launching your data-oriented (big data) startup


it always pains me to say ‘big data’ – especially as someone who started companies in the space before the label and worked with companies who have long dealt with massive amounts of data  – but i also know that i am not a marketer 🙂

big data keeps getting bigger. last year, VC firms invested $3.6B – 75 percent of what they invested in the previous five years combined. the pace has continued this year, with several firms announcing new funding rounds in the tens and even hundreds of millions of dollars.

for aspiring big data entrepreneurs, it’s exciting – and intimidating. i meet a lot of smart, talented engineers who want to work in big data but don’t know where to start.

i tell them to focus on an area where you can have a big impact, including feature engineering, mining email for B2B, applications for CRM, data governance, vertical integration, health care solutions – big data can drive health care savings of $300B according to a recent study – and tying into existing consumer properties such as facebook or linkedin to drive sales leads.

other areas, like data visualization or databases, are important but saturated, though there may be an opportunity to build next-generation databases using time series data. still others, like personalization technology, are better for established companies like google and facebook that have the data to train their image and voice recognition models.

once you focus and develop your big data idea, how do you turn that idea into a company?

turning your big data idea into a company

my advice in brief: be a painkiller rather than a vitamin, build and sell for enterprise customers, and remember that even with big data, less can be more.

be a painkiller, not a vitamin

like so many entrepreneurs, i love the technical challenge of programming. i started coding in fourth grade and have never stopped. so i understand how founders can be enchanted by the technical wizardry behind their products, especially in fields like data and machine learning.

but the corporate customers who are deciding whether to buy the product will be asking a set of questions with a very different focus. questions like: what’s the ROI here? will your proposed solution integrate well with our business culture? will it help move my production workloads?

one way to stay focused is to remind yourself to be a painkiller, not a vitamin. vitamins are great, but painkillers are vital. use technology to build a product that customers need – now.

i always ask founders in our first meeting why they made certain technical decisions. if you don’t know why you selected a particular technology and how your decision helps the customer, i would be hard-pressed to back your company.

build and sell for the enterprise

startups need to sell. in big data and machine learning, most customers will be enterprise customers. and most startups greatly underestimate what it means to be enterprise-ready.

my two bits of advice: first, if you’re an engineer, be sure to work closely with a product person, business person, or CIO so that you understand what it really means to sell to the enterprise. as a venture investor, i often introduce people to one another for precisely this purpose.

second, manage the gap between perception and reality. there are so many possibilities for big data, but there is also a lot of hype. manage the expectations of CMOs and CIOs so that you do not under-deliver at the start of what may otherwise be a lucrative long-term relationship.

understand the “why” of data storage

we all know how easy and efficient it is to store data today. in three decades, the cost of storing a gigabyte has gone from thousands of dollars to a few pennies. but now people have a tendency to store data without knowing how they want to use it. at some point, you enter a “data obesity” state where data storage, maintenance, and upkeep cost too much and slow you down.

even in a data-driven world, you shouldn’t default to storing every bit of data. instead, stop and ask yourself: do i have an idea of how i or somebody else in my company wants to use this data in the future? data storage still consumes energy and resources. before you store data, consider whether it will ever help you make a decision or deliver a product or service.

whether it is the onrush of data from sensors, advances in machine learning and deep-belief nets, or new modes of virtual reality, we are swimming in new information and need to imagine what will create the next wave of extracting knowledge and insights from all of it.

as i learned at twitter/cloudera/composite software/microsoft, building tools that allow more people to access and ask questions of the data enables everyone to make better decisions more quickly. as an investor, i often wonder: what are the new opportunities that will be created that we haven’t even thought of?

calling all aspiring engineers: ready to make an impact this summer?

TCP_4128 (1)

when students visit me in office hours at stanford, they often ask me what i think they should do over the summer. take an internship at a big company to learn how google or facebook gets things done? or join a startup and learn how to scale the next big thing? or take the ultimate plunge and start my own company?

i’m always happy when students pause and take a look at all of the options that they have, since the first key question in making a good decision is to be thoughtful about what you want to do in the future. when i was vp of engineering at twitter, i would often ask engineering recruits: “what do you want to do after twitter?” the response was always: “huh?” few people are used to talking about leaving a job they haven’t even started yet. but it’s the right question because you want to think about how to prepare yourself to succeed not just in the next job but in the jobs after that.

