We just made a new Series A investment in a company called Abgenial Systems. Dan Primack and Matt Marshall will pick this up soon enough. So I thought I should get out in front and discuss this a bit. Pardon me for being a bit oblique in what I am about to describe, but it is hard to strike a balance between transparency and pre-announcement when dealing with early stage investments. I don’t want to pre-announce or hype what they will do, but I do want to describe how we came to make the investment. It wasn't the usual "show us your Powerpoint and wait for the white smoke" dialog that characterizes most entrepreneur/VC interactions.
About a year ago I became interested in the phenomenon called mashups. VCs try to find Big Ideas that transform existing markets or create new ones. This seemed like a Big Idea, but packaged in a Small Box. I started thinking about something I eventually came to think of as The Recombinant Web -- a half step between Web 2.0 collaboration and the Holy Grail of a Semantic Web. After all, the motivation of mashup creators was to recombine otherwise useful silos into new, ever more useful experiences. Today’s web services are kind of like old, single-tasking PC applications. They exist largely isolated from each other on a common presentation screen. Then it was the CRT. Today it is the browser window.
The “integration” of Google’s web services or Yahoo’s web services is reminiscent the “integration” that existed in Lotus Symphony, the short-lived successor to 1-2-3. I especially like this description from 1985 because some things never change. A hit application spawned visions of a platform...
Lotus designed Symphony for access by other programmers. Hooks inside the software help developers write applications for the Symphony environment. Add-in applications directly from Lotus also are expected. Obviously, Lotus wants Symphony to be your only program. Whether this idea is valid depends on several factors, not the least of which is how well the package works with the expanded RAM of such computers as the IBM PC AT. With up to three megabytes possible, Symphony memory problems could become a thing of the past. (Creative Computing, February 1985, p.88)
Sounds vaguely familiar. "Hooks inside the platform" - the 1985 version of API, and "up to three megabytes possible. Memory problems could be a thing of the past." - the 1985 version of the infinite resource of the Googleplex.
Mashups emerged from the current developer community as a way to exploit the Web as that platform and break down the stack of proprietary, but "integrated" web applications from GYM. But “mashups” have lots of limitations as a methodology. Over time I distilled the issues to three fundamental objections.
- First, the author/user distinction doesn’t scale. You can’t possibly know what web services I might want to combine nor how. Only I do. I want to program the web. I don’t want you to do it for me any more than I want to go to a site that lists all known queries to search the web. A corollary to this point is that will be few (no) mashups that are viable standalone businesses, at least not at a scale that has an interesting economic consequences for an investor.
- Second, there is only so much "bookkeeping" that can be done efficiently in the browser. Integrating in the browser is like cooking on a camping stove. It's possible, but it wasn't designed for an elegant and synchronized multi-course dining experience. Complex and interesting re-combinations are not what browsers and AJAX were designed to support.
- Third, the idea of users combining arbitrary web services seemed too unbounded and too early. It seemed wrought with lots of nasty issues like service availability/fail-over, financial models, data formats, and a small issue of programming, programming, and programming.
I started blogging about mashups with the intention of learning-by-publishing. The more I wrote, the more I learned. The more I learned, the more convinced I became that these issues required a fundamental re-think of how to approach integration. For the concept of re-mixing to become mainstream, It had to be capable of handling sequencing of web services and flows of data, but with a user interaction model as simple as point and click or cut and paste. A tall order. The more I stared at concept of "mashups," the more convinced I became that the concept was too parochial.
Three or four years ago my friend Pete Kolstad introduced me to Rafael Bracho. Rafael was a co-founder of Active Software and a pioneer in the 1990’s evolution of enterprise application integration (EAI) software. Active Software went public with Rafael as CTO and eventually merged with WebMethods. Somewhere along the way, the original EAI vision of simplified application integration became bloated by layers upon layers of standards, registries, modeling languages, protocols, etc. and promise of nimble enterprise services-oriented architectures (SOA) was DOA.
We were looking at an EAI appliance company about 18 months ago and I asked Rafael to help us with the diligence. Luckily we lost that deal to another VC firm (it since has not gone well). Afterward, by pure serendipity for both of us, Rafael started telling me about a technology he and a partner had been working on for several years under the name Abgenial Systems. It was a radical re-think of how to integrate applications. It was so radical, in fact, that we didn’t understand where to use it, nor did the other VPs of Software Development that we asked to look at it. By starting with a new conceptual model for what integration and interoperability would mean, Rafael and his partner Jacoby were able to simplify the enterprise integration problem to the point of near triviality.
I was stumped as to how to make this an interesting business. WebMethods, Vitria, Tibco were yesterday’s news. Who cares if you could steal share from these guys? They were locked in a fight to the death on near-free software, bundled with low margin consulting services, and rewarded with market caps running at 1X revenues.
My most important observation from immersing myself in the concept of mashups was the recognition that mashups were just another form of application integration, done client side rather than server side. That launched me in to seeing a potential unification of innovative consumer applications with stultified enterprise applications.
