Mythic nets $40M to create a new breed of efficient AI-focused hardware

Another huge financing round is coming in for an AI company today, this time for a startup called Mythic getting a fresh $40 million as it appears massive deals are closing left and right in the sector.

Mythic particularly focuses on the inference side of AI operations — basically making the calculation on the spot for something based off an extensively trained model. The chips are designed to be low power, small, and achieve the same kind of performance you’d expect from a GPU in terms of the lightning-fast operations that algorithms need to perform to figure out whether or not that thing your car is about to run into is a cat or just some text on the road. SoftBank Ventures led this most-recent round of funding, with a strategic investment also coming from Lockheed Martin Ventures. ARM executive Rene Haas will also be joining the company’s board of directors.

“The key to getting really high performance and really good energy efficiency is to keep everything on the chip,” Henry said. “The minute you have to go outside the chip to memory, you lose all performance and energy. It just goes out the window. Knowing that, we found that you can actually leverage flash memory in a very special way. The limit there is, it’s for inference only, but we’re only going after the inference market — it’s gonna be huge. On top of that, the challenge is getting the processors and memory as close together as possible so you don’t have to move around the data on the chip.”

Mythic, like other startups, is looking to ease the back-and-forth trips to memory on the processors in order to speed things up and lower the power consumption, and CEO Michael Henry says the company has figured out how to essentially do the operations — based in a field of mathematics called linear algebra — on flash memory itself.

Mythic’s approach is designed to be what Henry calls more analog. To visualize how it might work, imagine a set-up in Minecraft, with a number of different strings of blocks leading to an end gate. If you flipped a switch to turn 50 of those strings on with some unit value, leaving the rest off, and joined them at the end and saw the combined final result of the power, you would have completed something similar to an addition operation leading to a sum of 50 units. Mythic’s chips are designed to do something not so dissimilar, finding ways to complete those kinds of analog operations for addition and multiplication in order to handle the computational requirements for an inference operation. The end result, Henry says, consumes less power and dissipates less heat while still getting just enough accuracy to get the right solution (more technically: the calculations are 8-bit results).

After that, the challenge is sticking a layer on top of that to make it look and behave like a normal chip to a developer. The goal is to, like other players in the AI hardware space, just plug into frameworks like TensorFlow. Those frameworks abstract out all the complicated tooling and tuning required for such a specific piece of hardware and make it very approachable and easy for developers to start building machine learning projects. Andrew Feldman, CEO of another AI hardware startup called Cerebras Systems, said at the Goldman Sachs Technology and Internet conference last month that frameworks like TensorFlow had  most of the value Nvidia had building up an ecosystem for developers on its own system.

Henry, too, is a big TensorFlow fan. And for good reason: it’s because of frameworks like TensorFlow that allow next-generation chip ideas to even get off the ground in the first place. These kinds of frameworks, which have become increasingly popular with developers, have abstracted out the complexity of working with specific low-level hardware like a field programmable gate array (FPGA) or a GPU. That’s made building machine learning-based operations much easier for developers and led to an explosion of activity when it comes to machine learning, whether it’s speech or image recognition among a number of other use cases.

“Things like TensorFlow make our lives so much easier,” Henry said. “Once you have a neural network described on TensorFlow, it’s on us to take that and translate that onto our chip. We can abstract that difficulty by having an automatic compiler.”

While many of these companies are talking about getting massive performance gains over a GPU — and, to be sure, Henry hopes that’ll be the case — the near term goal for Mythic is to match the performance of a $1,000 GPU while showing it can take up less space and consume less power. There’s a market for the card that customers can hot swap in right away. Henry says the company is focused on using a PCI-E interface, a very common plug-and-play system, and that’s it.

The challenge for Mythic, however, is going to get into the actual design of some of the hardware that comes out. It’s one thing to sell a bunch of cards that companies can stick into their existing hardware, but it’s another to get embedded into the actual pieces of hardware themselves — which is what’s going to need to happen if it wants to be a true workhorse for devices on the edge, like security cameras or things handling speech recognition. That makes the buying cycle a little more difficult, but at the same time, there will be billions of devices out there that need advanced hardware to power their inference operations.

“If we can sell a PCI card, you buy it and drop it in right away, but those are usually for low-volume, high-selling price products,” Henry said. “The other customers we serve design you into the hardware products. That’s a longer cycle, that can take upwards of a year. For that, typically the volumes are much higher. The nice thing is that you’re really really sticky. If they design you into a product you’re really sticky. We can go after both, we can go after board sales, and then go after design.”

