Part 2/2: AI UX Patterns Product Managers Should Know

by Shadab Farooqui


For years, we’ve evolved through familiar digital transitions— from CLI to GUIs, GUIs to touchscreens. Now, we’re entering a space where interfaces are increasingly adaptive, context-aware, and powered by AI. This isn’t just about layering AI onto existing interfaces; it’s about building around it. Here’s a look at some new patterns that product managers should consider as Gen AI reshapes digital experiences:

  1. Chat

Lightweight Chat: Chat can be useful for quick access to information or summaries without disrupting the workflow. For example, in Cursor’s AI IDE, developers can ask questions about the codebase directly within their coding environment, accessing answers immediately.

Integrated Chat: Beyond simple chat windows, integrated chat features blend with other core actions. Klaviyo’s segment builder, for instance, lets users define customer segments with natural language, skipping the need for complex logic builders.

2. In-Line Editing

Diff-Based Editing in Code: Tools like Cursor allow users to highlight sections of code and receive targeted changes, with the option to accept or reject each suggestion. This approach could extend to other fields where selective, detailed editing is important.

In-Painting for Visual Content: Platforms like Midjourney let users modify parts of an image selectively, which could be valuable in content creation and media editing. Users could adjust specific image elements without needing to recreate the entire design.

3. Visual Workflows

Node-based editors allow teams to map out workflows visually. Whether it’s creating multi-step onboarding flows or user engagement funnels, node editors make complex paths easier to manage.

APIs and Conditional Prompts in No-Code Platforms: Tools like Vellum.ai integrate API calls and conditional prompts into a single interface, allowing teams to customize workflows without backend development. This approach supports fast iteration and experimentation.

4. UI Primitives

For those who want to go further than basic chat, new UI primitives are adding more flexibility and personalization to Gen AI interfaces:

Select + Edit: Users can modify structured data by highlighting sections and applying prompts. Whether it’s a care plan in healthcare or financial data in SaaS, this approach allows users to make direct changes, with AI generating structured responses that are ready to use.

Inline Suggestions: Inline suggestions act like search suggestions or “Did you mean?” prompts. They can guide users toward better inputs or additional options without requiring extra effort. In a health app, this might mean prompting users to “Add nutrition goals” while they draft a wellness plan.

Personal Context: Gen AI becomes more powerful when it adapts to user-specific data. By integrating personal metrics—like health data from wearable devices—apps can create a more relevant, personalized experience. This makes the app feel smarter and more attuned to individual user needs.

Final Thoughts: Experiment and Iterate

Industry voices are already noticing these shifts. We are transitioning towards interfaces that are inherently AI-native. AI is used to create tiny one-click functionalities that are scoped to a particular task, reducing the mental load for users.

For product managers, the takeaway is clear: don’t just add AI, rethink around it. Each of these patterns—whether diff-based editing, node-based workflows, or in-painting—presents an opportunity to make user interactions simpler and smarter. The formula isn’t set in stone, which is exactly why it’s time to experiment.

We’re at the start of a shift in how users interact with AI-powered systems. The next step is to test these patterns, refine them, and see how they resonate with users. Start small, keep iterating, and build from there.

Get in touch:

Have ideas or experiences with AI-driven UI/UX? Reach out if you’re exploring Gen AI in your product and want to discuss ideas.


Part 1/2: AI Architecture Patterns all PMs should know

by Shadab Farooqui


I often find myself wishing there was a comprehensive catalog of design patterns for the AI applications we see in the real world. Sure, OpenAI offers its Cookbook and Recipes, but outside of that, the most practical way we usually learn about real-world implementations is by diving into open-source projects.

To simplify things, I’ve put together a list of common architecture patterns based on how AI is being integrated into production today. I know that “AI” can feel daunting, almost infinite in scope — and given the countless permutations LLMs can produce, it’s easy to see why. This list aims to cut through the noise and provide a clear, ranked overview of how companies are actually building with AI right now:

Ranked List of AI Application Architecture Patterns:

1. API-First Architecture

Example: Zakk.ai

Pattern: Uses external API (e.g., OpenAI) and vector databases (e.g., Pinecone).

Pros: Fast setup, scalable.

Cons: Limited customization, API dependency.

2. Fine-Tuned Model Deployment with API Gateway

Example: Enterprise Chatbots (e.g., ServiceNow Virtual Agent)

Pattern: Fine-tuning + API Gateway for custom, scalable solutions.

Pros: Domain-specific responses, scalable.

Cons: Needs substantial data for fine-tuning.

3. Microservices with Multi-Model Orchestration

Example: Netflix’s Recommendation Engine

Pattern: Independent services for different tasks, integrated by an orchestrator.

Pros: Modularity, flexibility.

Cons: Complex orchestration.

4. Retrieval-Augmented Generation (RAG)

Example: Bing Chat, Knowledge-based Q&A Systems

Pattern: Combines retrieval from a document store with generative models.

Pros: Grounded responses, factual accuracy.

Cons: Requires a well-maintained document store.

5. Serverless / FaaS (Function as a Service)

Example: Real-time Slackbots

Pattern: Serverless functions (AWS Lambda, GCP Cloud Functions) for inference.

