Introducing Technically Sentient - A Newsletter on the Tech and Business of A.I.

Posted on July 13, 2015
Filed Under About | Leave a Comment

The short version is, I have started a newsletter than you don’t want to miss. Go here to sign up.

Now for the longer explanation…

Back in 2002 I started work on a graduate degree in Computer Science, with a focus on Artificial Intelligence. I stopped after 18 months because I didn’t agree with the underlying philosophy of how they taught A.I. It was all top down, command and control stuff, at a time when bottoms up emergent models were just starting to show promise. But for the past 13 years, I have continued to follow the space, read papers, watch companies, etc.

Now that I’ve sold Backupify and have some time, I decided to make A.I. the primary focus of what I do going forward. I just made my first angel investment in Netra, which brings an interesting machine learning approach to video analytics, and I hope to make 20+ angel investments a year in this space. I think that over the next 2 years I can become one of the top angel investor in this area, and get really great deal flow as a result. I think I am better positioned than most early stage VCs to understand the technical and early market risk of many of these companies. Most importantly, I think over the next 5 years the focus of A.I. will move from “Machine Learning” to “Machine Thinking”, and as more higher level work becomes automated, providers of capital into these spaces will get disproportionate economic gains.

But my interest in A.I. is about a lot more than investing. What I love about the space is two things. First, it is a multidisciplinary space. People come to A.I. from Computer Science, Linguistics, Electrical Engineering, Robotics, Philosophy of Mind, Cognitive Science, Neurobiology, Medicine, Cognitive Psychology, and all the other fields you can get by combining those branches of study (e.g. Computational NeuroLinguistics). That cross disciplinary nature makes it incredibly interesting, because I get to follow advances in all kinds of areas.

The second thing I love about A.I. is that it is filled with all kinds of moral, legal, political, and ethical challenges as it moves forward. Building sentient machines will challenge many of our the underlying philosophies on which modern civilization is built, and as someone who is attracted to complex multi-dimensional problems, I think it will be exciting to help companies work through those.

To make sure I stay focused and move forward, I decided to start an email newsletter called Technically Sentient. It will be sent out a few times a month, but at my leisure, not on a regular schedule. It will focus on interesting businesses I’m seeing in the A.I. space (with some Robotics and Neurotechnology discussion as well - as these are the follow on areas of the intelligence revolution). I will also interview key people in the space, and write my own thoughts as well. I have found that writing often helps me clarify what I actually think about a topic, and I am excited to do more of it.

If you are in this part of the tech industry, go sign up. You won’t regret it. And if you are starting a company in the A.I. space, please send me a summary of it. That is where I want to focus my investing.

If you are raising capital for an early stage company that somehow incorporates artificial intelligence, please reach out to me. If you have cool stuff that you think I should include in my newsletter, send it to tips@technicallysentientdotcom.

Also, go here to sign up for my email newsletter.

Meetings and Communication Structures

Posted on June 14, 2015
Filed Under Business, Critical Thinking | Leave a Comment

The interesting thing about really small companies is that they like to boast about how much better they are than big companies. I did it too. When Backupify was 8 people and we never had any meetings, I talked about how cool we were and how much smarter we were than all those boring big companies with their meetings and structure and process, and how we would never have that crap. But that’s mainly because I was an idiot.

It turns out, big companies usually work the way they do for a reason. Think of it like a basketball team. If you are playing 2 on 2, you don’t need much of a strategy. One guy has the ball and the other moves around. You can run some simple plays like give-and-go, but mostly its improv. As you add players, it gets more complex. Say you were playing basketball with 10 guys on each side. Now there is no room on the court to move, particularly if everyone is running all over the place improvising. You either need to run coordinated plays, or you need to at least have a general philosophy about how the offense should work and where people should move based on what is happening.

The parallel in business is that only way to get big, really, is to master the processes and structures that make it easy for people to know what to do, and to get the context they need for their work.