There are probably going to be two big walls to Mythic, much less any of the other players out there. The first is that none of these companies have shipped a product. While Mythic, or other companies, might have a proof-of-concept chip that can drop on the table, getting a production-ready piece of next-generation silicon is a dramatic undertaking. Then there’s the process of not only getting people to buy the hardware, but actually convincing them that they’ll have the systems in place to ensure that developers will build on that hardware. Mythic says it plans to have a sample for customers by the end of the year, with a production product by 2019.

That also explains why Mythic, along with those other startups, are able to raise enormous rounds of money — which means there’s going to be a lot of competition amongst all of them. Here’s a quick list of what fundraising has happened so far: SambaNova Systems raised $56 million last week; Graphcore raised $50 million in November last year; Cerebras Systems’s first round was $25 million in December 2016; and this isn’t even counting an increasing amount of activity happening among companies in China. There’s still definitely a segment of investors that consider the space way too hot (and there is, indeed, a ton of funding) or potentially unnecessary if you don’t need the bleeding edge efficiency or power of these products.

And there are, of course, the elephants in the room in the form of Nvidia and to a lesser extent Intel. The latter is betting big on FPGA and other products, while Nvidia has snapped up most of the market thanks to GPUs being much more efficient at the kind of math needed for AI. The play for all these startups is they can be faster, more efficient, or in the case of Mythic, cheaper than all those other options. It remains to be seen whether they’ll unseat Nvidia, but nonetheless there’s an enormous amount of funding flowing in.

“The question is, is someone going to be able to beat Nvidia when they have the valuation and cash reserves,” Henry said. “But the thing, is we’re in a different market. We’re going after the edge, we’re going after things embedded inside phones and cars and drones and robotics, for applications like AR and VR, and it’s just really a different market. When investors analyze us they have to think of us differently. They don’t think, is this the one that wins Nvidia, they think, are one or more of these powder keg markets explode. It’s a different conversation for us because we’re an edge company.”


Bear Flag Robotics wants to sell an autonomous tractor for farms

Autonomous vehicles are increasingly becoming the shiny object in Silicon Valley. But the opportunity doesn’t just extend to cars driving around the streets of a major metropolitan area, and Igino Cafiero and Aubrey Donnellan hope to take it somewhere a little less obvious: the middle of an orchard.

Cafiero and Donnellan are building an autonomously-driven tractor as part of a startup called Bear Flag Robotics. The pair argue that there’s increasingly a struggle to find enough labor to work on farms, and even then, the costs are continuing to rise over time — leading to a need to increase those efficiencies on the actual field in addition to a lot of new technology like satellite imagery and computer vision to analyze the health of plants. The first product for Bear Flag Robotics is a self-driving tractor, and the company is coming out of Y Combinator’s winter class this year.

“We got a tour of an orchard and just how pronounced the labor problem is,” Donnellan said. “They’re struggling to fill seats on tractors. We talked to other growers in California. We kept hearing the same thing over and over: labor is one of the most significant pain points. It’s really hard to find quality labor. The workforce is aging out. They’re leaving the country and going into other industries.”

There are certainly a lot of technical challenges that go into it, and not just pertaining from having the right computer vision products in place in order to create an autonomous tractor. For example, the tractors have to be able to operate without a GPS signal, Donnellan said, simply because operating a tractor in an orchard may mean driving around with a ton of canopy cover — which could block the signal. It might be a little simpler to just drive down a path in an orchard, but there’s still quite a lot to consider, she said.

“We have this platform that we’ve plugged a ton of sensors into it,” Cafiero said. “That includes cameras. When you look forward, once we’ve automated the driving part, the sky’s the limit in terms of utilizing some of this technology once it’s out there. When we’re out there we can use these cameras, and be able to make recommendations and spot treatment in the field.”

When it comes to testing, Cafiero and Donnellan just go out to an orchard over in Sunnyvale a few times a week to see what some of the challenges growers face.

While finding labor has been a challenge, Cafiero acknowledges that there are still questions around undocumented labor when it comes to labor on those farms. He said, in the end, Bear Flag Robotics’ aim is to augment the workforce by taking away some of the more mundane tasks required on the fields. Cafiero also said that there’s a lot of reverse immigration happening from the U.S., leading to more of a labor shortage.

