Interview With Habib Haddad, Co-Founder Of Language Analytics LLC. / Yamli

Habib Haddad & Imad Jureidini

Continuing our series of interviews with Arab entrepreneurs, we bring you our interview with Habib Haddad, co-founder of Language Analytics LLC., the company that brought us Yamli, the cool web-based transliteration tool that solves the problem of many Arab users who don’t have an Arabic keyboard or who aren’t as comfortable typing in Arabic.

I’d really like to start by thanking Habib for taking some of his time to answer our questions, generously providing details and insight about Yamli, how everything fell into place, where they’re taking it, as well as sharing some tips and advice for other entrepreneurs.

How did you get the idea to build a tool like Yamli?

The idea came from my personal frustration in interacting with the Arabic language on the Web. It seemed to me unreal that I, an Arabic native speaker, had difficulty using my own language on the web. Living in Boston, access to an Arabic keyboard is not very easy especially when you are on your desk at work and want to search for news in Arabic. Even when I lived in Lebanon my interaction with the Arabic keyboard was limited, which is sad but it is the case with a large number of Arabic internet users. In fact, studies at the American University in Cairo shows that 78% of Arabic internet users have never typed in Arabic! Imagine if 78% of French never typed in French. Imagine how destructive that would be for the language on the web and how limiting it would be for local businesses, entrepreneurs and even publishers.

Could you tell us more about the steps you went through to make Yamli a reality?

Form the team: I was happy to have Imad Jureidini, a colleague from a previous startup to join me as a co-founder. My advice here is to pick someone who complements you, excel at what they do, who you enjoy hanging out with, and look for someone you think is smarter than you.

Identify the vision and goals: This is the thing that says that if we don’t get anything else right, this is what we’re going to do well and really excel at it.

This to me is the most important step and is broken down into multiple steps:

• Listen to your users’ problems: We spoke to potential real users (friends and family) to get a feeling on how they view the ideal solution for this problem. It’s amazing how unexpectedly unhelpful this was. Basically users don’t know the best solution to their problems, but they know what problems they have. Instead of asking what users want, try listening to their problems.
• Know your REAL competition: Transliteration in general, and even in the case of Arabic, is not a new concept but we felt it had not been solved the right way. There were a few tools out there that solved this problem by requiring the user to learn a one to one mapping table. However our real competition was the English keyboard. We wanted to come up with a solution appealing enough to convince the user, who is not accustomed to typing in Arabic, that now he can actually do it, that he can type real Arabic words using Yamli with no extra effort.
• Our vision:
  • Accurate solution that does not require the effort of learning from the users and that “magically” finds the right word.
  • Seamless solution, drop dead simple to start typing and not even feel an extra layer.
  • Available to all users (API)

Prototyping: Once the vision was defined we moved to execution. We ran into many technical problems both on the algorithm and the UI side but at no time did we accept to compromise on the vision that we defined. On the algorithm side the main challenge was to get it to understand the dialectal variations. Of course that meant we had to be aware of these. With limited resources we had to be creative; so for example I had to listen to YouTube movies on “درب الزلق” to train myself to the Khaleeji dialect 🙂

On the UI side, we did lots and lots of prototyping and focused a lot on getting the interactions and usability right. I remember our first UI prototype was drawn on a piece of napkin at a Pizza place near Boston. Before beta testing, we had friends come and use the tool (in exchange for a free can of Coke … and lots of coffee). We carefully identified their usage patterns and kept on iterating until we felt we got it right, and by right I mean easy enough for my mom to use without having to ask for my help (Seriously). One important thing to keep in mind here is that early users/testers might not reflect the target users.

Work on the production code: Scale, speed, reliability come into play here. Also this is the stage where we focused on the look and feel of the site, link placements, and text.

Private Beta Testing: We got friends and family to sign up. We sent out an email with instructions and waited for the feedback. In retrospect we made a mistake by signing up too many beta testers at once. Ideally you would want to divide them in small groups so that you have the time to act on the feedback from the first group, then have the second group test etc … We got a lot of good feedback and some bug reports. Major changes in the architecture are always tough to do, but are much better done at early stages. We decided to completely rewrite a module. Doing it at later stages would been much harder to control and would have caused a mess.

Turned on the switch and stayed awake for the next 36 hours!

What do you want Yamli to achieve and where do you see it a few years from now?

We aim to empower the Arabic web, tool by tool, innovation by innovation. Seems like a pretty general goal, right? You are actually correct to think so, but we are going to attempt to solve this problem by getting down and dirty, identifying every little annoying unsolved problem that affects a lot of users, and keep on banging our heads until we nail it down and solve it the right way. We initially started by solving the problem of typing and we are now moving into other innovations, mainly related to search so expect some cool innovations there pretty soon 🙂

How many people make up the Yamli team?

Currently the team is made up of the two co-founders. We are starting to look around for hiring. We don’t believe in outsourcing, and believe more in building a strong team that is passionate about the company and its mission, and that also has interest in seeing the company achieve it. Feel free to send us an email through our site if you would like to join our team.

How have you financed Yamli up to now?

Up until recently we have been bootstrapping the company although we had been pinged by a few angel investors and VCs. We recently received funding from two prolific angel investors: Georges Harik and Aydin Senkut. Both of them are early and ex-Googlers and were behind initiatives like Gmail, AdSense, Google Mobile, Google’s Asian syndication business … We are pretty excited and flattered to have this caliber of investors on board.

