sensors
3. Mobile phones / equipment of Associates in store
4. IR Motion Sensors
5. HVAC and Energy monitoring using sensing of temperature, pressure, humidity, Carbon Monoxide
6. Weight Mats
7. Parking Space sensor
8. Digital Signage
9. Gesture Recognition/ accelerometers
10. Door Hinge Sensor motion/ pressure
11. Wifi Router and connections
12. Shelf Weight
13. Air Filter/humidity
14. Lighting
15. Electricity, Water, gas meters
16. Spark (Temperature) for places this device is taken to

Pilot

We would perform A/B testing to measure sensor performance and outcome in the controlled environment. The specific experiment is for milk preservation and storage. We would like to measure energy saving, ensure compliance to FSMA and EPA and predict refrigerator breakdown and maintenance.

• • Refrigeration is the single most important factor in maintaining the safety of milk. By law, Grade A milk must be maintained at a temperature of 45 °F or below. Bacteria in milk will grow minimally below 45 °F. However, temperatures well below 40 °F are necessary to protect the milk’s quality. It is critical that these temperatures be maintained through warehousing, distribution, delivery and storage.
  • The cooler refrigerated milk is kept, the longer it lasts and the safer it is. As the product is allowed to warm, the bacteria grow more rapidly. Properly refrigerated, milk can withstand about two weeks’ storage.
  • Pilot program: For the cost of $15 per fridge/freezer we can monitor food’s temperature and receive audible and visual alarm when temperatures exceed its minimum or maximum temperature range. For $39 we could measure temperature, humidity as well as when the door is open and closed.

We would partner with two to three smaller to medium size organic grocery stores (Trader Joes) with the high traffic, ex: city to test the impact over two to three weeks period.

Compliance info for meat preservation and storage is provided below:

• https://www.fda.gov/downloads/Food/ResourcesForYou/HealthEducators/UCM109315.pdf
• https://www.fda.gov/aboutfda/transparency/basics/ucm194273.htm

Example use cases:

1. Predictive Device Maintenance to avoid compliance lapse (e.g. Fridge for Food Safety, Fire Safety equipment, lighting, etc.)
2. Hazard detection and prevention through monitoring of toxic substance spill and disposal (air filter, shelf weight and video sensor)
3. FSMA compliance across labels, food expiry, storage conditions, etc.
4. Health safety with store conditions like canopy use, weather, leaks etc.
5. Temperature, defrost and humidity monitoring for Ice-cream, meat, dairy, and pharmaceuticals
6. Video analysis to predict long lines and avoid bad customer experience or lack of lost customers increased productivity etc. by alerting and optimizing resource allocation
7. Video + Point Of Sale analysis for fraudulent transactions avoidance

A central monitoring within stores, and centrally can be created, to mimic the Nasa base in Houston, is always able to support all adventurers within the store. Roger that?

---

Sources

FMI U.S. Shopper Trends, 2016. Safe: A32. Fit health: A12. Sustain health: A9, A12. Community: A12, * The Hartman Group. Transparency, 2015.

http://www.cnsnews.com/news/article/wal-mart-pay-81-million-settlement-what-epa-calls-environmental-crimes

https://www.fda.gov/food/guidanceregulation/fsma/

https://www.epa.gov/hwgenerators/hazardous-waste-management-and-retail-sector

Amazon store  https://www.youtube.com/watch?v=NrmMk1Myrxc

https://foodsafetytech.com/tag/documentation/

http://www.securitygem.com/cao-gadgets-inexpensive-and-tiny-sensors-for-your-smart-home/

http://www.clemson.edu/extension/hgic/food/food_safety/handling/hgic3510.html

---

Team – March & the Machines

Ewelina Thompson, Akkaravuth Kopsombut, Andrew Kerosky, Ashwin Avasarala, Dhruv Chadha, Keenan Johnston

By: Trevor Gringas, Prashob Menon, Dheeraj Ravi, Joanna Si, DJ Thompson

Opportunity:

Fake news is “a made up story with an intention to deceive”. It originated in the BCs when Julius Caesar’s son-in-law, spread false rumors about Marc Antony, culminating in Marc Antony being declared a traitor in the Roman senate. So fake news is not a new phenomenon. However, in the last year, the engagement of users across fake news websites and content has increased significantly across all mediums and types (headlines, sources, content). In fact, in the final 3 months of the 2016 election season, the top 20 fake news articles had more interactions (shares, reactions, comments) than the top 20 real articles.¹ Furthermore, 62% of Americans get their news via social media, while 44% use Facebook, the top distributor of fake news.² This represents a major shift in the way individuals receive information. People are led to believe misleading and often completely inaccurate claims. Decisions that are made off of this information are more likely to be incorrect, leading to a serious threat to our democracy and integrity.

Media corporations are recovering from playing a part in either disseminating this news or inadvertently standing by. Governments have ordered certain social media sites to remove fake news or else face a hefty punishment (e.g. €50 million in Germany).³ Companies such as Google and Facebook are scrambling to find a solution and are investing millions in internal ideas and external partnerships. However, it is extremely difficult to come to consensus on what defines fake news. Often times, ideological underpinnings define one’s proclivity to call something fake or not. This is why our solution focuses on identifying:

• claims which are 100% false (e.g., “there are millions of illegal voters in the US”),
• scientific studies which have been disreputed (e.g., “power poses reduce cortisol levels and increase confidence”), and
• conspiracy theories (e.g., “the moon landing was staged”).

Satire and opinion pieces such as  articles from The Onion or a statement like “Russians undermined the US political process” are currently out of scope given that Artificial Intelligence (“AI”) is still far from being able to semantically understand words like a human. Human beings cannot even agree on such things; thus, it is unreasonable to expect AI to be able to do so in the near future.