“The work itself is really tough work,” Donnellan said. “You’re in the field all day long, sometimes in inclement conditions. One of the tasks we’re automating is spraying, fungicides, herbicides, and these people out there, they’re wearing hazmat suits. It’s not good for their health to be doing these tasks in general. When you’re presented in higher paying jobs in other fields, there’s less of a case to go into that job, and there’s demand in a lot of other industries like construction [and other industries] where it’s easier work and better pay.”

Selling the actual tractor can also be a challenge, simply because potential customers will be buying their equipment down the road at sellers they know. If something breaks down, they need someone to come over, in person, as soon as possible to fix it or risk losing yield. And the major equipment providers may too see the need to start working on autonomous tools. Cafiero’s hope is that the startup will be able to work with local sellers and get into those channels, and that’s the only logical place to start. There might be some aim to scale up over time, but the company hopes to just get started with local dealerships for now.


Qualcomm’s former exec chair will exit after exploring an acquisition bid

There’s a new twist in the BroadQualm saga this afternoon as Qualcomm has said it won’t renominate Paul Jacobs, the former executive chairman of the company, after he notified the board that he decided to explore the possibility of making a proposal to acquire Qualcomm.

The last time we saw such a huge exploration to acquire a company was circa 2013, when Dell initiated a leveraged buyout to take the company private in a deal worth $24.4 billion. This would be of a dramatically larger scale, and there’s a report by the Financial Times that Jacobs approached Softbank as a potential partner in the buyout. Jacobs is the son of Irwin Jacobs, who founded Qualcomm, and rose to run the company as CEO from 2005 to 2014. Successfully completing a buyout of this scale would, as a result, end up keeping the company that his father founded in 1985 in the family.

“I am glad the board is willing to evaluate such a proposal, consistent with its fiduciary duties to shareholders,” Jacobs said in a statement. “It is unfortunate and disappointing they are attempting to remove me from the board at this time.”

All this comes following Broadcom’s decision to drop its plans to try to complete a hostile takeover of Qualcomm, which would consolidate two of the largest semiconductor companies in the world into a single unit. Qualcomm said the board of directors would instead consist of just 10 members.

“Following the withdrawal of Broadcom’s takeover proposal, Qualcomm is focused on executing its business plan and maximizing value for shareholders as an independent company,” the company said in a statement. “There can be no assurance that Dr. Jacobs can or will make a proposal, but, if he does, the Board will of course evaluate it consistent with its fiduciary duties to shareholders.”

Broadcom dropped its attempts after the Trump administration decided to block the deal altogether. The BroadQualm deal fell into purgatory following an investigation by the Committee on Foreign Investment in the United States, or CFIUS, and then eventually led to the administration putting a stop to the deal — and potentially any of that scale — while Broadcom was still based in Singapore. Broadcom had intended to move to the United States, but the timing was such that Qualcomm would end up avoiding Broadcom’s attempts at a hostile takeover.

BroadQualm has been filled with a number of twists and turns, coming to a chaotic head this week with the end of the deal. Qualcomm removed Jacobs from his role as executive chairman and installed an independent director, and then delayed the shareholder meeting that would give Broadcom an opportunity to pick up the votes to take over control of part of Qualcomm’s board of directors. The administration then handed down its judgment, and Qualcomm pushed up its shareholder meeting as a result to ten days following the decision.

“There are real opportunities to accelerate Qualcomm’s innovation success and strengthen its position in the global marketplace,” Jacobs said in the statement. “These opportunities are challenging as a standalone public company, and there are clear merits to exploring a path to take the company private in order to maximize the company’s long-term performance, deliver superior value to all stockholders, and bolster a critical contributor to American technology.”

It’s not clear if Jacobs would be able to piece together the partnerships necessary to complete a buyout of this scale. But it’s easy to read between the lines of Qualcomm’s statement — which, as always, has to say it will fulfill its fiduciary duty to its shareholders. The former CEO and executive chairman has quietly been a curious figure to this whole process, and it looks like the BroadQualm saga is nowhere near done.


NexGenT wants to rethink bootcamps with programs for network engineering certifications

Developer bootcamps — several-month training programs that are designed to help people get up to speed with the technical skills they need to become a developer — exploded in popularity in the early part of the decade, but there’s been a bit of a shakedown on the space recently.