of course, none of this is completely predictable: the world will change, the valley will change, and you will change. in less than twenty years, i’ve started a ph.d. program in computer science, worked on microsoft’s early cloud-computing efforts, founded and sold my own company, led software development of webOS at palm, led the engineering team at twitter as it quadrupled in size, became a venture investor, and even helped fix

but while i couldn’t have imagined all of this, i always considered where i wanted to go and then contemplated, job by job, the starting places that would give me the best experience to get to that point. i think this is even more important today, when there are more opportunities than ever for talented engineers. and it’s especially important for students looking at internships, where the pace of learning is so much faster and the new opportunities will present themselves in weeks and months rather than years.

there are some rules of thumb: if you want to start a company one day, joining a startup of 10-20 people and seeing what does and doesn’t work will help you understand what it takes to be a ceo and get a company off the ground. if you are thinking of graduate school in computer science or engineering, a big company will often give you the best technical problems to work on.

but if there is one rule of thumb that i emphasize to students, it’s to focus less on company size or stage, or even the product or service, and instead to think more about the people you’ll work with and what you can learn from them. are they really mentors? how much time will they spend with you? does the company offer opportunities for growth and for impact and responsibility?

these questions are not easy to answer in advance, but they are critical. summer should be about so much more than checking a resume box. it’s a chance to improve your skills, discover what you really enjoy doing and how you want to do it, and introduce you to the people who will help you get there.

to that end, i want to encourage all aspiring engineers to consider the kpcb engineering fellows program. it’s the kind of program i wished existed when i was coming out of school. it not only gives students a great experience at one of our portfolio companies but enables them to get cross-company experiences and puts them at the center of the vast kpcb network.

i’ve told every team that i’ve ever led that the strength of the valley is not any particular company. rather it’s the people who make the valley what it is. at a time when this ecosystem is just buzzing with activity, the kpcb engineering fellows program connects you with the people and experiences that matter most. for a budding technologist just stepping out into the world, it’s a pathway to achieving your dreams. the link is here. i hope that you’ll apply.

calling all engineers: uncle sam needs you


a year ago today, i began one of the most interesting stints of my career: i became part of the technology surge team brought in from the private sector to fix my time in washington was an eye-opener because i saw firsthand the endemic problems that undermine the government’s IT efforts, and i also saw what we need to do to overcome those problems.

time magazine did a great job describing the problems and how we tried to solve them in an article called “Obama’s Trauma Team.” one big mistake the government made was a mistake the private sector never makes: the government opened the website to everyone at once. people in the tech industry know to launch to an expanding population of users over time. when you start small with your launch, you can see the system under strain, fix it, and then scale it.

led in significant part by Mikey Dickerson, then a google site-reliability engineer, the team applied standard silicon valley rigor, behavior, and approaches to the salvaging effort. together with engineers from the contractors hired to build the site, we fixed by the administration’s deadline.

the Obama administration is now applying what it learned from all of us to its other challenges. in august it announced the U.S. Digital Services Playbook. consisting of 13 “plays” — each with a series of checklists and key questions — the playbook helps government agencies apply the tech industry’s best practices for product development.

the administration also launched the U.S. digital service, a small team of industry experts who will work with the agencies to improve how the government does IT. (the idea is similar to the UK’s Government Digital Service, which has transformed that country’s IT efforts.)  Mikey Dickerson is now both administrator of the digital service team and deputy federal cio. 

In addition, the administration has launched 18F, an internal consultancy that Federal agencies can hire to develop and deliver state-of-the-art digital services, and a new Veterans Affairs Digital Service team to help improve services for our nation’s heroes.

this is our industry’s chance to help federal IT become easier for people to use. citizens deserve technology services from the government that actually work. there are so many ways we can use the skills that we’ve honed in silicon valley to make a difference in our nation’s capital. is the one project we’ve all heard about, but it’s far from unique. the new open payments website, for instance, is supposed to “to help consumers understand the financial relationships between the health care industry, and physicians and teaching hospitals.” unfortunately, it’s so obtuse that even professional health care journalists can’t figure out how to use it. according to the new york times,  if it “were a consumer product, it would be returned to the manufacturer for a full refund.”

i say that this kind of performance is unacceptable today. and yet the government continues to hire expensive contractors who have shown that they’re great at meeting complicated regulations but don’t have a clue about solving real-world problems.

now, with the new U.S. Digital Service team, 18F, and the VA Digital Service, our industry has a real opportunity to help modernize government IT, make federal websites easier for citizens to use and more effective in solving their problems, and put everyone’s tax dollars to good use. It looks like things are beginning to change.

here’s where you come in. 

the U.S. Digital Service, 18F, and VA Digital Service are looking for a few good people who:

a: will join these teams full time

b: are willing to leave the valley for a year or two to work in public service

c: can find other tech experts to fill either of those first two capacities

earlier this month i hosted Todd Park, tech advisor to the White House based in the Valley, at the kleiner perkins offices in menlo park to introduce him to potential recruits for the government’s new digital service teams. we had a great turnout, but we still need more.