 

 

Solution:

Check Yourself (“CY”) provides real-time fact checking solutions to minimize the acceptance of fake news. It combines natural language processing techniques with machine learning techniques to immediately flag fake content.

CY’s first approach will employ semantic analysis. Often times, fake news articles are purely clickbait and meant to induce someone to click on an article to generate ad revenue. These articles will have gibberish or unrelated content from the headlines. Our solution will first examine whether the headline and the body of an article are related/unrelated and then whether the content supports the headline. Furthermore, the CY solution leverages fact-checking websites or services to determine whether the actual content itself has anything explicitly fake. Verification would happen against established websites, academics, and other website attributes (e.g. domain name, Alexa web rank).

The second approach involves (i) identifying platform engagement (Facebook, Twitter), (ii) analyzing website tracker usage (ads, cookies, widgets) and patterns over time, and (iii) generating links between those items to predict relationships. In the past year, the proliferation of ad trackers has led to many domains being created for clickbait and then quickly being abandoned to avoid detection. Furthermore, these websites often link to common platforms and other websites where one can find patterns in fake news sources that are distinct from those created by established news sites. This will result in a neural network through which the CY algorithm may predict the probability that the source is fake.

Combining the above two approaches leads to a novel solution as it semantically analyzes the text and assesses the veracity of the source to generate a probability score for how fake an article is.

The first phase of this will be designed in-house by a data scientist. After devising a baseline result and target, we will then use crowdsourcing to improve upon the algorithm. Given our limited in-house resources and the novel nature of this problem, we want to maximize our potential for success by generating ideas from individuals from all disciplines and encouraging collaboration either through an in-house crowdsourcing platform, or through existing platforms such as InnoCentive. We also intend to build out a mobile application through which users may curate and select news subscriptions that would automatically be scored using the CY solution. The mobile app will allow users to submit feedback on the accuracy of CY’s probability scores.

The next stage in the company’s roadmap is to continuously improve on the product and incorporate other features beyond a mere probability score and in-article highlights. These could include a list of corroborating sources or a list of “real”/factual news articles on the same subject. In the long-term, the goal is to be able to apply the algorithm not only to written text articles, but to be able to convert verbal speech into text, subsequently run the algorithm, and have CY call out inaccuracies on an almost real-time basis. This long-term solution would take into account not just textual relationships but other things such as verbal intonations and facial muscle movements so that factors such as mood and facial expressions can help determine the likelihood of fake news. CY intends to be a real-time lie detector for all types of news mediums, print, video, and yes – even live in-person interviews. Impossible you say? Tell that to the folks at Google Brain who created the computing power to essentially perform real-time language translation. The computing power available today is rapidly increasing such that aspirations of this sort are indeed achievable.

Implementation/Pilot:

Pilot 1 will be run with articles on the 2016 election. Subject matter experts will be asked to evaluate our algorithm in real-time. We will place the experts in four conditions – liberal, conservative, independent, no affiliation – and run two experiments.

Experiment 1.  Assessing speed and accuracy of the CY algorithm on fake news sources. We will present the exact same news stories to each group. The algorithm, along with the experts, will evaluate the article and both the speed of the human experts’ comments and their assessments will be compared against those of the CY algorithm. Both fake news sources and legitimate sources will be tested.

Experiment 2.  The second phrase will involve snippets of phrases in the various articles (not opinionated statements, but facts or lies).

Pilot 2 will be conducted with an academic journal or newspaper. In line with our propensity for crowdsourcing and desire to collaborate across disciplines, our team will test the algorithms against a team of faculty and students fact-checking sources for publication.

The Competition:

Many companies are trying to solve this problem. As noted above, Facebook and Google are key developers in this space. Existing solutions largely consist of human fact checkers, but they are not as comprehensive in their approach as we are. Furthermore, human fact checkers are rarely able to provide feedback in real-time. Universities are also trying to solve this problem, and are doing so with small teams of students and faculties. The advantage CY has over universities as well as the tech giants is two-fold. First, we intend to create neural networks that span various news sites and search engines (Google, for e.g., currently relies only on its search algorithms and platforms)[1] Second, our focus on crowdsourcing the solution and crowdsourcing for further feedback allows for the best ideas in a newly emerging area.

Market Viability:

Even though our value proposition affects companies and customers, we will primarily start with a B2B product. We anticipate collaborating with a news aggregator as an initial keystone customer. Given the strength and connections of our Advisory Board, CY is confident that initial keystone customers will not be an issue. As more media and news aggregators adopt a fake news notifier, content producers themselves will be incentivized to use such a service as well. Large media companies have around 10-20 fact checkers on staff for any live debate. The media company cost for fact checkers alone results in about $600K-$1.2M (assuming they spend $60k per checker per year). Furthermore, these customers often use Twitter and Reddit and would find our service invaluable to confirm the veracity of statements/claims immediately. Even more staff is on hand for research publications and institutions to verify academic journals and articles prior to publication. We anticipate that CY would reduce at least 50% of the fact checking resources of a media company. Key to CY’s continued success is to gain quick adoption and serve as the go-to platform for real-time fact-checking solutions so that additional features (such as a suggested sources feature described above or a social sharing aspect) have distinct and sustainable value. The product will be offered as a subscription service for lower-usage customers, and then as a combination of a subscription + usage cost basis for larger customers.

Conclusion:

At this time, we are asking for $225K to cover development costs and expenses over the next year. The bulk of this funding would go towards hiring a data scientist with the remainder covering administrative cost including IT and Server costs. To supplement this, we are also working on securing grants from agencies who are keen to address the problem of fake news.

 

[1] For a selection of articles on the efficacy of crowdsourcing and its potential, please see: King, Andrew and Karim Lakhani. “Using Open Innovation to Identify the Best Ideas.” Sloan Management Review 55(1):SMR 466.