And that could be a product of a lot of things, but for Jacob Hess and Terry Kim, it’s just not enough time to become a fully-fledged developer. With training in the Air Force, where both had to work on these kinds of compressed programs for entry-level technicians, both decided to try their own approach. The end result is NexGenT, which is own kind of bootcamp — but it’s for getting a certificate in network management, and not a one-size-fits-all sticker as a developer. That approach, which includes a 16-week class, is considerably more reasonable and helps get people industry-ready with a skill that’s teachable in that compressed period of time, Hess says. The company is launching out of Y Combinator’s winter class this year.

“There are 500,000 open IT jobs, but when you look at that number, what’s more interesting is so many of them are IT operation roles, and the remaining is software development,” Hess said. “The bigger pie in IT is non-software programming jobs. Cyber security is also huge because of the automation and AI. We want to create the stepping stone. Network engineering becomes a foundation for a lot of these jobs, whether you want to be a cloud architect and work for Amazon, it all starts with understanding and building a foundation around networking.”

The end result is a 16-week program where a batch of applicants gets a review, and a percentage of them are accepted into a cohort of students. They go through an engineering module, which teaches them the basics and mechanics of network engineering and learn about the IT industry. Students can go faster if they want — it’s primarily online — and then start working on labs where they are building their own lab, either physical or virtual. The process culminates in a project where the students have to roll out an HQ facility in two branch offices from design to technically implementing it.

The next phase is about getting them certifications for various technologies, which help them basically show that they are ready to start entering the workforce. Think of it as something similar to having a Github account where prospective employers can review the work, except the process is a lot more formalized and you end up with something concrete on the resume. The final phase is around career coaching and helping them get a job, which can last up to 6 months. Throughout this process, students have access to a mentor and live coaching where students can ask whatever questions they wish.

So, the process is not so dissimilar from the notion of a developer bootcamp. But at the same time, there’s a small-ish graveyard of developer bootcamps and some with issues. Galvanize in August said it would lay off around 11% of its staff, while Dev Bootcamp and Iron Yard shut down altogether. The knock on these camps is it’s hard to get developers ready to start shipping code in such a small period of time — but Kim argues that getting them certified and ready to be a network engineer is definitely something that’s doable in around 16 weeks.

“It’s more realistic,” Kim said. “For coding bootcamps, you have to go by off the portfolios and check their Github, and they have to pass that technical interview. In our world of IT operations, it’s not about the bachelor’s degree, it’s about the person having the knowledge. But the industry certifications come from third parties, and when they come out of our program and have two or three certifications. It’s enough to get into that entry-level job.”

It remains to be seen if this kind of an approach is going to work. NexGenT charges a tuition — around $12,000, which with maximum discounts hits around $6,500. The company offers a 36-month payment plan as well that comes with an enrollment fee, which stretches out that very steep ticket price. In reality, these zero-to-60 programs are designed to be for-profit, though there are some different models that take in a percentage of salary among other approaches. With that in mind, though, there’s always an opportunity to build a strong pipeline with certain companies, and if they can identify high-performing students they can offer more of a proof point and potentially use that as an opportunity to offer some variation of scholarship.

While this is more of a bootcamp-ish style program, there are already some IT certification programs through tools like Coursera. Google, in one instance, is offering financial aid for a batch of those students, and companies with deep pockets might be able to build out these kinds of pipeline programs on their own. Hess and Kim hope to offer some kind of high-touch approach, instead of just a class on a platform of many, that will give them an edge to be a preferred option.


Google expands its Cloud Platform region in the Netherlands

Google today announced that it has expanded its recently launched Cloud Platform region in the Netherlands with an additional zone. The investment, which is worth a reported 500 million euros, expands the existing Netherlands region from two to three regions. With this, all four of the Central European Google Cloud Platform zones now feature three zones (which are akin to what AWS would call “availability zones”) that allow developers to build highly available services across multiple data centers.

Google typically aims to have a least three zones in every region, so today’s announcement to expand its region in the Dutch province of Groningen doesn’t come as a major surprise.

With this move, Google is also making Cloud SpannerCloud BigtableManaged Instance Groups, and Cloud SQL available in the region.

Over the course of the last two years, Google has worked hard to expand its global data center footprint. While it still can’t compete with the likes of AWS and Azure, which currently offers more regions than any of its competitors, the company now has enough of a presence to be competitive in most markets.

