Design Patterns Meet Startup Ideas

Are there design patterns hiding behind successful startup ideas?

If you are a software engineer or architect, you would be quite familiar with design patterns. Design patterns are abstract guidelines or high level description of a solution that can be reused in similar situations. Are there design patterns that are hiding beneath the so-seemingly different surface of the millions of startup ideas out there? Can they be distilled and abstracted for reusability?

My hypothesis that there are quite a few design patterns that can be distilled from successful startup ideas. If we can identify design patterns from startup ideas, it would be possible to leverage them to put together a new startup idea. We can also validate a proposed startup idea, by identifying the design pattern elements contained in it. Given the phenomenal growth of startup eco-system, we do have sufficient data points to analyze and validate this hypothesis.

I started by looking at the bunch of recently funded startups from Y-Combinator and see whether we can spot any common design patterns.

1. Localyze is a “company that supports international employees throughout the whole relocation process with an all-in-one software solution covering everything from visa to social integration.”
2. FloRecruit is an applicant-tracking platform for in-person recruiting events.
3. Mighty is a $20 per month cloud computer streaming service that’s just for Google Chrome.
4. Apero provides machine learning driven payment billing services to physician offices.
5. Puzzl is campaign tracking platform for brands.
6. Marble Technologies provides self payment kiosks for restaurant services.
7. Obie is an one stop platform for commercial real estate asset management.

Can you spot the common pattern beneath these different startup ideas? I term this as the Offloader pattern. What is an Offloader pattern? It is essentially offloading of a non-core work by an entity to a specialized service provider, so that the entity can focus on the core work.

Consider the startup, AperoHealth Jacinda Shelly. They enable the physicians to offload the cumbersome billing task so that physicians can focus their energy on treating more patients better. What about the startup Mighty? This is an browser resource offload for your device so that your device resources can be used for other core work.

Consider Localyze. They are offering an onestop relocation services to corporates. Corporates have to deal with multiple vendors for immigration (legal services), transportation (freight), new location familiarization concierge services. Now they can offload all these non-core work to an one stop, single platform Localyze. Marble Technologies Matthew DiMarcantonio does this for restaurants, enabling restaurants to offload their billing, inventory management, customer feedback tracking.

Offloader pattern is also applicable for tasks such as software development. LangAPI is a startup which enables multi-lingual localization of any mobile app, so that the app developer can focus on the business logic, whereas multi-lingual app localization can be offloaded to a third party vendor.

Resource Monetizer pattern: An entity may be owning resources which are currently unutilized by it. They can be the spare room in your home (Airbnb), your backyard or your houseparty (myscoot.in). Can a platform be built to monetize these non-conventional unutilized resources? Think about cars that remain unsold at an automobile dealer? Can these cars be rented out on daily basis? This is exactly what the startup ‘Carve’ did.

Another pattern that keeps cropping up is the Overhead Eliminator pattern. This is about eliminating space-time overheads through technology. What Swiggy did to food delivery industry in India, MyPetrolPump Nabin Roy is attempting in delivering fuel to your door by eliminating the time/space overheads for large fuel consumers. How many physical visits does it take to get your customized clothing fitted by the tailor? Binks is custom fit tailoring delivered to your door without physical visits. Is getting construction supplies taking lot of time/space? The startup curri delivers construction supplies directly to your customer on demand.

Another pattern that I saw is the PAI pattern. PAI stands for Process Automation for Individuals. Just as Robotic Process Automation tries to automate the repetitive manual tasks in business workflow, the PAI pattern tries to apply this task automation to individuals to help improve their productivity at a small bearable cost. If you are a college student applying for a loan, you have to fill in multiple forms, in different formats, with the same inputs. ScholarMe is single app for multiple college tuition loan applications. Metacode provides better code search for developers, simplifying a task that they perform often. Zapier is the text book example of this pattern.

A pattern that crops up often is the ‘Expert Replacer’ pattern, where expensive human resources needed to perform a highly complex task in a niche domain are replaced/augmented by machine intelligence. This pattern is quite different from standard automation where repetitive tasks done by humans get replaced by machines. WellPrincipled is a startup which replaces highly skilled MBAs with AI. HolyGrail is a full stack autonomous research system driven by AI to find solutions to complex research problems in chemistry focussing currently on battery manufacturing. Think of AI driven new drug discovery in pharma industry. This is a form of Replacer pattern in action.

Another pattern that I saw in some of the startup ideas is the V2P-Aid (Virtual to Physical World Transition Aider) pattern. Virtual entities often need physical presence to communicate better with humans in real world. Matagora is a startup delivering pop-up physical storefronts for online brands. Prenda provides in-home physical micro-schools which can aid virtual education content, with teaching guides.

An important pattern that crops up in many startup ideas is the Skin In The Game pattern. This relates to the payment for a service rendered by the startup idea. Your solution’s pay off should be tied directly to your customer’s outcome. Think of Lambda school or Blair which finances college education through income sharing agreements. This is a classic example formoving away from ‘volume based payment’ (pay money for the number of classes attended) to ‘value based payment’ (pay a percentage of your salary once you get the job).

Another pattern pretty common is the ‘Targetter’ pattern which targets a specific underserved group with services they currently lack. Just check the ideas of the following startups.

1. OutTalent provides job services for engieers from developing countries.
2. Elpha — Female professional Community Network
3. Revel — Senior Citizen Community Network
4. Compound — wealth management services targeting startup employees
5. Zippi — loan services specifically designed for gig workers in Brazil,
6. ShortStory — petite women clothing needs.

A startup idea may consist of a combination of patterns. Among the startup ideas we looked at so far, OutTalent Tilek Mamutov combines ‘Skin In the Game’ for pricing, with Targeter pattern (by offering services to engineers in developing country). There is a little bit of Resource Monetizer pattern lurking there as well since existing developer country employees in Tech Giants are able to monetize their skills by providing coaching to the job aspirants through the OutTalent Platform.

I have been looking through a small set of samples to mine for design patterns in startup ideas. Given the limited set of data points I am looking at, some of the patterns I find, can be erroneous or they may be overfitted for these specific startups. But it is a small start on this important topic.

The moonshot question: Can we build an AI platform that can automatically propose startup ideas based on the design pattern templates and analysing real world scenarios? Automated hypothesis generation has been around for many years, Can we apply these techniques along with startup design patterns to creating/validating startup ideas?

VCs/Startup Incubators, Y Combinator, Michael Seibel: While evaluating startup ideas, do you map them to implicit design patterns? What are the ways you abstract the startup ideas? Also would it be possible to share information on what tools you use to filter & validate startup ideas? Do you use tools which do automatic asssement of a startup idea pitch deck as a first level filter for you (similar to automatic essay grading tools)? And if so, how useful are they?

If you know of any existing work on design patterns for startup ideas, please do point me to it. Share with me, any design patterns in the startup ideas you have come across.