Boudeau, Kevin J. and Karim R. Lakhani. “Using the Crowd as an Innovation Partner.” Harvard Business Review April 2013 R3104C.

Parise, Salvatore, Eoin Whelan, and Steve Todd. “How Twitter Users Can Generate Better Ideas.” MIT Sloan Management Review (2015): 21.

Schlack, Julie Wittes. “HBR Web Article: Ask Your Customers for Predictions, Not Preferences” Harvard Business Review January (2015).

Sources:

¹https://www.buzzfeed.com/craigsilverman/viral-fake-election-news-outperformed-real-news-on-facebook?utm_term=.nbR6OEK6E#.ghz5aZk5Z

²https://techcrunch.com/2016/05/26/most-people-get-their-news-from-social-media-says-report/

³https://yourstory.com/2017/04/faceboo-google-fake-news/

⁴https://www.ft.com/content/ba7d4020-1ad7-11e7-a266-12672483791a

 

                                                                                                       Scorch

 

Problem

Wildfires are getting worse each year. Acreage burned by wildfires and the number of wildfires burning more that 50,000 acres has continued to increase over the last 30 years.

 

Beyond the sheer devastation, forest fires are also a massive expense to the US government. Suppressing the fire requires hand crews, tankers, chemicals, and helicopters, not to mention firefighters. The fire management budget in the US is already up by 60% from a decade ago. More and more frequently, the Forest Service spends close to $2 billion per year on fighting fires.

Fire seasons have grown much longer, both in the US and across the globe. Earlier this year, more than 90 blazes scored 180,000 hectares, razed hundreds of homes, destroyed villages, cattle, and crops in Chile. As climate change continues to take hold, we will see more droughts and bigger and more severe fires will cause greater destruction of people’s lives and the economy of entire regions. CoreLogic estimates that US homes in the high or very high risk for wildfires could cost up to $237 billion to rebuild.

 

Opportunity

In the last decades, wildfire fighting has evolved by taking into account a more data driven approach to fighting, detecting and predicting these types of incidents. The first data driven models focused primarily on the study of weather. As the weather can determine how dangerous the fire can become, specialists started to analyze weather conditions during wildfires to fight them back.

Subsequently, other data were aggregated to build models for fire-behavior analysis to predict how flames will spread. Most of these models were reinforced with other data like the topography and fuel conditions, as well as smoke distribution forecasts. The problem of these approaches is that they look to mitigate the problem rather than prevented.

Then, the era of satellite imagery came and since then, satellite imagery has been used to try to detect forest fires. The problem is that it can not be used in real-time for fire prevention due to the amount of time needed to gather the data. In the same way, data from this type contain a high number of false positives for fires, ranging from hot asphalt parking lots to house fires to farmer burn piles.

 

What if we could use weather forecasting, topological data and real-time drone imagery to aid in forest fire protection?

 

Solution

Scorch will solve the wildfire problem with a technology that scouts fires before they flare up so firefighters can stop the flames early on. Evidence from Gamaya, a crop scouting company, and satellite imagery from NASA show that data science could effectively identify and detect fires. Scorch will use drones to fly over areas at high risk of wildfires and capture information through our hyperspectral camera (measures the light reflected Plants reflected by plants; reflection patterns change according to different stages in the life cycle and conditions of the plants). The image data will be processed through our algorithms that will be trained through crowdsourcing/ human intelligence tasks on Amazon Mechanical Turk and verified by fire experts. The results will be immediately sent out to the nearest fire station with recommendations to the firefighters and updates as the fire progresses.  As we collect more data, we will use machine learning algorithms to increase accuracy even further.

 

 

Concerns

One of the concerns is the frequency of false positives. Prior imaging software could falsely detect wildfires that are actually farmer burn piles or hot asphalt parking lots. This could lead to falsely deploying units that could actually cost more money. The Scorch algorithm would use historical fire perimeter boundaries and weather data to hone in on actual wildfires.

 

Another concern is the U.S. regulatory environment regarding drone use that might difficult initial testings. This could be solved by implementing the pilot program outside the U.S., notwithstanding this changes in international drone regulation might affect our ability to expand to relevant regions. This problem can be mitigated by working with authorities due to the public interest surrounding the problem that we are trying to solve.

 

Next steps and ask

In order to build this product, we need to hire developers and to set up crowdsourcing to train our imagery algorithm to properly focus in on wildfires. Additionally, we will do some testing to determine how much a drone could cover in order to be efficiently capture images useful for our purposes. In order to do this and develop our MVP, we are looking for a seed round of $150,000.

 

Sources

1. http://www.cnbc.com/2015/08/19/what-the-wildfires-are-costing-us.html
2. https://www.theguardian.com/world/2017/jan/25/chile-fire-firefighting-international-help
3. http://blog.galvanize.com/fighting-forest-fires-with-data-science/
4. http://www.popsci.com/year-wildfire#page-4
5. https://github.com/sallamander/forest-fires
6. https://www.ncbi.nlm.nih.gov/pubmed/12148070
7. http://lias.cis.rit.edu/projects/wasp
8. http://gamaya.com/

Everyone’s familiar with a class-action lawsuit where a bunch of families get together to sue a pharmaceutical company. Well, class-action lawsuits can be filed by shareholders, too. When a company acts fraudulently by issuing misleading statements to investors or hiding negative information about their firm, they can cause economic injury to shareholders (since stock prices will generally drop upon eventual disclosure of the information.) When that happens, the primary means of recourse is through the legal system. These cases are called securities class-action lawsuits.   

