Personal blog of Yzmir Ramirez

Forethought, a 2018 TechCrunch Disrupt Battlefield participant, has a modern vision for enterprise search that uses AI to surface the content that matters most in the context of work. Its first use case involves customer service, but it has a broader ambition to work across the enterprise.

The startup takes a bit of an unusual approach to search. Instead of a keyword-driven experience we are used to with Google, Forethought uses an information retrieval model driven by artificial intelligence underpinnings that they then embed directly into the workflow, company co-founder and CEO Deon Nicholas told TechCrunch. They have dubbed their answer engine “Agatha.”

Much like any search product, it begins by indexing relevant content. Nicholas says they built the search engine to be able to index millions of documents at scale very quickly. It then uses natural language processing (NLP) and natural language understanding (NLU) to read the documents as a human would.

“We don’t work on keywords. You can ask questions without keywords and using synonyms to help understand what you actually mean, we can actually pull out the correct answer [from the content] and deliver it to you,” he said.

One of first use cases where they are seeing traction in is customer support. “Our AI, Agatha for Support, integrates into a company’s help desk software, either Zendesk, Salesforce Service Cloud, and then we [read] tickets and suggest answers and relevant knowledge base articles to help close tickets more efficiently,” Nicholas explained. He claims their approach has increased agent efficiency by 20-30 percent.

The plan is to eventually expand beyond the initial customer service use case into other areas of the enterprise and follow a similar path of indexing documents and embedding the solution into the tools that people are using to do their jobs.

When they reach beta or general release, they will operate as a cloud service where customers sign up, enter their Zendesk or Salesforce credentials (or whatever other products happen to be supported at that point) and the product begins indexing the content.

The founding team, mostly in their mid-20s, have had a passion for artificial intelligence since high school. In fact, Nicholas built an AI program to read his notes and quiz him on history while still in high school. Later, at the University of Waterloo, he published a paper on machine learning and had internships at Palantir, Facebook and Dropbox. His first job out of school was at Pure Storage. All these positions had a common thread of working with data and AI.

The company launched last year and they debuted Agatha in private beta four months ago. They currently have six companies participating, the first of which has been converted to a paying customer.

They have closed a pre-seed round of funding too, and although they weren’t prepared to share the amount, the investment was led by K9 Ventures. Village Global, Original Capital and other unnamed investors also participated.

McCarthyFinch sounds a bit like a law firm — and with good reason. The startup has developed an AI as a Service platform aimed at the legal profession. This week, it’s competing in the 2018 TechCrunch Disrupt Battlefield in San Francisco.

The company began life as a project at a leading New Zealand law firm, MinterEllisonRuddWatts. They wanted to look at how they could take advantage of AI to automate legal processes to make them more efficient, cost-effective and faster, according to company president Richard DeFrancisco.

“They were working on leveraging technology to become the law firm of the future, and they realized there were some pretty tremendous gaps,” he explained. They found a bunch of Ph.Ds working on artificial intelligence who worked with more than 30 lawyers over time to address those gaps by leveraging AI technology.

That internal project was spun out as a startup last year, emerging as an AI platform with 18 services. MinterEllison, along with New Zealand VC Goat Ventures, gave the fledgling company US$2.5 million in pre-seed money to get started.

The company looked at automating a lot of labor-intensive tasks related to legal document review and discovery such as document tagging. “Lawyers spend a lot of time tagging things with regards to what’s relevant and not relevant, and it’s not a good use of their time. We can go through millions of documents very quickly,” DeFrancisco said. He claims they can lower the time it takes to tag a set of documents in a lawsuit from weeks to minutes.

He says that one of their key differentiators is their use of natural language processing (NLP), which he says allows the company to understand language and nuance to interpret documents with a high level of accuracy, even when there are small data sets. Instead of requiring thousands of documents to train their models, which he says law firms don’t have time to do, they can begin to understand the gist of a case in as little as two or three documents with 90 percent accuracy, based on their tests.

They don’t actually want to sell their platform directly to law firms. Instead, they hope to market their artificial intelligence skills as a service to other software vendors with a legal bent who are looking to get smarter without building their own AI from scratch.

“What we are doing is going to technology service providers and talking to them about using our solution. We have restful APIs to integrate into their technology and do a Powered By-model,” DeFrancisco explained.

The startup currently has 10 trials going on. While he couldn’t name them, he did say that they include the largest law firm in Europe, largest global provider of legal information and the fastest growing SaaS company in history. They are also working on agreements with large systems integrators including Deloitte and Accenture to act as resellers of their solution.