It’s usually around 15 employees that you start to realize not everyone knows everything that is going on. Prior to that, you all sit around two or three tables all day, close together, and hear everything that everyone says. Around 15 people, for at least a few key hours a week, someone is always out - sick, vacation, travel, whatever. So you realize you need a way to recap all the key points for people. Most companies start having a weekly or monthly “all hands” meeting about this time. The All Hands Meeting becomes important, so people try not to miss it, and its not too hard to find some time every week or two that everyone can make it.

After 15 people, every time you double, you face this same problem. Your communication architecture breaks down again because the channels you use to communicate no longer reach everyone, so you have to add more channels. But, you don’t want to add so much overhead that it slows everything down. I know some companies that at 50+ employees, still meet every single day for a few minutes in the morning. I think that’s tough.

What I want to share in this post is what Backupify looked like at 80 people. There are probably lots of right ways to do this, so don’t take this as gospel, but this structure worked for us, so I am sharing it in case you find it helpful. This doesn’t include every meeting we had, by any means, but instead, is all the key meetings that I drove, or was deeply involved in as CEO.

Keep in mind that we were highly experimental with meeting and communications structures, and every quarter we tried new meetings (some stuck, some didn’t), new formats for existing meetings, and new configurations of attendees. What you read below is just where we were at exit.

1. Weekly Reports
Each executive emailed the group a weekly report. We had no standard format across teams. Everyone developed a format that worked for them. It was supposed to take no more than 10 minutes to put together. It basically tracked the progress of their group that week. Some execs used their quarterly goals, and tracked progress against them. Others were more detailed. These went out between 3pm Friday and 9am Monday. We never had “status” meetings because I think having a meeting just so people can report on status is inefficient. It’s easier to just read it.

2. Weekly Management Meeting
This was a 90 minute meeting that happened every Monday at 1pm. The first 45 minutes were the senior team, and the second 45 minutes alternated each week between either Sales/Marketing or Product/Engineering/Support. Junior executives and key individual contributors joined in the second half of the meeting if it was relevant.

We shared a Google Doc every Monday morning that had the following headings:

- Questions on Weekly Reports: This was anything that an exec wanted to discuss from someone else’s report.

- Customers: This was anything having to do with getting and keeping customers - things that impacted them.

- Product: This was new releases, problems, changes to scope or schedule, etc

- Team: This covered new hires, demotivated team members, HR issues, etc

- Stakeholders: This covered key partnerships, investors, funding, Board of Directors, etc

- What’s Not Working: I think this is self explanatory

- Other: A catch-all heading

Each week, executives went in before the meeting and put topics under each area, with their initials beside their topics. I would then choose which topics we covered at the meeting, based on what I thought was most important on the list. The format was one we stole from another company, and what I liked about it was that it forced us to look at the business through different lenses in each section. Sometimes issues would bleed across sections, and this forced us to acknowledge that Customers and Team might have different perspectives with a certain issue.

The “What’s Not Working” section was also useful because it encouraged people to bring up broken processes or problems that may have otherwise gone unnoticed.

3. Weekly Coffees
Once a week we had a 15 minute optional meeting in our main open office area where we quickly introduced new hires and covered any key topics that might be of interest to the team. It was mainly announcements.

4. Monthly All Hands Meeting
These meetings started around 4:30pm and ran about an hour. Every month we started by highlighting our goals for the year and our progress against them. We covered key metrics, new hires, product launches, etc. Sometimes an executive would give a 10 minute talk about a key topic. It could be a topic related to the business, or something more general about careers or whatever. For example, our VP of Engineering gave a talk called “Managing Up” which was all about how to make your boss successful, and it was one of the most popular talks we ever had. We usually went out to the bar for an hour after this.

5. Monthly Communications Course
I realized about two years in that many problems arise in companies because people don’t know how to talk to each other. So every month I personally taught a Communications Course that every employee had to take once a year. Usually it was filled with whoever started that month. The focus was on concepts like: how to have difficult conversations, or driving to clarity in your conversations, or how to sell ideas internally, etc.