The Problem:

Currently, there are a handful of firms that specialize in this type of litigation. These plaintiff firms basically throw bodies at the problem by keeping tons of lawyers on staff. These employees manually comb through the news, read reports from industry bloggers, and keep their eye on the stock exchange, hoping that they can identify a possible opportunity for a suit before one of their competitors. This is an incredibly time-intensive and inefficient process. If a suit isn’t identified immediately, it takes an average of 77 days for a plaintiff firm to file.

Since law firms invest so much manpower in the identification process, they are incentivized to only take on the largest companies with the highest potential settlements. The average market capitalization for firms targeted by securities class action suits in 2016 was $9.08 Billion. This means that plaintiff firms are neglecting to hold the majority of malfeasant companies accountable for their actions because either they’re not big enough fish or lawyers are simply missing these opportunities because fraud in smaller companies doesn’t dominate the news cycle.

These firms generally take a shotgun approach to bringing these securities suits to court, leaning on high frequency and volume over quality. Historically, 44% of all securities class-action suits are dismissed, meaning that the courts are being inundated with frivolous cases. And it’s increasing year over year:

*Source: Stanford Law School

 

The Classy Solution:

Instead of relying on plaintiff lawyers and industry blogs (like Lyle Roger’s The 10b-5 Daily) to manually scan and analyze stock price data, we believe there is an opportunity to use machine learning to drive a uniquely efficient, highly competitive plaintiff firm that can hold more corporations accountable for their actions.

We propose the creation of Classy, a revolutionary algorithm that utilizes machine learning and human input to classify and predict securities lawsuits.

Our value proposition is threefold:

1. Bring Suits Faster

The Classy algorithm will be able to identify potential suits much more efficiently than traditional means. Filing the strongest, most profitable suits before a rival firm gives our firm a huge competitive edge.

The more efficient the system, the faster we can hold companies accountable for their actions and address client grievances.

2. Cover the Whole Market

Classy allow the plaintiff firm to hold companies of all sizes accountable, not just the largest, most profitable ones. Using augmented perception, we can analyze the entire market for stock patterns that demonstrate the potential for a suit (something that is impossible to do manually).

The algorithm can also cover a much larger breadth of news sources than a human analyst. This gives us a better ability to match a news item with a companion drop in stock prices that captures the potential for fiduciary misconduct.

3. Choose Better Cases

Classy can reduce the volume of frivolous lawsuits that get filed by allowing the firm to better prioritize their staffing structure and redistribute their human and financial resources away from searching for potential fraud and towards suit selection and execution.

Once the algorithm has enough data, we can extend its functionality to issue a prediction score to the firm, which would encompass the likelihood of winning the case as well as the estimated settlement value. The plaintiff firm can use this tool to help guide them away from filing unsuccessful, unproductive cases.

A Classy Design:

Classy combines external sensors with machine and human algorithms to predict the likelihood of securities misconduct of various firms and help analyze the success of a suit.

1. Monitor stock prices (sensory input)

We would use supervised machine learning and deep learning to flag precipitous stock price drops throughout the whole market.

2. Track relevant sentiment (sensory input)

We would use natural language processing and sentiment analysis to analyze relevant news items, identifying patterns of negative disclosures by a firm in the past or public apologies issued by CEOs.

3. Predict Outcomes (machine algorithm)

These sensory inputs would then be analyzed by a machine algorithm, which would use the data to create a likelihood score of disclosure malfeasance by the firm and the predicted settlement value.

4. Supervise with experienced plaintiff attorneys (human algorithm)

This information would then be transmitted through a human algorithm – plaintiff lawyers with years of experience and relationship expertise – who would then verify and expand upon the potential suits flagged by the machine algorithm. They would also provide feedback to the machine algorithm in order to improve its efficacy and accuracy over time.

Action Steps:

Validation: We want to assess our hypothesis by using historical data to test the validity of our claims: that Classy can bring cases faster, with a lower margin of error, and extend to a broader set of companies.

Poach a Partner: Rather than offer Classy as an available service to all plaintiff firms (which could lead to a race to the bottom), we would like to partner with a set of experienced plaintiff lawyers and start our own firm. This will give us a competitive edge over rival firms and a greater potential to monetize our efforts.  

Start Bringing Suits!

Our Funding Target:

We believe that validation can be achieved in two months and we have budgeted accordingly:

• Bloomberg Subscription:   $24,000
• Dow Jones DJX Subscription:   $800
• Two Months of a Developer’s Time: $25,000

     Total:                                                      $49,800

 

 

Sources:

http://securities.stanford.edu

https://ycharts.com/dashboard/

http://www.dandodiary.com/2017/01/articles/securities-litigation/2016-securities-lawsuit-filings-surge-record-levels/

https://10b5daily.com

SDR.ai: YOUR SMARTER SALES ASSISTANT

AUGMENTED INTELLIGENCE FINAL PITCH

TEAM CJEMS

I. Executive Summary

Business to business (B2B) sales continue to grow rapidly, but the services available to improve the process and success rate have been slow to keep up. As an intelligence platform that enhances the sales process, SDR.ai is here to change that. SDR.ai helps companies personalize customer messaging and better leverage data, resulting in improved sales results with less spending on internal resources.

II. The Problem and Opportunity

Sales Development Representatives (SDRs) help companies find and qualify sales leads to generate sales pipelines for Account Executives, who work with the customer to close deals. SDRs are a vital part of the sales process, as they weed out people that will not buy to find the ones that likely will, but their work is often repetitive and cyclical. SDRs work with large data sets and follow a clearly defined process, making them ideal candidates to integrate aspects of their jobs with automation. While it is still on the human SDR to understand the pain points of the prospective customer, an opportunity exists to better personalize messaging and make use of the available data to increase the final close rates for sales teams. Current SDR emails already utilize templates, but they do not consider what works and what doesn’t, and while it is possible to analyzing open / click rates of emails, linking this to revenue, or even spending time tweaking emails to add extra personalization, detracts from the time SDRs could spend on the phone with customers.

