Birth of a New Product-2 Refining Ideas by Jack Krupansky In my previous article I focused on filtering and evaluating new product ideas. These articles are not so much about the product ideas themselves, but an attempt to capture the feeling of turning raw ideas into products. Many success stories give the impression that the entrepreneurs went directly from a simple sketch on a cocktail napkin to a working model with almost no effort. Two months after my previous article I'm on to the next phase, taking a few selected product ideas and refining them a bit. I'm still not to the point of having the single, best product idea which will be passionately pursued, but at least I feel a little closer. There are three product ideas that have survived to this point: an online catalog for entrepreneurial software developers, middleware that allows applications to execute code scripts on a database server, and software agents. But before I go on, I wanted to describe a little of what was going on before I put the first ideas down on paper. I had decided to finish up with my current consulting client that I had worked with for four and a half years and move on the something "new". Unfortunately, I had no new client lined up. I made a few calls and sent out a bunch of resumes, but didn't come up with anything right away. I also was feeling a bit depressed about the state of the software industry and decided that I really didn't want to continue much longer as a "mere" consultant. I really wanted to do a new, really cool product. After thinking about my own experiences and those of others, I decided that I really didn't want to have a small "hobby" product that could not provide me with full-time income and that the only way for small-scale software entrepreneurs to "make it big" is to do something "insanely cool" that literally sells itself rather than requiring a large scale marketing effort. The goal would be to develop a chunk of technology that would be licensed to one or more of the "big guys" or maybe to hope to be bought out by one of "them." I am still focusing most of my attention on paying my bills by consulting. In other words, I'm a true Midnight Engineer since I have to do all the neat, cool stuff during odd hours. I gave four months notice to my client, so now I have to spend a lot of time looking for new work in addition to doing the current work. As important as all that is, that doesn't affect my need and desire to have my own product-based business for the long-term. I also decided to give up my apartment in Paris, since I had no idea where I'd find new work and I didn't want the fact that I was living overseas to in any way interfere with finding work. By the time I come back from Paris at the end of July, I'll have been over there for eleven months; not quite a year, but long enough to have gotten most of the benefit. By the time I write my next article I hope to have my "day job" (still an independent consultant) lined up and make some real progress with the new product. My financial pressure only increased when the hard disk on my notebook PC died in Paris. The computer was actually on loan from my client, so I just returned it to them. I was going to do that anyway soon in preparation for finishing my work with them. Unfortunately it happened just three days after I got to Paris in April, so I had almost three weeks before I was to return to Boston. I decided to rent time on a PC at a service bureau a couple of hours a day. Normally I would have had my system backup on a CD-ROM, but my client had some kind of software problem with their CD-ROM writer when I tried to write the backup a couple of hours before I had to go catch my plane to Paris. I did have my previous backup which was two weeks old, but I had done a lot of work in those two weeks. I loaded CompuServe WinCIM from my old backup and had my client email the most important files which I had changed in those two weeks. They also figured out how to get around the software problem and wrote my new backup CD-ROM and sent it by Federal Express, but that took another two days. The service bureau had a 90 MHz Pentium box running Windows 95. But it was the French version! I never did learn much French, but I was able to figure out most of the menus and messages. The guy who ran the service bureau spoke English and was very helpful. The day before my disk died my Paris CompuServe access number stopped answering. I got a different number from another American staying in Paris and was doing okay with it for two days before it stopped working also! I gave up for the day and the next day bit the bullet and called overseas to the Boston CompuServe access number. I did that for two days and only then thought of checking with CompuServe to see if they might have some other access numbers in Paris. They only had two before. It turned out that the old numbers were no longer on the list and they had three new numbers which worked just fine. That weekend I took the high speed TGV train down to Aix-en-Provence and ran into some Americans at dinner. They had been living in Paris for seven years and had signed up with CompuServe in Paris. CompuServe had sent them email announcing the phone number changes in advance. In fact, a CompuServe rep had actually come out to his apartment and tweaked his modem settings so he could get higher throughput. A high-tech house call? In Paris? Unbelievable. At least I now know enough to check my access numbers once in awhile. Back in Boston, I went shopping for a new notebook computer. At Nobody beats the WIZ I found a Toshiba 120 MHz Pentium with 16MB RAM, 1.25 GB disk, integral 10X CD-ROM, AccuPoint pointing device in the middle of the keyboard, integral universal power supply, Lithium ion battery, 11.3-inch screen, and fairly lightweight. The only real negative is the dual-scan screen, but the color is fairly good and only motion is somewhat annoying. The floppy is external, but you can insert it in place of the CD-ROM. It was the last one, the floor model, so they gave me $100 off and the whole thing cost less than $1,700, including tax. The same model with an active matrix screen would have cost about $2,600. I'm quite happy with it, especially considering the low cost. But, I still have to worry about paying off another $1,700 that I didn't have in my