I've always been intrigued by claims of programming for non-programmers--no coding skills required
or perhaps just a little knowledge of simple scripting to bake a relatively sophisticated application. Jeff Nolan, who recently left ERP giant SAP, is chasing that do-it-yourself application building dream as CEO of an early stage startup, Teqlo.
The path to DIY composition of applications for Teqlo is way beyond Excel marcros, and depends on loosely-coupled Web services wired together in a service-oriented framework. It’s the promise of software as services that can be easily assembled and reconfigured in ways to execute useful business processes, from simple map mashups to more complex application scenarios.
Another startup company, Coghead, is taking a forms and database approach to creating DIY applications, more along the lines of salesforce.com. Users won’t get to play with Teqlo for a while, however. It will go into private beta in January, and open up for a public beta in mid-2007, according to Nolan. "It's an early phase and an immature market, so we have headroom," he said.
Applications, dubbed Teqlos, are assembled out of Web services (Teqlets), connecting components together on a “canvas board,” Nolan explained. Teqlets get their smarts from XML wrappers that semantically normalize the Web services, Nolan said, expressing the data inputs and outputs as microformats (specific data, such as longitude and latitude in a map Web service or date information in a calendar), and then applying a routing algorithm that determines the proper sequencing of the services in order to properly run an application.
“The technology goes back a few years. It was developed by company in UK. They came up with a fundamental break in approaching the problem," Nolan told me. "Rather than EAI, the software assumes services can talk to each other via standard protocols and just worries about how the data moves. Automated sequencing in simplest form; you have a service like eBay that takes an action and outputs something, such as item and location. Another service like Google Maps waits for something to happen in order to react to it, such waiting for an eBay component to select something that it can then plot with a map marker. Teqlo knows the order of actions by inference and treats it like a routing problem. Google Maps needs to have the location data in a specific format. We know from eBay’s microformats that there is a piece of data called ‘location,’ so the Teqlo engine looks at the two services and infers what the sequence should be, the one logical way to do it.”
Teqlo's prototype formula, or interaction, editor allows users to add email-like rules to the components with a point-and-click interface
Click here for more screen shots
Nolan gave other examples of how the Teqlo service could work, such as taking sales content and merging it into a document and processing outbound messages for a marketing campaign follow up. A more complex assemblage might take customer data from Salesforce, place it in a calendar, create a hard goods delivery schedule, send out emails or SMS messages to drivers and provide a Google Map for truck routing--without writing any code or minimal scripting.
"We have patents on the automated sequencing, but the real innovation goes across the continuum of usability," Nolan said. "A complete novice or system integrator can both get something from Teqlo without having to give up something to satisfy the other." He considers the audience for Teqlo's technology as 90 percent consumers of applications; 9 percent application assemblers, such as end users, system integrators, domain experts or vertical consultants; and 1 percent building the core back end services, exposing their Web services.
Teqlo is not targeting large enterprises or software vendors. “It’s too hard to sell into global 5,000. Selling to IT people not where our solution will thrive,” Nolan said. Teqlo is focusing on companies or divisions with 100 to 250 employees.
Nolan compared Teqlo’s market development approach to an iTunes music store. “We want to give business users the kind of feeling of an iTunes store where we merchandize the applications they might find useful," Nolan said. He also wants to replicate the iTunes model with a standard pricing model, consolidated billing and a recommendation engine that encourages consumption of Teqlets and gets Teqlo in the transaction flow. Consolidated billing is essential given that a composite applications running through Teqlo’s engine might require payments to several different suppliers.
Teqlo’s business model is still embryonic. "When it comes to revenue, that's step 63 and we are on step 8," Nolan told me. Relatively commodity services, such as map mashups could be free, but Teqlo could pre-package sets of services that work with various application platforms and take a referral fee. Nolan didn't rule out advertising, but said he was skeptical. "We have to do something more meaningful than Google AdSense," he said. "We would have to have our own ad network. It's more complex than a lot of companies would have us believe in terms of empowering individuals and selling to users rather than IT."
In the beginning Teqlo will pre-assemble many composite applications, Nolan said, focusing on the biggest pain points for the target audience, mostly small and medium business without significant IT support services. He cited verticals such as contracting, local retail operations and small business financial services as fertile ground for its pre-assembled Web applications.
Nolan, who is also a blogger and a fellow member of the Enterprise Irregulars, is taking a fairly transparent approach to developing his company and product. At this point the Teqlo product is what he termed a “rough alpha” and in the proof of the concept phase. “Right now we want to prove that it can work and learn from it, such as the provisioning requirements and bandwidth per user,” Nolan said. “Overtime the challenge will be how to peel back and enable more complex functionality.”
“We have a couple of huge question marks. How well will it handle complex applications? At what point does the curve of user programming get too steep? We don’t know where the inflection point is and we have no way to know until get people using it. We may find that it doesn’t get as much mileage as we originally thought. There are also issues, such as if a service is not responding. Do you find a substitute service on the fly? We haven’t figured that out yet. Also, when we have hundreds or thousands of components available, we need to structure and organize them in logical ways. We can have tags and surfaces the most popular services and enable search, like Apple's Spotlight within our repository." In addition, the company is working on ways to run composite applcations assembled in Teqlo outside of its namespace.
As Nolan suggests, the boundaries of end-user, DIY application composition and delivering sophisticated Web applications are not very clear. But the potential exists to for users to become productive application assemblers if enough of the heavy lifting, or "muck," as Amazon's Jeff Bezos says, can be eliminated. Teqlo and others exploring this territory have their work cut out for them.