6. Monthly Management Deep Dives
As the company grew, we inevitably had problems that required more than 45 minutes of management discussion. But, it was very difficult to block out 2-3 hours of time on the calendars of 5-8 busy executives on short notice. So we start putting placeholders in far in advance for monthly deep dives. We would hold 3 hours on a Tues or Wed afternoon. A few days before, we would decide what we would actually talk about at that meeting. Occasionally we would cancel them because we didn’t have any key issues to discuss, but that was rare.

7. Quarter Operating Reviews
We only started these our last year, but I really liked them. We blocked out one and a half days per quarter and had each department come in and present last quarter’s results, progress this quarter, and next quarter’s plan. It was a rolling quarterly thing. The exec team would be in all sections, and then each section also had key individual contributors come in from that team. It was a great meeting to get people in alignment, and helped me a ton as CEO. For example, knowing that once a quarter I got to take a Marketing deep dive and look at things like detailed analysis of PPC campaigns and leads from events we ran and landing page A/B testing results meant that I got to learn, and give my input, without having to nag the marketing team all the time about the details of what they were doing.

8. Twice a year management offsites
We did two offsites a year. The spring offsite in April was 2 days and was focused on whether or not the plan for the year was working, and also on executive development. The executive team would do presentations on areas where they had some expertise, and we talked a lot about team and culture. The fall offsite was 3 days and was focused on strategic planning. We started on the first day with a refresher about “what is strategy” and reviews of the business performance, market changes, and technology changes. Day two was all about taking that analysis and building some ideas for the next year, and Day three was all about finalizing that information into a formal plan.

The only other thing I will add to this is that we had an email address alignment@, and we used it to have important discussions that were far reaching, or send out key information about changes in the market, news, etc. All the execs were on it, as were about 15 key individual contributors. It was a good channel for debate and discussion.

The purpose of all this was to accomplish a few things. First of all, I was always trying to get lots of cross functional discussion going, because it was amazing to me how small a business can be and still have teams in silos. Secondly, I wanted to make sure there was ample opportunity to discuss key issues and problems in the business. One of the hardest things about being CEO is getting people to share bad news, so, you have to create forums where they can do that openly without being punished. Otherwise, problems fester and just get worse.

Over my time running Backupify I became fascinated with organizational structure and communications architecture for companies, and how to use it to get the best performance out of people. I’ve given it a lot of thought, and so, if you are interested in it too, or want to talk more about this, drop me a note and I’m always happy to chat.

Advanced Startup Management Techniques

Posted on May 2, 2015
Filed Under Business | Leave a Comment

Yesterday I spoke at an event and some of the stories I told about how I managed the team at Backupify got a good response - so much that I thought I should write about them. What follows is 3 management ideas that I wouldn’t suggest you use if you are just starting out, but might be useful to add to your toolkit if you are already competent at basic management tactics. The ideas are useful in specific situations that may occur as you build a startup, but probably don’t have the broad applicability of other management ideas.

Before I outline them though, I should let you know how I think about management at startups. I think most managers spend time coordinating work, checking up on work, and trying to motivate each individual. But the individual is part of a team and you have to constantly think about the team dynamic. Most people don’t.

I once read a comment by Gina Bianchini, and subsequently met her at a First Round Capital event where I got to discuss it with her for 5 minutes. Her comment changed the way I looked at management forever. What she said was “world class team chemistry trumps world class individual talent every time.” It’s brilliant. And it’s true. If you have a rockstar who isn’t a good fit with your team, pass on her.

As a manager, I always thought about team chemistry, how to get people to work better together, and communicate better. I felt it was a waste of time to try to teach them detailed things about their areas of work, because tech companies change so fast that even if you are an expert in marketing (or sales, or programming, or whatever) today, in 5 years the world will be different. So instead I tried to teach them new frameworks for thinking about things.

I wrote that so that you understand that if your view of management is dramatically different, these ideas may not work for you.