consider this a call to action.

i encourage you to reach out to Todd and Jennifer Anastasoff at to learn how you can help the government deliver a better digital experience to all Americans. maybe you’ll just want to cycle in for a year or two; maybe you can suggest someone for the team. either way, this is a chance to make a real difference in the life of our country. isn’t that why we got into technology in the first place?

designers + engineers + empathy = greatness


designers and engineers need a deeper understanding of each other’s craft to create truly great products. i’m convinced that engineers need to understand the experiences designers aim to create, even as designers need to understand just know how engineers will make those experience come to life. when the two groups interact for the greater good: they build phenomenal products, with minimal time resources.

it’s all a matter of empathy — loosely defined as understanding the feelings and thinking of others. during my time leading engineering and design on webOS, and later at Twitter, i’ve learned that empathy is core to a product team’s ability to move quickly from designers’ “what” to engineers’ “how.” said differently, a designer knows what to make, and an engineer knows how to make it. when they overcome the communication barrier that separates the what and the how, good things are certain to come.

at palm, for example, we had to deliver a complete reset of webOS, moving the entire platform to a web-centric model. to do that, we put together a unified group of four teams: one team on the kernel, another on the apps, a third on infrastructure and the fourth on design. by working as a unified group, the engineers could empathize with what designers wanted the experience to be, while the designers understood the constraints of the OS.  and because of that empathy, we delivered an entirely new webOS in less than a year. we had achieved a virtuous cycle of product design, the goal of every product company.

the notion of deep, cross-discipline understanding isn’t limited to software development. it can be just as effective when developing hardware, hardware/software systems, and even for manufacturing. it’s not even new. design for manufacturability and assembly methodologies – where designers actually consider whether their designs can be easily assembled and built – have been around for decades.

but without empathy — where the different roles innately understand each other’s goals, assumptions and constraints — those cross-discipline development teams are still prone to misunderstandings and delays. recognizing that there is craft in the what and the how is key – and for the leader to help their design and engineering teams seamlessly understand each other is key.

which brings up the question: how can people with different mindsets and goals understand each other’s thinking? my colleague john maeda, who also happens to live at the intersection of design and technology, suggests early stage companies let designers code and engineers design. while not everyone can make this crossover, those who do will bridge the groups and help accelerate development.  i’ve lived in the valley long enough to see these kind of hybrid design/engineers make a huge difference in companies and now in the startups that they are founding.

for even slightly more mature companies, i believe it’s the leaders who have to crossover  and interact with other teams. they become the bridges who make sure everyone’s on the same page, with the same understanding of goals and constraints. and they also make sure design comes at the beginning of the process. without it, empathy becomes a one-sided proposition, and that just won’t work. so it’s not enough for a leader to keep the “why” in focus for everyone anymore – they’re going to have to get their hands dirty in the what and how, or at least serve as a solid communication bridge.

great execution: balancing order and chaos


it’s human nature to prefer order over chaos. as a general rule, people want everything to be calm and predictable. we are generally as a species uncomfortable with turmoil.

ironically, in the tech industry upheaval is valued – we love turmoil. as entrepreneurs and venture capitalists, we’re on a constant hunt for new ways to disrupt markets and upend the status quo. the companies that we launch and fund are in a nonstop race for more engineers, more customers, and never-ending revisions that challenge the edges of the CEO’s sanity. working at these places can feel like riding a one-way ticket to crazy-town: an absolute sh*tshow of disorganization with no panic button to be found.

yet here’s the thing. If the environment at a startup isn’t crazy, then something’s wrong. It may seem counter-intuitive, but chaos is an essential ingredient in a startup – it is what catalyzes the innovation. think of it as a necessary state of brownian motion where ideas collide with other ideas as fueled by deadlines and desperation. there’s a limit to the chaos, however, and you immediately see what that looks like because products stop shipping.

the key to a great startup environment is finding balance, as steve jobs did at apple, jon rubinstein did at palm, and elon musk is doing at tesla motors. special kind of leaders know how to orchestrate the simultaneous demands of time, scope and quality. consider what happens when any of those elements go out of balance:

  • time/ fall too far behind a deadline, and you could miss a critical service level agreement.
  • scope/ allow unchecked scope creep, and you could end up with a bloated mess satisfying no one.
  • quality/ let quality decline, and your company’s reputation could get permanently damaged.

great leaders know how to keep the chaos driven by these three factors in balance, and constantly modulate the demands to make the chaos manageable for their company leaders and employees.