All of a sudden the idea of Abgenial pivoted. Last Fall, Web 2.0 was white hot on “collaboration and mashing” and Enterprise Software was moribund with “legacy vendor consolidation and broken distribution models.” Software as a Service was the VC industry’s Great White Hope, combining Web distribution with Enterprise functionality. (Read this as low cost of distribution and tangible value.) Google, Microsoft, and Yahoo were increasingly veering towards web applications (calendars, maps, photo sharing, etc.) – free form of SaaS.
The Abgenial Pivot was to see the dots were connected. Consumer web….enterprise web…..Saas….legacy apps…. Mashups….SOA were all variations on a theme and that theme was <hype>Programming the Web</hype>. It was then that I went to my partners and asked to provide a $250K bridge loan to blow away the haze and find the business in this combination of market intuition and technology.
We did.
And then we screwed it up.
Thinking that the real innovation here was software, we designed a software business. Sell a product and make money. A time-honored tradition. We approached a bunch of VCs whom I know with (1) a team with a record of success (2) real, protectable technological innovation and (3) a short productization plan. No bites for a co-investor in Series A. The market said, Wrong! Software Still Sucks.
We went back to the Business Model Drawing Board to re-consider how we could accomplish this grand(iose) vision of programming the web, but with a more palatable adoption model. In retrospect, I think the first pass of delivering the technology as software was a default option. It was easiest because this is what the team had done before, not because it was what the market wants today. So, of course, we moved to hosting - but hosting what exactly?
We went through several theories of what to deliver, looking at various other companies as role models. We flirted with being a "better this" or a "better that" but constantly came back to the belief we could be so much more than a Web 2.1 company, precisely because the technology was so radically better and different than anything we had seen. Along the way I managed to get confirmation of this point from folks I knew at Salesforce, Microsoft, Google, and Yahoo, as well as the single most-informed thinker on Web 2.0 software architectures I have met - Dion Hinchcliffe.
I was stretched with my partners. We were thrashing. We knew we had something Big, but somehow it was the Business We Dare Not Name. Technology, Platform, Application, Solution -- these are all gobblegook businesspeak that don't define anything concrete.
Then we had a breakthrough. Oddly enough it came at meeting at one of the largest software companies back in May. Rafael and I were both there schmoozing and being schmoozed. A few conversations into the day, a light went on for both of us. All of a sudden the fog lifted and it was clear to both of us how to go to market and the natural business model. It was a kind of a Steve-Jobs-in-Xerox-PARC moment.
That stunning moment of clarity was enough to get my greed glands going. I no longer wanted a co-investor. VCs alternate between fear and greed. Now that the opportunity, differentiation, value proposition, and path to monetization are clear the next steps are about establishing proof and building a team. I outlined the concept to my partners and we agreed fund this post haste. We closed Series A two weeks later. The team is hard at work building out the business. Soon we will begin looking for some senior management, including a world class CEO (hint to reader).
Abgenial isn't really in stealth mode. It is not a state secret, but it is clearly in gestation and explaining it is just a distraction right now. One thing I have learned after 25 years in software is that innovative software defies description. It has to be seen to be understood. Imagine Bricklin and Frankston describing Visicalc before you saw it. Imagine Ray Ozzie describing Lotus Notes before you saw it. So please be patient. Abgenial will emerge from the shadows soon enough, and then you'll see why this Aha! was so hard to find, and so obvious once we did.
Chance favors the prepared mind. Blogging as a form of thinking out loud definitely helped me prepare mine for this one.
Interesting way to set out the guidelines and why not this will be called "web 2.1" to help more people to understand.
Congratulations
Posted by: Jose Bracho | June 20, 2006 at 04:39 PM
we should talk. we're industry analysts that keep on eye on where we're going- here i spoke to the idea of a mashup server.
http://www.redmonk.com/jgovernor/archives/001787.html
any relevance?
Posted by: James Governor | June 21, 2006 at 04:50 AM
Google does search.
Cisco networks networks.
What is Abgenial Systems really up to?
Posted by: Dimitar Vesselinov | June 21, 2006 at 12:36 PM
Integrating web application has been detected a few years ago as an interesting market to enter in. Some basic tools are simply lacking but it is technically feasible.
For instance, Web applications already have a lingua franca, it is called HTML and Javascript... This is really different from integrating application from the Windows era where you had to work at the OS layer thus the plaform.
I kind of see what you are trying to achieve. If it is what I think, it is really a neat project. Where could I get update on this project?
Posted by: Nicolas Toper | June 21, 2006 at 04:46 PM
All, thanks for the comments. I will make sure we have a mailist on the Abgenial site for people to subscribe for info.
Posted by: Peter Rip | June 21, 2006 at 05:20 PM
collaboration , or enterprise web 2.0 as you put it, is not just collaboration, it is part consumption, aggregation, modification, and collaboration...you can really collaborate without a fundamental structure in place at several levels...data layer...presentation layer...business layer....if you are just doing web stuff,,,good luck, you also need a viral network, ie : skype app2app, ms live p2p ect...building your own network is likely not going to leverage anything.
Posted by: jccodez | July 21, 2006 at 08:12 AM