In the near future, Google also plans to open regions in Los Angeles, Finland, Osaka and Hong Kong. The major blank spots on its current map remain Africa, China (for rather obvious reasons) and Eastern Europe, including Russia.


Cloud security startup Zscaler opens at $27.50, a pop of 72% on Nasdaq, raising $192M in its IPO

The first post-billion, big tech IPO of the year has opened with a bang. Zscaler, a security startup that confidentially filed for an IPO last year, started trading this morning as ZS on Nasdaq at a price of $27.50/share. This was a pop of 71.9  percent on its opening price of $16, and speaks to a bullish moment for security startups and potentially public listings for tech companies in general.

That could bode well for Dropbox, Spotify and others that are planning or considering public listings in the coming weeks and months.

As of 3:45PM Eastern time, the stock has gone significantly higher and has just reached a peak of $30.61 as it approaches the end of its first day of trading. We’ll continue to monitor the price as the day continues to see how the stock does, and also hear from the company itself.

Initially, Zscaler had expected to sell 10 million shares at a range between $10 and $12 per share, but interest led the company to expand that to 12 million shares at a $13-15 range, which then moved up to $16 and Zscaler last night raising $192 million giving it a valuation of over $1.9 billion — a sign of strong interest in the investor community that it’s now hoping will follow through in its debut and beyond.

Zscaler is a specialist in an area called software-defined perimeter (SDP) services, which allow enterprises and other organizations to better control how they allow employees to access apps and specific services within their IT networks: the idea is that rather than giving access to the full network, employees are authenticated just for the apps that they specifically need for their work.

SDP architectures have become increasingly popular in recent years as a way of better mitigating security threats in networks where employees are using a variety of devices, including their own private mobile phones, to access data and apps in corporate networks and in the cloud — both of which have become routes for malicious hackers to breach systems.

SDP services are being adopted by the likes of Google, and are being built by a number of other tech companies, both those that are looking to provide more value-added services around existing cloud or other IT offerings, and those that are already playing in the area of security, including Cisco, Check Point Software, EMC, Fortinet, Intel, Juniper Networks, Palo Alto Networks, Symantec (which has been involved in IP lawsuits with Zscaler) and more — which speaks both of the opportunity and challenge in the market for Zscaler. Estimates of the value of the market range from $7.8 billion to $11 billion by 2023.


Hootsuite nabs $50M in growth capital for its social media management platform, passes 16M customers

Over the last several years, social media has become a critical and central way for businesses to communicate, and market to, their customers. Now, one of the startups that helped spearhead this trend has raised a round of growth funding to expand its horizons. Hootsuite, the Vacouver-based social media management company that counts some 16 million businesses as customers, said today that it has raised $50 million in growth capital — specifically through a credit financing agreement — from CIBC Innovation Banking.

We asked Ryan Holmes, the co-founder and CEO, for details about its valuation and funding, and said that it will be used for more acquisitions in the near future, and with it the valuation is unchanged.

“We opted for to go with non-dilutive credit at this point and found a great partner and terms in CIBC,” he wrote in an email. “The company is cash flow positive and the facility will primarily be reserved for M&A purposes. There is no associated valuation, however our latest 409a is up from last year and growth is very strong.”

Notably, the last time Hootsuite raised money — way back in 2014 — the company was already valued at $1 billion. For some context, at the time it had 10 million businesses as customers, and today it has 16 million including what it says is 80 percent of the Fortune 1000, so it’s likely that its valuation has grown as well.

“This financing is a testament to the strong fundamentals behind Hootsuite and our ongoing commitment to innovation and growth as the clear leader in social media management,” said Greg Twinney, CFO of Hootsuite, in a statement. “The additional capital will help us scale even faster to bring the most innovative products and partnerships to market globally and help our customers strategically build their brands, businesses and customer relationships with social.”

The funding, according to the release, will also be used to expand its business in Asia Pacific, Europe and Latin America. It also plans to add in more tools to serve the needs of specific verticals like financial services, government and healthcare.

You may not know the name Hootsuite but you might recognise its mascot — an owl — and more specifically its corresponding shortened link — it starts with ‘’ — that is used a lot on Twitter, the social network that gave Hootsuite its first customers and ubiquity.

Things have moved along quite a bit since those early days, when Hootsuite first started as a side project for Holmes, who himself was running a marketing and advertising agency when he started it.