budget. At the time of my presentation at ENTCON'97 back in April, I had pretty much discarded all the other product ideas and made up my mind that I really wanted to focus all of my attention on software agents. But that's not what I'm doing right now. Having made that decision, I proceeded to do some research (mostly with the AltaVista search engine on the web) and decided to reconsider that decision. The one company which had bet heavily on agents was General Magic with their TeleScript language. Started by some ex-Apple software developers, the company has so far been a spectacular failure and is reorganizing to focus on a monthly service that allows mobile professionals to keep in touch with e-mail, voice mail and scheduling via Wildfire-like phone service, personal electronic organizers, and Internet browsers. The bottom line is that even with significant funding, software agent technology is not quite ready for prime time and the potential market is not ready for it either. You might ask how I decided to take an interest in software agents. People have been talking about them for many years, there's been a lot of research, and the term has even appeared in Business Week and Forbes magazine. Basically, I had another product idea and realized that it could be extended to support software agents. That first idea came from a series of email exchanges with a friend who has a client/server database package and wanted to be able to execute a code script on the database server on request from the client application on another machine across the network. We discussed some approaches and I realized that the infrastructure needed to support the execution of scripts could be extended to support software agents. So off I went on agents. Rather than try to drag my friend off on my quest for the software holy grail, I immediately decided that the remote script execution would be a separate project from the new agent project. We thought there might be significant overlap, but I wanted to preserve the option of pursuing a more mundane, but more easily realizable and more easily explained product without interfering with the far-reaching potential of the agent concept. That was a wise decision, since the agent project will be rather stop-and-go for awhile with no short-term payoff. The remote database script project is much more manageable, but with more limited potential. Besides the failure of General Magic, I think the fatal blow to my short-term interest in agents came when I realized that software agents suffer from "NOTES disease." Lotus came out with a really neat communications/database product that essentially defied description and analysts and customers alike have had great difficulty understanding what the product was all about. Most people just used it as a fancy email system without fully realizing its ultimate potential. You need to be able to describe your product in a single paragraph and demonstrate it with just a few screen shots. Even I barely understand what software agents are really all about, so there is currently little hope that typical software purchasers or analysts will fully understand what the concept is really all about, let alone what my specific product would accomplish. To most people, if they think they understand agents at all, a software agent is a mini-program that goes searching across the net for things that interest the user. The potential of agents seemed so huge that I have actually been considering venture capital. But the bottom line with the venture capitalists is that you really need to be able to rapidly ramp up to a high level of sales within a year or two. Agents are so new and so much new technology is needed that even two years seems too short. And it's just as unlikely that most corporations will be able to change course (and paradigm) and come up to speed on agents in two years. They're still struggling to adapt to the Internet/web which are actually quite simple. I couldn't honestly tell a VC that I could achieve $100 million in less than five years and even then only if everything went right. If I were a VC, I wouldn't go near software agents, yet. I also considered whether to approach a Microsoft, IBM, Novell, HP, etc. But I decided that would probably be so much effort and require laying out the complete blueprint for them. Some of them are already doing some form of agents. IBM has a Java- based research tool for agents called Aglets. The bottom line is that I wouldn't feel comfortable negotiating with these companies at this stage and don't even want to clue them into the possible agent strategies that I can envision. I'd rather develop something real and get it out on the web for the "lunatic fringe" to play with and get it "blessed" as being "really cool". Then do a bit of PR and be in the position where the big guys come courting me. I have talked to a number of people that I know to get their feedback on my overall agent concept. With something this new and unproved, it's hard to find many people who can really discuss the concept intelligently in anything more than an academic manner. But, I find that explaining a new concept to someone is one of the best ways to figure out how the concept needs to be tuned to make it palatable to the widest possible audience. Most people thought it was a really good idea. One person with a lot of experience starting new companies was wary of focusing on the low-level infrastructure for agents since that should eventually become part of the operating system and Microsoft and other companies could probably just do their own rather than license the technology. But, that person really thought that agent-based applications could be a big opportunity for capturing the interest of VC. As hard as I tried, I couldn't find a way to immediately start making a lot of money from software agents. Nonetheless, I still believe in their potential and maybe in two more months I'll have cracked that nut. I think that within ten years it will be a pervasive technology. I will continue to evaluate this area, looking for an "angle" that will minimize both technical and marketing risk. One intriguing idea is to identify a "killer app" for agents and bundle the agent infrastructure development into the app project. But I don't yet have a clue what such a killer app might be. I still