III.  Our Solution

SDR.ai aims to solve this problem by creating emails that mimic what actual SDRs sound like, without the template, considering the available data on what works vs. what doesn’t. It will integrate with existing popular CRMs, including Salesforce and Microsoft’s CRM, to learn from previous email exchanges and aggregate data in one place. Messages can be personalized to the recipient to create a more authentic message. Additionally, and most importantly, SDR.ai can send many more messages, increasing the volume of potential leads and the chances of bringing in additional revenue.

After initial training and manual emails, SDR.ai will continue to build smart responses, with the goal of handling everything up except phone calls, including scheduling and even finding the right person for SDRs to email from a prospective company (by using integrations like LinkedIn and Rapportive). Unlike real employees, SDR.ai is online 24/7, thus making it easier to connect with clients abroad, who normally must take time differences into account, losing valuable time and creating even longer sales cycles.

IV. Target Customer

SDR.ai is ideal for mid-market, high-growth small to medium businesses seeking increased sales and better conversion rates. More specifically, we are targeting B2B software-as-a-service companies (SaaS) that have defined sales processes that include a focus on inside sales. These types of companies are typically much more reliant on inside sales (37% of high-growth companies use inside sales as primary sales strategy). Moreover, many often lack the financial and human capital resources that make them ideal target customers for SDR.ai.

V. Business Model

As a software business, we will rely on a subscription-based business model that will also offer add-on consulting services to customers. On average, compensation for an SDR is high, with a base of $46k and an OTE variable comp of around $72k. Our pricing, therefore, will be based on our ability to reduce overhead spending on SDR’s and our ability to increase revenue per SDR.

VI. Product Development

We will first build a minimum viable product (MVP), focusing on developing initial automatic email creation, Salesforce integration, ability to send a high-volume of messages, and 24/7 availability capabilities. We plan to pilot SDR.ai after an MVP is created to gauge early feedback. To ensure we collect enough data to make the prototype of the product useful and accurate, we plan to partner with software (SaaS) companies that handle a large volume of leads. Given that this product can be tied directly to revenue generation, companies will likely be willing to try the prototype. From here, we could collect data on the most common language used, tied to deals that have been closed historically. By integrating with popular CRMs like Salesforce that already store historical data and emails used, we can determine how many emails on average it takes before deals are progressed from the SDR to the Account Executive. We also can take things a step further by looking at what is useful across different industry verticals, as CRMs already store this type of information.

After the pilot runs its course for a month or so (or whatever the average sales cycle length is), we can review the validation of the emails that were created with SDR.ai compared to those that were not. In short, we can validate that emails were (a) more readily responded to by either picking the right person in the organization (i.e. less emails that pass SDRs from one employee in a prospect client to another) and / or due to shortened response times or (b) opened and responded to by analyzing the language used in each response. The language can be continually refined and tweaked based on #2 until SDR.ai finds the right optimization of length, follow up, and personalization.

VII.  Competition

Currently, no direct competitors exist. Instead, there are three categories of potential “competitors.” Existing legacy CRM and sales management software providers do not provide the AI-enhanced capabilities of SDR.ai and will not move into such offerings, as they are focusing on their bread and butter offerings. Companies that offer predictive marketing software are growing, but they currently offer no clear solutions for sales teams. Finally, sales automation solutions providers provide valuable automation tools, but lack the AI-powered learning that SDR.ai would provide companies.

 VIII.  Risks and Mitigation Strategies

We believe SDR.ai can be an industry catalyst and fundamentally transform sales at large corporations. We recognize, however, that there are a few key risks we must assess and mitigate.

First, CRM providers such as Salesforce could potentially develop their own AI solution and effectively integrate it into their CRM offerings. To successfully mitigate this threat, we propose a two-part plan. First, SDR.ai will immediately file for patent protection in order to prolong our first mover advantage in order to secure as many customers as possible. Second, we will spend the coming years investing in an intuitive UI and integrating with customer databases outside of traditional CRM (e.g., BI tools). This will make customer switching costs higher and help address competitive threats.

Second, as a new platform SDR.ai may face scalability risks. Specifically, customers may find that the platform works well when their sales organization is smaller but that it fails to effectively scale as the database grows. In order to mitigate this, we propose a carefully designed pilot program that chooses rapidly growing companies. This strategy coupled with investment in a talented engineering staff will help address this risk. SDR.ai will not commit to aggressive growth until the pilot program demonstrates that the technology can effectively scale.

VIII. Investment Ask

We’re confident that we have the right team and resources to bring this product to market and grow the business; however, we are seeking $300,000 in seed funding to hire additional engineers to continue to build and refine the SDR.ai offering. While we have the correct business team in place, in order to bring this product to fruition, we need additional engineering resources.

 Sources

http://blog.persistiq.com/the-rise-of-sales-development?

https://www.salesforce.com/blog/2014/08/what-is-sales-development-gp.html

https://www.saleshacker.com/day-life-sales-development-rep/

https://resources.datanyze.com/blog/preparing-for-future-without-sdrs

 

 

• Genesis and Overview

Our team initially did a review of Pokemon Go’s success as an augmented reality platform. The game was superimposed over a layout of a real-world location on a user’s cellphone, and would interact through augmented reality. The game was widely popular and one of its unintended side-effects was that it motivated a large number of players to be more physically active. As each of the different goals were in different locations, players had to walk significant distances if they were particularly competitive. While not the primary purpose of the game, it got us thinking that there was definitely an angle to be tapped on.