seeing is believing


today, i’m going to riff on a topic i haven’t written about before, but has interested me for years: computer vision on mobile phones. i believe advances in computer vision — combined with the compute power we now take for granted on our cell phones — could improve people’s lives in ways most of us haven’t imagined. i’d better explain how i reached this conclusion.

scientists have known for years that increased blink rates are a great predictor of tiredness or fatigue. that fact came in handy when i wanted to figure out if my 4-year-old daughter would go to bed at 8pm or at 9pm, since that usually meant a big difference in her bedtime routine. i decided to build a mobile app that would record her face and let me count how often she blinked, helping me predict when she would fall asleep — and making a happier evening for parents and child. that’s a pretty simple example of what I mean.

cardiio is a more-sophisticated app that leverages mobile phones’ cameras and compute power. hold up your iPhone to your face in a well-lit area, and cardiio uses the front-facing camera to look at the capillaries on your cheeks. the app then measures the light that’s being reflected to determine your heart rate — useful for tracking fitness levels, calorie burn, and even estimate your life expectancy.

i should mention i don’t have any investments in cardiio or any other mobile computer vision app. i just find the whole space really interesting. MIT, for example, has developed computer-vision algorithms that can tell the difference between frustrated and pleased smiles. now imagine mobile apps that interpret shoppers’ smiles and help retailers fine-tune their merchandising. retailers could also use mobile apps to analyze foot traffic for optimum cross-selling and impulse buys. and thanks to community efforts like PubFig and Labeled Faces in the Wild, computer vision software can recognize faces — with a high degree of confidence — across a wide variety of poses, expressions and conditions (recent NYT article on the advances). it won’t be long before that capability shows up in commercial-grade mobile apps.

mobile computer vision can also help us model our environment and improve crop yields. for years, scientists have been finding new ways to use near-infrared reflectance spectroscopy to detected crop mold and fungi contamination and insect infestation. it’s easy to imagine drones fitted with infrared cameras detecting early signs of infestation.

and then there’s augmented reality — potentially giving humans a sixth sense for understanding the world around us. Google Glass may be the best example so far, as developers continually add new apps that overlay information on what the wearer sees. But I wonder about the effect this sort of enhanced vision has on us. if you wear Oculus for seven hours, does it rewire your brain? for pro and con, mobile computer vision could have a dramatic impact on us and the world we live in.

here is a fun application of a convolutional neural net that i setup with caffe last weekend.

game of drones

Airware color logos

people all across the tech industry are talking about mobile  — one of the largest markets we’ve ever seen. a side-benefit of this mobile revolution is that, in the race fit computers in our pockets, the competitive forces also drove down the cost of components. components, by the way, that have advanced technology in unmanned aerial vehicles, or “drones.” and so we have an early market forming and hundreds of commercial applications just waiting to be built.​

i grew up tinkering with hardware (meccano ftw!), and the prospect of drones make me feel like a kid again. forgive the pun, but it doesn’t take a rocket scientist to know drones will be a big deal. but how? and, when? many smart folks in tech have been giving this a lot of thought. fred wilson posted this great pic of a drone in his apt. twitter engineer @stammy has been surveying all of the bay area with his setup. and chris dixon wrote this post about a16z’s investment in a new OS company for drones. so, there’s plenty of buzz about the commercial drone space and we’ve had our eye on it for some time,  and I’m now lucky enough to jump into this game of drones!

today, kleiner perkins is proud to announce our investment in airware. i will join chris ​and john (airware’s CEO) ​ on the board, and will get to work with the ​growing team. we see ​many great opportunities here at kleiner, and i’m privileged to be able to partner with airware​ to build a world-class aerial hardware company and information platform.​

we chose to invest in airware as the frontrunner for a few reasons:

  • platform: unique in their approach, they’re building a platform with software and hardware that will be flexible enough for any commercial application. that’s especially attractive to me because I have built platforms and know how powerful they can be in fueling an industry
  • talent: team is equal parts aerospace and silicon valley – they are plugged in to the industry, know what customers need, have assembled the right people
  • regulation/compliance: they are also unique in that they’ve taken a very proactive approach with regulatory agencies and insurance providers – getting at customers’ needs and addressing them. they will catalyze an ecosystem of commercial drone development – first internationally and then in the US following FAA regulation

for me personally, the interplay between hardware and software will trigger memories of my time with webOS as well as my time building composite software and the data processing product we built.  i’m looking forward to getting started on supporting jonathan and the team on their journey. it’s a dream come true for me to focus on a company like airware.

we are just beginning to scratch the surface of all the dazzling and world-changing commercial applications for drones. airware will help drive the OS for these machines and provide a platform for others to build on.