Social media is now the fastest-growing category for marketing spend — partly because of the popularity of social networking services like Facebook, Snapchat and Twitter; and partly because “eyeballs” can be better tracked and quantified on these networks over more legacy channels like print and outdoor ads. At the same time, presenting yourself as a business on a social network is getting harder and harder. Sites like Facebook are focused on trying to improve engagement, and that is leading it to rethink how it shares and emphasizes posts that are not organically created by normal people. On the other side, we’re seeing a new wave of privacy and data protection regulation come in that will change how data can be used across and within these sites.

All of this means that Hootsuite, and others that it competes with, need to get a lot smarter about what it offers to its customers, and how it offers it.

Starting as a modest tool that plugged into Twitter, Hootsuite itself now integrates with just about all of the major social platforms, most recently finally adding Instagram earlier this month. Its customers use a dashboard to both monitor a variety of social media platforms to track how their companies are being discussed, and also to send out messages to the world. And they now use that dashboard and Hootsuite for a growing array of other purposes, from placing ads to content marketing to analytics across an increasing number of platforms — a range of services that Hootsuite has developed both in-house and by way of acquisition.

One challenge that Hootsuite has had over the years has been the company’s focus on the freemium model, and how to convert its initially non-paying users into paying tiers with more premium offerings. Some of that expansion into new services appears to have helped tip the balance.

“In the past year, Hootsuite has seen tremendous growth from acquisitions like AdEspresso, to strategic partnerships with market leaders such as Adobe, to recognitions such as being named a leader in the Forrester Wave and G2 Crowd,” said Holmes in a statement. “This financing allows Hootsuite in continuing to create strong value for customers looking to unlock the power of social.”

Another challenge has been the fundamental fact that Hootsuite relies on third parties to essentially “complete” its offering: Hootsuite offers analytics and tools for marketing, but still needs to connect into social networks and their data pools in order to do that.

This makes the company somewhat dependent on the whims of those third parties. So, for example, if Twitter decides to either increase the fees it charges to Hootsuite, or tries to offer its own analytics and thereby cuts off some of Hootsuite’s access, this impacts the company.

One solution to this is to continue to integrate as many other platforms as possible, to create a position where its stronger because of the sum of its parts. Unsurprisingly, Hootsuite also says that some of the funding will be used to increase its partnerships and integrations.

More generally, we are seeing a trend of consolidation in the area of social media management, as several smaller, and more focused solutions are brought together under one umbrella to improve economies of scale, and also to build out that “hub” strategy, becoming more indispensable, by virtue of providing so much utility in one place.

As part of that trend, we’ve seen two of Hootsuite’s rivals, Sprinklr and (not an owl but another bird of prey), also grow by way of a spate of acquisitions.


Pilot raises $15M to bring bookkeeping into the modern era

The first time Waseem Daher, Jessica McKellar, and Jeff Arnold worked together on a startup, they built one that allowed administrators to patch security updates to a system without having to restart it.

So it might come as a bit of a surprise that the next big technical challenge the three MIT graduates want to tackle is bookkeeping . But after selling Ksplice to Oracle back in 2011, it was actually the financial software they had built internally that made the jaws of the finance teams at Oracle drop, Daher said. They had created a continuously-updating internal version of QuickBooks, keeping a close eye on their spending and accounting and not having do hire a bookkeeper to do so, out of pure frustration with the process. And today that’s basically launching as Pilot, a startup that has now raised $15 million in a financing round led by Index Ventures.

“If you look at the history of bookkeeping, it goes back to the 1400s,” Daher said. “Probably the oldest written records were of transactions. Around 1400s, we invented double-entry bookkeeping, a system for how money moves into and out of various accounts of companies. That system, as articulated in 1400 in Venice, is basically still what people do in every American business today. You hire a bookkeeper or bookkeeping firm, you send them all your stuff and they track and produce the set of books. The way it’s done today is the same way it’s done in the 90s, the 40s.”

When a company starts working with Pilot, the actual core experience on the customer side doesn’t really change all that much: they still work with a human on the other end. But the bookkeeper from Pilot is working with the internal tools they have built to bring in the data from the company, organize it and structure it, and produce a set of books that are more accurate than someone might have produced than just doing it by hand. Customers will get the kinds of questions you might expect from a normal bookkeeper as they look to clarify what’s happening, but in the end the process happens much more seamlessly. They can integrate directly with their existing services like Expensify or Gusto (or ask Pilot to help out with that) and then go from there.