With the increased adoption of wearables, the cult-like fitness industry is ripe for an AI integration. Tribe is our vision of an adaptive workout platform that combines machine learning and augmented reality. Tribe knows about all your physical attributes, such as height, weight, how much you ate yesterday, how much you slept and even what your fitness goals are. Based on all these, it can propose a workout for you that’s tailored to what your body needs at that very point in time. Each person will have a completely unique interaction with Tribe.

• The Problem

There are a whole range of fitness apps on the market now that attempt to guide users towards individual fitness goals. Most of them are fairly rigid and expect you to follow a set training program without any deviation. The classic couch-to-5k running guide is a clear example of this, where you can’t skip or modify a day’s workouts. Given the rigidity, it’s also not surprising that the attrition rate from these programs are remarkably high.

There are several solutions to the individual who desires varied, customized workouts.  seemingly simple solution to the individual that wants a customized, flexible workout, is to hire a personal trainer. However, that is a costly route which wouldn’t work for most people. To figure out how our platform could help, we tried to understand exactly what a personal trainer does, and how it could be replaced by AI. A personal trainer is usually well educated with a substantial background in kinesiology. They’re training style has also evolved based on the number of clients they’ve seen, and their ability to prevent injury before it happens, as well as understanding each client’s limits, and trying to push them farther. In essence, it was a skill that was based on interpretation after being experienced with multiple repetitions – sounds a lot like machine learning.

• The Solution

Enter Tribe, a software based solution that provides the first-ever augmented reality workout. Tribe will leverage a system of virtual gyms similar to Poke gyms with particular workout functions at each gym. This means that as the user base grows, groups will exist at certain gym locations and users will be performing the same exercises simultaneously. While this is not a group workout, there is a social component to it. Members, in a sense, are part of a workout tribe, similar to those that have been popping up around the globe in the social fitness craze of the last decade.

App-based fitness challenges are still virtual. One of the most well known virtual experiences is Fitbit’s “Adventures.” Through adventures, users can take a virtual journey through a wilderness hike, motivating them to reach a step goal. The experience is based on images, and there is no integration with the user’s surrounding environment.

• How it works

Step 1 – Collecting User Data. Users sign up to the platform and integrate it with their existing fitness tracker. For the purpose of this illustration, we’ll assume that Tribe is synced to a Fitbit. The user is then able to confirm the height and weight inputs that are retrieved from Fitbit as well as add their own personal information such as location, workout preferences and any past injuries. Based on all the information provided, Tribe is able to access historical fitness data from the Fitbit (heartrate, step count, workout intensity), and combine this with the workout goals to create a custom workout plan for the user.

Step 2- Location recommendation. When a user wants to have a workout, they can access Tribe and let it know that they’re trying to spend a given time on a workout – 60 min for example. Tribe will then process where you are, as well as the workout you need and give you directions to a workout location. Users will be able to elect from free locations (a public fitness area or an open field), or paid locations (a gym that Tribe has partnered with). The prompt to go to the location will be very similar to that of PokemonGo which engages the user in a game-like environment using augmented reality.

Step 3 – Workout Recommendation. Once you’re at the location, Tribe’s guided workout begins. This is where the platform’s core competency is. Tribe will already have all your data from the current day, and even the day before. It’ll be able to understand that you may have had a long night out with little sleep, or that you haven’t worked out in 4 days and have been sleeping 12 hours a day. Based on this information, as well as your physical attributes and goals, Tribe will be able to propose a workout that’s customized for you at that very point in time.

Step 4 – The Guided Workout. Very much like PokemonGo, Tribe will be able to guide you through your workout through augmented reality. Users will be able to place their phone in a visible place and see objectives on the screen. Perhaps an animated balloon 4 feet off the ground as a target for box jumps, or a cartoon character that holds a 2-minute plank with you. In addition to all this, the use of the front and back facing cameras on the phone will keep users honest by tracking their movements – very much like a more modern, portable version of Dance Dance Revolution.

• Challenges and proposed alterations

One of the challenges for this app is that unlike Pokemon Go, there will need to be workout specific spaces that account for the safety of users. A gym cannot exist in the middle of an intersection, for example. An exercise that requires the user to impact the ground cannot be on concrete.

• Marketing and partnerships

Some obvious marketing opportunities exist, starting with existing gyms and personal trainers. As the workouts could be anywhere, one potential location is existing gyms. Additionally, the trainers at the gym or independent trainers could craft workouts that can be integrated into the app and advertised to its users. Our target customer would already be interested in working out, and therefore our marketing campaigns will be targeted at the places they already are – in addition to gyms, also 5k races, farmers markets, and healthy food stores where athletes looking for more personalized workouts congregate. Lastly, we will use social media, particularly fitness influencers, to gain traction.

• Funding Requirement and Timeline

While we will need $2M to fully develop this technology, we are asking for $200K in funding to get us to a first pilot in one mid-sized city. This will enable us to create the initial exercise programming prototypes, develop the initial software prototype, identify locations within a city for gym activities, and partner with local organizations to gain a following. Once we can prove the model in a single city, we will expand to additional geographies.

 

A source of inefficiencies is an opportunity for MediLinx

MediLinx aims to be the leading data solutions provider for the healthcare market in Latin America. To do so, MediLinx serves as a third-party SaaS provider for private insurance companies in Latin America. Claims currently are filled out manually, submitted via mail to insurance companies, and then manually re-entered into the computer. The MediLinx solution removes excess inefficiencies by supporting digital record creation, claim automation, and data analytics in medical claim processing. Through this, we will be able to streamline the claim process to make it more efficient and help prevent fraudulent claims.

In Latin America, the current claim processing is slow, outdated and inefficient, thereby generating high administrative costs for the insurance company.