1. Create alignment by being the bad guy, or causing a commotion.
There is a great book called Us And Them: Understanding Your Tribal Mind. The book looks at many of the psychological studies that underpin group dynamics. At startups, there are lots of tribes, even though we don’t always realize it. There is the “speed wins, get it out the door, fix it later” tribe. They fight the “don’t release it until it is ready, even if that means missing the date” tribe. There is the “all press is good press (even if its bad press)” tribe. They fight the “what people say about you on social is your destiny” tribe, who believe that a few bad reviews sink a ship.

There are also workflow tribes. The “meetings suck and are a waste of time” tribe is always battling the “meetings are efficient if done right” tribe. There is the “plan the work and work the plan” tribe, who always battle with the “companies are made by last minute heroics” tribe.

Even though this book is mostly about ethnic, religious, and political tribes, I found the principles to be very adaptable to all the different philosophical tribes I encountered in the startup world. What’s great about “Us and Them” is it teaches you how to get tribes to work together. And one of the best ways to get tribes to work together is to have them align around a common enemy. That enemy could be an event, or a person, but either way, when they have to do it, they forget about their differences.

On multiple occasions at Backupify, I decided the team wasn’t working well together for some reason, and made myself the bad guy so that they could align against me on something. For example, there was a time when people were battling about budgets and couldn’t agree where we should cut. I announced that since they couldn’t agree, I was going to cut an area of the budget that none of them wanted cut. Soon enough they have fixed the budget on their own.

2. Force people to take another perspective.
The number one thing I look for in executive hires is the ability to see the world through someone else’s eyes, and engage on that level. It’s that corporate empathy that makes teams work well together. But in startups people tend to get territorial, and committed to their ways of doing things, and it is important that you break down those behaviors.

I once had two people who had very different ideas about how work should flow. They weren’t getting along and didn’t like each other despite both doing great work. I made them have lunch together, once a week, and gave them a series of topics to discuss each week. The rule was that they each had to take 10 minutes just listening to the other person’s point of view on that topic, and couldn’t speak. The first item to discuss on the list (which ties back to point #1 above) was for them to discuss how angry they were at me for making them do this. Their working relationship improved.

3. Gain support for ideas by making teams pitch both cases.
Startups face a lot of tough decisions. Sometimes people grumble, or even worse, leave the company, when a major decision doesn’t go their way. This is usually preventable if they have an understanding of how the decision was made, and feel like they had some input into it. Some CEOs try to do this by just meeting with everyone and listening to their ideas. I used a different process.

I like to take my top execs and individual contributors, split them into two teams, and give them each 2 hours to discuss a topic and come up with a presentation to justify their point of view. It’s important that you don’t let people pick which team they join - just assign them to a point of view to defend. This process has 3 major benefits.

First of all, each team will have some people who inevitably disagree with the point of view they are asked to defend. Those people actually make the team better by providing good counter points for the analysis, but more importantly, their resistance to the point of view tends to soften when they hear the analysis of others who support it.

Secondly, each team has to do a really good job of their analysis, because no one wants to be criticized by the CEO in public for having done a shoddy job. Teams have to be thorough even with they disagree with the point of view they are defending, or else they look stupid.

The third benefit is that by having each team present, everyone is forced to listen to detailed reasons, counterpoints, and counter-counterpoints, before a decision is made. Plus, when you get to the end and make a decision, you have all your top execs and key players aligned on all the issues, and why the decision was made. It is soooo much more effective than making a decision yourself and trying to communicate it down the chain of command where it will become misinterpreted and misunderstood.

The other nice thing about this process is, you can usually control it to get the outcome that you want, by asking the right questions and laying out the right ground rules. In the end, if you define the situation correctly, the team will usually all agree on the outcome you were hoping for.

There you have it. I used all three of these techniques multiple times at Backupify, and found them to be very useful. If you have questions, or want to buy me a beer and talk more, reach out to me on twitter.

keep looking »