have hope, but I don't feel comfortable committing to this idea at this time. My friend initially proposed the remote database scripting idea as a way for a client application to request that a script be executed on the database server. But I extended the idea to include support for executing scripts based on trigger events that occur when the database is changed or updated. Some databases support triggers, but in a proprietary manner. That's where middleware comes in. Middleware can act like a switch box to hide differences between two pieces of software. Let the customer pick the database and the scripting language without the choice of one dictating the choice of the other. All through this process of evaluating and refining product ideas, marketing is THE key concern. The technology may be cool and we may be able to build the product, but will we be able to sell enough to make a decent profit. What I really want to do is develop a chunk of technology that can be licensed to a few software venders who then pay us to consult on integrating the technology with their products. My current thought for a marketing strategy is to develop our middleware and package a full-featured demonstration version and make it available for "free" on my web site. The demonstration would use my friend's relational database and my CodeScript scripting language. The middleware wouldn't require either his database or my scripting language, but the demo would show how any database and any language can be hooked up once the appropriate "plug-in" interface modules are developed. Oracle and SQL Server are the obvious candidates for the production database and Java is the obvious candidate for the production scripting language. But since our interest is attracting the attention of the venders, it's not necessary for us to do all the plug-ins in advance. Some of the plug-ins might be developed independently of the database and language venders, but some customization will probably be needed. I think this middleware will also provide a great lead-in for software agents. Although people think of agents as searching for information, I think their ultimate benefit will come when they can "wait" for information to become available (e.g., waiting for a cheap airfare, hotel room availability, a buyer for a good.) And I think that can best be implemented by allowing the trigger portion of agents to hang out at the database and be triggered by the database rather than having the search agents constantly querying the database. It's not quite that simple, but I don't want to give away too many of my ideas for free! The web-based catalog for entrepreneurial software developers continue to limp along. It is operational at http://www.basetechnology.com/esdmall.htm, but it's still not very fancy. I was calling it a "mall", but since you can't actually buy anything, it's really more of a catalog or showroom. There hasn't been a gigantic level of interest, but it is a good tool and allows me to get experience with the web. I'm sure that once I get back from Paris in August I'll be able to get it moving a little better. I'm still offering to add entrepreneurs to the catalog for free. You might also ask what happened to the other ideas I mentioned in the previous article. Well, they just vanished off the radar screen. It wasn't that they were discarded for specific reasons, but that they just didn't seem to be that likely to be wildly successful. It's just like when VC review business plans - it doesn't matter what's wrong with a plan, but just whether the overall plan really gets them excited. My discarded ideas just didn't excite me and I didn't perceive that a wide audience would get too excited by them. I continually agonize over whether I'm spending too much time trying to figure out what to do and that maybe I should just actually do something. But I'm constrained by needing to spend my prime time doing consulting work (and enjoying France!), so it's easy to defer the new projects. I also keep trying to see if I can get one of the ideas in good enough shape to interest the VC. That's not a priority and doesn't consume very much of my time, but it's like an annoying mosquito that keeps buzzing in my ear and won't go away. Maybe the idea is that if I could get VC funding then I wouldn't have to do the time-wasting consulting. But I think the biggest concern holding me up is that I just don't feel that I have the right angle of attack on any of my ideas and that I just need to put in a little more effort to find the sweet spot. What's Next The technical issues of the product ideas need to be elaborated, but the marketing strategy will get just as much attention. Due diligence research needs an equally high priority. Lining up consulting work has absolute top priority, unfortunately. I also intend to spend a little time trying to come up with even better product ideas to test whether the current ideas are really the ones I should pursue. I really, intensely want to figure out the proper angle of attack to go after software agents, but I know better than to let that desire consume all of my attention. And I definitely will be talking with more people to get more feedback before it's too late. I also feel the need to step back and look at an even bigger picture than I can currently see in front of me. I would like to talk to some high-level corporate managers to get a better feel for what worries them about the technologies and products that they currently have to work with. I don't trust the filtered and refined "facts" that analysts write in their research reports, books, and articles. I also need to try to figure out what ways I've been too patient with the status quo and how I haven't acted confidently enough in proceeding with my best judgment. The bottom line is that in two months I should have made a lot of progress and have something more to show for that time than just some refined ideas. We'll see... Jack Krupansky runs a one person software business, Base Technology, which develops and markets the Liana object-oriented programming language and CodeScript scripting language interpreter and offers Windows software development consulting. He may be reached at 800-786-9505 (messages), jack@basetechnology.com on the Internet, or at http://www.basetechnology.com on the World Wide Web.