We plan to pilot in Mexico, where we have found and analyzed the main pain points for the insurance companies:

1. Claim processing is manual-intensive, requiring excess time spent reviewing and inputting claims
2. Fraud (15% of claims)
3. Hospitals take too long to fill and send needed forms

 

Entry strategy and Market Size

Our end customers are private insurance companies with operations within Latin America. Our initial target customer will be located in Mexico City, and once the concept is proven we will roll out to other major cities in Mexico.  After Mexico, we will scale through other major countries in Latin America. Most of our potential clients in Mexico have also a strong presence in Latin America, which will help us to achieve our international expansion.

The healthcare industry in Mexico is estimated in $77bn USD.  We are focusing on the private healthcare sector that holds a significant weight over the total expenditure (represents 45% of total health expenses).

Our client are private insurance companies. In Mexico there are over 40 health insurance companies, although 5 players control most of the market share.  They serve around 8% of the total population in Mexico, however their penetration in our initial market is much higher, 18%. This market is growing and is expected to double by 2030.

Product Overview

There are two main features of our product, 1) Digitization and 2) Automation.  The real value creation comes from the automation; nonetheless digitization is a main feature that enables the automation process.

The claim processing platform consists of a three user system 1) Insurance Company, 2) Patient, 3) Doctor.  The patient can upload their documents and fill out the claim forms directly on the platform, as well as the doctor.  There (s)he can make the requests to enter a claim for instant processing.

There are two components to the product the front end and back end.

• Front End (Sensors): This is basically the platform for users, composed of three different dashboards.  Each dashboard is designed and built to meet each customer’s requirements of the product.  Here doctors will input the patient’s claim report.  Patients will have a dashboard to view, edit and sign the reports.  Insurance will be able to access the final reports with the analytical results attached to each one.
• Back End (Algorithm): The core system is designed to process claims automatically.  The system will work through a machine learning process, by analyzing historic and new data to define the parameters that make a claim valid or invalid. The system will review the claim and the insurance policy of the patient in addition to his current status with the insurance to define the validity of the claim. The data is captured (digitized) through either by user-input in the platform or through a scanned picture of the documents. The system will process scanned documents and convert it to digital data, to then be analyzed by the automation process. Once a new claim goes through the system, it will automatically process the claim and determine if it is to be preapproved or subject to manual analysis. The system will build up the reports to output to the different user dashboards.  

The ask

The company requires 350k to launch.  30% Will be used for product development, which includes an MVP and the final platform.  We will use 20% for the pilot testing, before going into full development, to minimize risk.  The rest will be allocated for first year’s salaries and marketing expenses.

“Alexa, sing me a song.”

The Amazon echo is but one of several voice-powered devices that has been gaining wide customer appeal over the past year. During the holiday season, Alexa responds with a Christmas Carol. Imagine though if Alexa could respond with a song or playlist that corresponded with your mood just based on the tone of your voice.

Enter Beyond Verbal, a company going beyond data analytics to Emotions Analytics: “by decoding human vocal intonations into their underlying emotions in real-time, Emotions Analytics enables voice-powered devices, apps and solutions to interact with us on a human level, just as humans do.”[1]

How does it work? Beyond Verbal’s software examines 90 different markets in the voice that according to the company, can show loneliness, anger, and even love. The software simply needs a device with a microphone and then measures attributes including, Valence, Arousal, Temper, Mood groups. To date, the software has analyzed 2.3 million recorded voices, across 170 countries and included 21 years of research.

Recently, the company launched a free mobile phone app – Moodies. The app runs continuously and supplies a new emotion analysis every 15-20 seconds. I tested it out while researching for this post and was informed that I was exhibiting feelings of loneliness and seeking new fulfillment, searching warmth. Amusingly, it also stated that I exhibited signs of needing to be right and was ignoring reality — this was shortly after I said that I was skeptical. In a 2-3 minute span, the app proceeded to flash through the following assessments of my speech:

In terms of commercial use cases, the simple Alexa example above is one of many. Others have proposed its use in customer service call centers, executive coaching sessions, and as of late, even to diagnose disease! In fact, the Beyond Verbal team collaborated with the Mayo Clinic to analyze the speech of patients who had coronary artery disease and blocked arteries against those who were healthy to find that there was a distinct factor in the voice that was associated with the likelihood of heart disease! The research analyzed the voices of 88 patients, an additional 9 undergoing other tests, and 21 healthy people. The distinct identified voice factor was found when an individual was 19X more likely to have heart disease. [2][3]

The company has also identified another use case that will help hospitals, health/fitness app developers. They’re hoping to assist individuals understand how changes in a person’s environment may impact a patient. “We could see that people were tired in the morning, became more enthusiastic, active and creative throughout the day, but that their anger levels spiked around lunchtime,” says CEO Yuval Mor. [4] One of Beyond Verbal’s recent funders – Winnovation – mentioned that they could see smart devices continuously monitoring a person’s voice and sending alerts or making emergency calls if a medical problem requiring immediate attention occurred.

Competitors in this emotional analytics space include Swiss startup nViso SA, that developed a prediction process for patients needing tracheal intubation using facial muscle detection and Receptiviti, a Canadian startup that uses linguistics to predict emotions. [5]

Suggestions/Improvements:

A more forgiving industry – Voice recognition has come a long way since the first generation of Siri on Apple and even now, voice recognition is far from perfect. Assessing intonations in voice and attributing them to emotion is an even more herculean task. It’s interesting that the Company has jumped immediately to application in the healthcare space (recall the collaboration with the Mayo Clinic above), an area that arguably demand a high level of diagnostic accuracy. A more forgiving industry might be a good place to start commercializing the technology.