the future is unwritten. my partners and i at kleiner are fortunate to have our chance to help a great team grow and write their own story. our goal is that one day you be able to lift up the hood of any commercial drone and see airware powering every one of them. this really is a game of drones.


coreos: the os of warehouse-scale computing


sometimes as a VC, you can start to feel overwhelmed by all the old and new business and technical acronyms you hear. it’s a new language, of sorts. well, i just happened to find one that sounds like music to my ears:


try saying that out loud. there are a few variants. the version i’m most partial to is “operating system as a service.” OSaaS. now, is that even possible?

turns out, it is. the folks at CoreOS know a thing or two about this emergent field, and they have built the world’s first “operating system as a service.” that’s the kind of creative, technical breakthrough that gets me excited and reminds me of my time as an infrastructure engineer — at composite, microsoft, but especially my days at twitter.

one of CoreOS‘ key innovations is providing updates and patches without the need for major operating system migrations. for some enterprise Linux customers, this will be the last integration they’ll ever need. this is the type of solution i would have given my right arm for in previous roles, and i’m excited to be representing my colleagues at kleiner perkins on the board of CoreOS as we lead a Series A investment into the company and it’s CEO, Alex Polvi.

when i started in venture a few years ago, i did not envision doing a deal like this, largely because i didn’t expect a team to be tackling this. i should’ve known better…as i got to know alex and his team during their time in ycombinator, it was quickly apparent to me that tech was moving even faster than i was in industry. in a relatively short period of time, the CoreOS team has created a foundation for the next generaiton of warehouse-scale computing.

the “s” in SaaS can now mean many things…even entire operating systems. i’m excited that even after my days as an operator, i get the chance to partner with alex and his team at CoreOS, to dust off my knowledge of infrastructure and my days at twitter and microsoft to learn more about the next wave of technologies and help the team take the company to the next level.

the bigger story behind ClearStory’s Data Intelligence

i know what you’re thinking — “another big data startup.” yes, it’s true, but beyond the cliche, there is truth and opportunity, and when it comes to larger enterprise businesses, small decisions can have tremendous effects — good and not so good — on the trajectory of a business. that’s where a company like ClearStory Data comes in. as i wrote about earlier at the time of our investment, larger amounts of data will undoubtedly lead to more noise, so businesses may place a premium on harnessing deep machine learning technologies to not only help with data sifting and analysis, but to also facilitate discovery of insights that the human eye may be incapable of observing.

ClearStory’s approach, announced today, is novel. by combining the power of its platform with various data sources and collaboration, Data Intelligence is enabled via a simple and intuitive user experience. up to this point, while decisions in the enterprise were often in consultation with the data, the data itself may not have been so good or as good as it could be – let alone collaboration around the data. here, there’s a larger trend afoot in the market for data intelligence and relative implications for how this data can inform decisions which directly tie to real returns on investment decisions within the enterprise.

the discovery from harnessing big data in the enterprise will spawn many new opportunities. as an investor, i’m obviously excited to meet founders who are thinking about using data to architect new systems and solutions, as well as new business models. it’s the deep learning from the data which excites me most — as the saying goes, “in God i trust, all others must bring data.” well, i’d put a spin on that — others may also want to bring data analysis tools and systems that can answer questions about the data in a collaborative manner in real time. in other words, the data is great, but not good enough anymore. it’s this “anymore” which creates the opportunity for the next set of founders to build something new, backed by hard data, which couldn’t have been known before.

proud to partner with Nivi, Naval, and the AngelList crew.


well, summer vacation in technology is certainly over. it’s already been an exciting month with news from apple, news from twitter, and as i’m writing this post about our investment in angellist — blackberry is being taken private. and it’s only monday!

back to angellist. as you may have seen, nivi and naval announced a new funding round for angellist, and i’m proud to be leading the investment in nivi, naval, and their colleagues on behalf of kleiner perkins. i’ve known naval for a number of years (back to when he was  starting vast an early investor in composite software jim armstrong introduced us) and we have always wanted to work together. so, when this opportunity developed, there was no way we were going to miss out on partnering.

everyone by now knows that a platform like angellist democratizes financing for startups. it also saves founders time. it enables investors outside of the valley to get access to early-stage deals and helps companies outside the valley echo chamber gain visibility. at kleiner, (like many other firms i imagine) we watch the angellist newsfeed and network with other peers.

on a personal level, as a former seed investor (before joining kleiner), i find the concept of syndicates to be incredibly compelling, both as a follower or the lead on an opportunity. if this existed years ago, maybe i would’ve started my own fund for syndicates!

again, i am very excited to work with nivi, naval, and the angellist crew!