You recently launched small banner ads in the Yamli tool, how successful would you say they’ve been so far?

The ads are not yet open for commercial use. We are currently using them to promote our API partners, and also to support initiatives and blogs in the Arabic web space.

Could you give us a little idea about the underlying technology powering Yamli’s typing technology?

There are two major components to Yamli’s typing technology:

Server side: The transliteration algorithm runs on the server. The patent pending algorithm, which we like to call “our secret sauce”, takes a particular Latin word, breaks it down and analyzes its phonetic spelling, applies Arabic grammatical rules along with phonetic rules and comes out with a set of results ranked by likelihood. The algorithm also improves with time and adapts to dialects, new typing patterns, new words. So you can think of it like a small baby when it was launched, which was then fed and taught by users over time. Now it’s more like a smarter, fatter young kid. Well, maybe just smarter.

We also spend a lot of time thinking about scalability. We strive to stay well ahead of our current needs, and plan our server infrastructure in a way that insures that we can handle very large increases in traffic.

Client Side: Yamli uses the latest AJAX techniques to perform its magic in real-time. While in principle the Yamli user interface is simple, it is actually quite a challenge to be able to integrate functionality like ours in a way that is compatible with all the web sites out there, as well as the variety of web browsers on the market. In fact we have had to come up with new and unique solutions to browser issues that are specific to Arabic and other right-to-left languages. We probably spend half of our development time working on the Javascript client code. We also spend a lot of time testing the code with each release, and in a sense we are in a permanent QA mode. This is vital for code that is distributed through an API and which third parties depend on.

Why did you decide to launch an API and how useful has it been to you?

Our mission statement is to “Empower the Arabic web”, all the Arabic web, not just Yamli.com but all of it. We believe in collaboration and in providing your technology for others to freely use. Now you might say this is utopist but it helped us as a company in many direct and indirect ways. It helped us get the word out about Yamli and it helped us build strong relationships with some of the API users.

Is there an official Yamli WordPress Plugin in the plans?

Yes, and other plugins as well. Actually we plan on releasing tools specifically tailored to bloggers. Bloggers are what makes the internet a more real and democratic place and what would be better than bloggers expressing their ideas and views in their own mother tongue without losing the crucial nuances that get lost in translation? And what would be better than have their readers comment and interact with them in the same fashion? We are constantly thinking of tools to encourage blogging in Arabic. We of course want to hear your comments and suggestions and would love to have you be part of this effort, so for all the bloggers out there we want to hear about all the problems you face in blogging (info AT yamli DOT com) in Arabic and think together about the best solutions for them.

What other main features are you looking forward to adding to Yamli?

As I mentioned earlier, we are gradually releasing tools that will help us achieve our goal. Our next innovation will be related to Search. Search is a very important, if not the most important, aspect of the user’s web activities. Search drives the internet, but it has been very weak in the Arab world and the low number of queries especially in Arabic confirms that. The main problem behind that is that the Arabic content on the web is very low, with less than 0.5% of the worldwide content. This has created a vicious circle where the lack of content discourages people to search, which in turn discourages content creators to publish in Arabic. We have identified a set of problems we are working on to solve these issues. We hope that by doing so we will break the vicious circle and help people search, and help businesses, publishers and bloggers create Arabic content and have confidence that it will be reached. We also have other cool ideas that we will be working on very soon.

Are you seeing any meaningful usage in countries where keyboards are Arabic like Saudi Arabia?

A large portion of our users come from Saudi Arabia. In fact about 75% of our users come from Arab countries like Saudi Arabia, Egypt, Jordan, Morocco, and Lebanon. The Arabic internet population is mainly dominated by a younger generation of users, and those users are usually more prone to using the English keyboard unfortunately. What we provide is a quick, easy and “cool” way of typing real Arabic words instead of English or Arabizi. In a sense we provide a disruptive technology that will hopefully overturn the existing dominant method used to type Arabic (Arabizi) and replace the web content with real Arabic words.

Our typing technology is not targeted at those who know, or wish to learn, how to type Arabic. In fact we will soon be releasing tools that target the whole spectrum of Arabic users, but I can’t say more at this moment 🙂

Are you making money?

Not really 🙂 as we are not currently focused on revenue , but more on releasing products that tackle fundamental problems for Arabic internet users. From a business perspective we see a lot of opportunities in the Arabic web, and we are really excited about the fast pace at which it is growing. However we believe our time and energy at this moment is best spent on coming up with useful and innovative products regardless of whether we can monetize them or not. That said, we believe we will be well positioned to extract the business opportunities that will arise.

What lessons did you learn from launching Yamli that you’d like to share with other entrepreneurs?

• Be ready for criticism.
• Talk to as many other entrepreneurs as you can.
• Pick people who complement you, excel at what they do and who you enjoy hanging out with.
• Be general in drawing the vision for the company, but focus on the tools and the products that will help you achieve it.
• Spend time identifying the few things that you absolutely have to get right and relentlessly refine them. Those few things usually make a big difference.
• Do not underestimate details. Don’t be afraid to spend time working on the little details that affect a large number of users.
• Products should be drop dead easy to use. “Simple things should be simple, complex things should be possible.” (Alan Kay)
• Don’t rush into taking funding. Pick the right investors for your company.
• Use your own products every chance you can. Be your Number 1 customer.
• Listen to your users’ problems.
• Just do it!