Accuracy across cultures and limited applicability for non-English speakers – As of now, it isn’t clear how the software accounts for cultural nuances in communication. Presumably testing on other languages will help some of this, but research should also be done on English-speakers from a variety of backgrounds and geographies as well. The company has begun to do research and run test on mandarin-speaking persons. Testing on other languages and different cultures will be important to expand the potential user base.

Mechanism for feedback on Moodies – While the mobile app is a step in the right direction to collecting more data points to confirm its research and algorithm, the app currently isn’t optimize for improvements or further learning. Users can’t confirm nor deny the accuracy of the predicted emotions. Moodies also doesn’t allow users to enter in their environmental conditions or what activity they are performing. If the goal is to eventually also use the data to see how changes in a person’s environment affects mood, a way to collect this data through the app ought to exist.

Privacy and Personal Data – Another area for concern and improvement is clarity around privacy concerns, particularly if Beyond Verbal intends to target solutions in the healthcare space. According to Matthew Celuszak, CEO of CrowdEmotion, a competitor to Beyond Verbal “in most countries, emotions are or will be treated as personal data and it is illegal to capture these without notification and/or consent.” Consent is only one area of concern as the laws and regulations around storage and use of personal data create additional burdens to companies. [5]

1. http://www.beyondverbal.com
2. http://www.beyondverbal.com/can-your-voice-tell-people-you-are-sick/
3. http://www.beyondverbal.com/beyond-verbal-wins-frost-sullivans-visionary-innovation-leadership-award-using-vocal-biomarkers-to-detect-health-conditions-using-tone-of-voice-2/
4. https://blogs.wsj.com/venturecapital/2014/09/18/beyond-verbal-raises-3-3-million-to-read-emotions-in-a-speakers-voice/
5. http://www.nanalyze.com/2017/04/artificial-intelligence-emotions/
6. https://thenextweb.com/apps/2014/01/23/beyond-verbal-releases-moodies-standalone-ios-app/#
7. http://www.mobihealthnews.com/content/emotion-detecting-voice-analytics-company-beyond-verbal-raises-33m
8. https://techcrunch.com/2016/10/17/science-and-technology-will-make-mental-and-emotional-wellbeing-scalable-accessible-and-cheap/

IntelliTunes – Predictive Analytics for Music Composition

 

The Problem

The music industry is notoriously fickle, and despite attempts to predict the next future chart topper, a significant amount of resources are devoted towards training and nurturing a group of artists with the hope that just a small percentage of them succeed. While this may seem like a challenge, the real challenge lies behind the scenes with the songwriters and musicians that come up with the tunes that we hear on the radio.

 

We believe that the music industry has seen a consistent genre shift every few years, from 70;s disco, to 80’s ballads, 90’s rock, 2000’s hip-hop and today’s Bieber. Given that we know a trend will last for a period of a few years, it would be worthwhile to invest in a system that would be able to predict these trends, and reduce the inefficiencies with song writing. Based on the type of music that’s on the top of the charts, and the historical chart toppers, we’re proposing an AI system that would be able to definitively compose a range of songs that would likely constitute future chart hits. Bieber would just be a mouthpiece for a significantly more intelligent machine.

 

Existing Platforms

There have been a few iterations of the proposed solution so far, but most of them have dealt with a library of past and present music, while IntelliTunes aims to be a predictive model of tomorrow’s music.

The Sony Computer Science Laboratory (“Sony CSL”) was probably the first commercial endeavour made to integrate AI and music composition. By analysing the musicality, tone, pitch and symphony in a range of music that was trending on top charts, the program was able to consolidate and create a unique pop song. It had all the markers from other top-ranking songs of the period, and hence should have also been a hit. However,much like Chef Watson, the result was something akin to serving caviar with peanut butter.

Platforms such as Pandora and Spotify also serve a need in the market by making an assessment of your future listening trends based on predictive analysis of your past music choices. While an astute use of AI, these platforms only serve to match your future listening needs with music that’s available on the market. It does not attempt to create songs that could be personalised for the individual listener.

 

The Proposed Solution

Much like how Chef Watson was the proposed AI solution to the culinary world, we expect IntelliTunes to be the the solution for the music industry. By design, IntelliTunes would be constantly consolidating the movements of songs on the charts, and identifying parameters such as time on the charts, sudden climbs, sudden drops and most importantly,region.

 

The program would utilise Deep Learning, which is particular type of machine learning whereby multiple layers of “neural networks” are programmed to process information between various input and output points – similar to a loose imitation of the human brain’s neural structure.This allows the AI platform to understand and model high-level abstractions in data, such as the patterns in a melody or the features in a person’s face.

The system would then be able to host a virtual library of micro-attributes of musicality, compose songs that would be expected to be desirable in the near future, given the movement and trends of today. An additional interesting application would be the ability to compose unique songs given a particular time period or genre for the listener that would like to create their own personal rendition of Metallica meets 40’s swing. Apps like Spotify have shown that giving users the ability to independently curate their music is a valuable proposition and creates especially sticky customers.

 

One of the great things about the proposed solution is that it would be language agnostic. IntelliTunes would be able to make predictions and compositions for songs across multiple geographies, because all it does it put together the notes to form a melody. Given the melody, a human songwriter will be able to piece in words with the pre-requisite amount of human emotion that an AI would not be able to replicate. An AI may have been able to come up to the tune of Nicki Minaj’s “Anaconda”, but it’s highly unlikely that it could have fathomed the lyrics.

 

Resources:

• http://www.flow-machines.com/
• https://www.csl.sony.fr/music.php
• https://www.nytimes.com/2017/01/22/arts/music/jukedeck-artificial-intelligence-songwriting.html?mcubz=1&_r=0
• https://futurism.com/a-new-ai-can-write-music-as-well-as-a

 

Team Members:

Joseph Gnanapragasm, Cen Qian, Allison Weil &  Rachel Chamberlain