That kind of human-software mix is something that’s increasingly common in services businesses — like Pilot — as the tech industry figures out what should be automated and what should still be handled by a person. There are still a lot of things that a person can catch, but there’s also the actual human relationship, which isn’t a kind of repetitive task you’d want to automate with an algorithm. To begin, Pilot isn’t trying to force companies to completely rip out their bookkeeping software and start from scratch, and instead start to collect the electronic information they already have.

“Uber’s like that, the drivers are humans but the software makes them much more effective,” Index Ventures’ Mike Volpi said. “You can see it in a lot of applications where in IT support there’s a few businesses like this, you troubleshoot using software, and when you can’t you fix it pass it to humans. In customer service chats, a lot of times it’s an AI, and when the questions get tricky enough it rolls over to humans. It’s interesting because there are tasks which humans are fundamentally needed and there are tasks that are mundane that software can do and the human can avoid doing. It’s an interesting thesis around this hybrid.”

Prior to Pilot, the team sold another company to Dropbox called Zulip, and spent some time at the company as it continued to scale up (Dropbox is now in the process of going public). Some of the challenge alone was somehow assembling a team that found some fascination with the intersection of accounting, machine learning and working directly with customers, but so far McKellar said that they’ve been able to put one together thus far. And, more importantly, now that they are starting to roll out their service they can start getting some perspective on the industry as a whole.

“I think people can get motivated by almost any problem if you know you’re tackling a big problem for many people,” McKellar said. “But there’s quite a lot of subtlety to what we’re building. The rules and principles of bookkeeping are well define but the real world is really messy, and designing the right systems to automate bookkeeping at scale is actually a tricky thing. We have an incredible engineering team that is able to tackle this with the right mindset it. The analogy you can draw is self-driving cars — that’s a system normally done by a human, everyone understands what it takes to drive a car, what actions you should take. It’s difficult for people to put into words, what are the rules given a set of inputs, but it needs to work and be reliable.”

As more and more of this information comes in, and more and more companies start to work with Pilot, they can start spotting trends in the industry. For example, if a 17th SaaS business with a similar business model to other Pilot companies signs up, they could down the line take a look at their info and spot potential discrepancies based on anonymized trend data picked up from other comparables in the industry — or do a better job of spotting inefficiencies or others. And there are some obvious funnels for this already, like getting the right information for tax purposes to accountants.

There’s going to be a lot of increasing activity in this space, though. Already you’re seeing some funded projects like botkeeper, which are looking to find some ways to automate a bookkeeping service. There’s nothing quite so formalized and an obvious tool that looks to take out QuickBooks (and, again, a lot of these seem to be playing nice for now), and there’s always the chance that Intuit could try to take on the space itself. But at the end of the day, Volpi says it’s based on the team that they’ve assembled — and that combination of humans and algorithms — that gives them a shot at succeeding.

“If you look at a fundamental level, the bookkeeping for the doctor’s office or florist, it is really all following the same underlying principles,” McKellar said. “One of the engineering challenges is to build the tooling and systems and software in a way that’s intelligent. It has to be a set of processes that can flexibly accommodate every vertical over time. In some sense this company, why we raised this, was to validate a huge hypothesis — it’s possible to automate bookkeeping at scale across a range of industries.”

Here’s the rest of the investors in this round, since it’s a long list: Patrick and John Collison, Drew Houston, Diane Greene, Frederic Kerrest, Hans Robertson, Adam D’Angelo, Paul English, Howard Lerman, Joshua Reeves, Tien Tzuo, as well as many others.


WeWork expands its Flatiron School education business to London with £1M in scholarships

 WeWork — the co-working startup valued at $20 billion with some 200,000 members across 200 locations globally — is continuing with its strategy of expanding into a wide array of adjacent operations to grow its business. Today the company announced that it will be expanding the coding-focused Flatiron School abroad, starting in London this June.
Alongside this, it’s also… Read More


Qualcomm’s pivotal shareholder meeting is bumped up following Broadcom hostile takeover block

 After getting delayed by at least a month or so following a regulatory investigation just days before the meeting was scheduled to take place, Qualcomm’s timetable to hold the meeting is now bumped up to ten days from now — and possibly before Broadcom may complete its move to the U.S. The new change is coming as part of a presidential order by the Trump administration to block… Read More

Powered by WordPress | Theme: Aeros 2.0 by