While they are based in New Zealand, they plan to open a U.S. office in the Los Angeles area shortly after Disrupt. The engineering team will remain in New Zealand, and DeFrancisco will build the rest of the company in the U.S as it seeks to expand its reach. They also plan to start raising their next round of funding.

Meet PoLTE, a Dallas-based startup that wants to make location-tracking more efficient. Thanks to PoLTE’s software solution, logistics and shipment companies can much more easily track packages and goods. The startup is participating in TechCrunch’s Startup Battlefield at Disrupt SF.

If you want to use a connected device to track a package, you currently need a couple of things — a way to determine the location of the package, and a way to transmit this information over the air. The most straightforward way of doing it is by using a GPS chipset combined with a cellular chipset.

Systems-on-chip have made this easier as they usually integrate multiple modules. You can get a GPS signal and wireless capabilities in the same chip. While GPS is insanely accurate, it also requires a ton of battery just to position a device on a map. That’s why devices often triangulate your position using Wi-Fi combined with a database of Wi-Fi networks and their positions.

And yet, using GPS or Wi-Fi as well as an LTE modem doesn’t work if you want to track a container over multiple weeks or months. At some point, your device will run out of battery. Or you’ll have to spend a small fortune to buy a ton of trackers with big batteries.

PoLTE has developed a software solution that lets you turn data from the cell modem into location information. It works with existing modems and only requires a software update. The company has been working with Riot Micro for instance.

Behind the scene PoLTE’s magic happens on their servers. IoT devices don’t need to do any of the computing. They just need to send a tiny sample of LTE signals and PoLTE can figure out the location from their servers. Customers can then get this data using an API.

It only takes 300 bytes of data to get location information with precision of less than a few meters. You don’t need a powerful CPU, Wi-Fi, GPS or Bluetooth.

“We offer 80 percent cost reduction on IoT devices together with longer battery life,” CEO Ed Chao told me.

On the business side, PoLTE is using a software-as-a-service model. You can get started for free if you don’t need a lot of API calls. You then start paying depending on the size of your fleet of devices and the number of location requests.

It doesn’t really matter if the company finds a good business opportunity. PoLTE is a low-level technology company at heart. Its solution is interesting by itself and could help bigger companies that are looking for an efficient location-tracking solution.

After two years operating in stealth, Mint .com founder Aaron Patzer’s new startup Vital Software is open for business.

Patzer made the announcement Wednesday while on the Next Stage at Disrupt SF.

Patzer’s company, which he co-founded with Dr. Justin Schrager of Emory University, is an enterprise software business that aims to make emergency rooms visits easier and more efficient for patients and doctors. The company is tackling the ER experience first and sees opportunity for the software in a hospital or health care facility, Patzer said.

“It’s a terrible experience, and not just because of the emergency,” Patzer said while on stage.

The software features an easy patient check-in system and uses AI natural language processing to find out more from the incoming patient. The system is dynamic, meaning it can ask follow up questions to the incoming patient to gather more information. By the time nurses see the patient, they’re already equipped with the information they need. The software also provides updates to the patient, such as possible wait times.

The idea is to give doctors and nurses software that is useable, Patzer said, noting that software found in hospitals is outdated. “It’s literally Windows 98 software.”

The company is self-funded, although Patzer noted off stage that they plan to raise funds next year. The company has one customer, a large hospital system he couldn’t name, that is now trialing the software.

In his view, software is constant need for disruption. His timeline: about every 10 years. That just happens to put Mint.com, he said, in a spot ripe disruption.

Dropbox is a critically important tool for more than 500 million people, which is why we’re so excited to have founder and CEO Drew Houston on the Disrupt stage in September.

Dropbox launched back in 2007 and Houston has spent the last decade growing Dropbox to the behemoth it is today.

During that time, Houston has made some tough decisions.

A few years ago, Houston decided to move the Dropbox infrastructure off of AWS. In 2014, Houston chose to raise $500 million in debt financing to keep up pace with Box, which was considering an IPO at the time. And in March 2017, Dropbox took another $600 million in debt financing from JP Morgan.

Houston also reportedly turned down a nine-figure acquisition offer from Apple.

All the while, Houston led Dropbox to be cash-flow positive and grew the company to see a $1 billion revenue run rate as of last year.

And, of course, we can’t forget the decision to go public earlier this year.

Interestingly, Houston first told his story to a TechCrunch audience at TC50 in 2008 as part of the Startup Battlefield. In fact, you can check out the original pitch from TC50 right here.

At Disrupt SF in September, we’re excited to sit down with Houston to discuss his journey thus far, the decision to go public and the future of Dropbox.

The show runs from September 5 to September 7, and for the next week, our super early-bird tickets are still available.