Pros: Cost-effective, easy to scale.

Cons: Cold start latency, limited execution time.

6. Vector Database Integration

Example: Semantic Search Engines

Pattern: Embeddings stored in vector databases for fast similarity search.

Pros: High-speed search, scalable.

Cons: Embedding generation overhead.

7. Embedded AI / Edge Deployment

Example: Apple’s Siri (on-device inference)

Pattern: Models run on-device for low latency.

Pros: Offline capability, enhanced privacy.

Cons: Limited model size.

8. Hybrid Local-API Inference (On-Device + Cloud)

Example: Google Assistant

Pattern: Lightweight model on-device, API call for complex tasks.

Pros: Low latency for common tasks.

Cons: Sync issues between edge and cloud models.

9. Multimodal Fusion Architecture

Example: Pinterest Lens (text + image input)

Pattern: Combines outputs from models handling different input types.

Pros: Supports diverse user inputs, richer context.

Cons: High computational complexity.

10. Monolithic Application with In-House Model Hosting

Example: Traditional ERP Systems

Pattern: In-house hosting of AI models within the main app stack.

Pros: Full control, direct integration.

Cons: Difficult to scale.

11. Continuous Learning System

Example: Self-Driving Car Systems

Pattern: Live data ingestion, model retraining loop.

Pros: Adaptive, improves over time.

Cons: Resource-intensive, risk of instability.

Happy Building!


“Always be automating”

by Shadab Farooqui


I love automation because it creates capacity to do higher value things, and helps create optionality. My company Good Vibes Only AI (https://gvo.dev) is the realization of that passion. It was founded with the principal of making product experiences that enable humans and organizations be at their 100x best.

Automation has always been a defining force in the evolution of human workflows. Changing how we work, live, and even think about the concept of labor itself. From simple tools to sophisticated machines, humans have always been automating. However to truly learn about the progression of automation, I thought of getting a history lesson. Let’s dive in!

Yet, while automation became essential to industrial and business processes, the systems were limited to highly structured, rule-based environments.

The Evolution of Automation: From Mechanization to Intelligence

  1. The Age of Mechanization (1760–1850)

    The Industrial Revolution marked the first era of automation, where mechanical inventions replaced manual tasks in agriculture, textile manufacturing, and metal production. Innovations like the steam engine and the mechanized loom revolutionized productivity, enabling workers to output more goods with less effort. Though the systems were rudimentary, the framework for modern automation was laid here: machines doing repetitive, labor-intensive tasks. The next big innovation — electricity came after ~100 years.

  2. The Battery Revolution (1800–Present)

    The invention of the battery in 1800 marked a turning point, enabling portable, on-demand energy. Early batteries, like the lead-acid cells developed in the 1850s, paved the way for mass electrification and transportation, laying the groundwork for modern automation. The next great leap came nearly a century later with lithium-ion batteries in the 1990s, which revolutionized portable electronics by offering compact, rechargeable power. Today, advancements in solid-state and sodium-ion batteries promise safer, more efficient energy solutions, especially for electric vehicles. New fast-charging and recycling innovations further signal a shift towards a sustainable, battery-powered future, as society moves from static power to a dynamic, distributed energy landscape.

  3. The Age of Electrification (1870–1920)

    With electricity, automation moved beyond basic mechanization into a more flexible, distributed form of work. Electrification enabled the widespread adoption of assembly lines, and industries like automotive production became entirely reorganized. Automation allowed for unprecedented production volume and efficiency, with some of the earliest examples of programmed, repetitive tasks within a controlled environment. Yet, while machines began to replace more complex human labor, they still required constant human oversight and intervention. The next big innovation — computers came after ~50 years.

  4. The Age of Computation (1950–2000)

    The invention of computers marked a quantum leap in automation. Programmable machines and robotic arms were now capable of performing precise tasks with a degree of adaptability. Computerized control systems in factories became ubiquitous, and software-driven automation found its way into offices with data management systems. Companies like IBM led the charge in digitizing and automating back-office functions, from payroll to data processing. Yet, while automation became essential to industrial and business processes, the systems were limited to highly structured, rule-based environments. Creativity and context-dependent decision-making remained out of reach. The next big innovation — big deep learning and intelligence came after ~20 years.

  5. The Age of Intelligence (2000–2020)

    With the rise of machine learning and, more recently, deep learning, automation gained the ability to analyze unstructured data, make predictions, and improve from past actions. This was a pivotal era when tools like Google’s predictive algorithms, Amazon’s supply chain automation, and the first chatbots demonstrated the potential of intelligent automation. These systems learned patterns but lacked a true understanding of context or intent—traits essential for handling tasks that require nuanced human-like reasoning. The next big innovation — agentive AIs came after just ~4 years!

  6. The Age of Agentive AI (2020–Present)

    Today, automation has entered a stage characterized by agentive intelligence—systems that can perform complex, context-sensitive tasks without constant human intervention. This generation of AI doesn’t just follow programmed instructions; it perceives, interprets, and takes autonomous action within defined parameters, adapting on the fly.

  7. The Age of Intelligent Robots

    Robots with human like movements integrated with vision and sound AIs are evolving — aiming to replicate human tasks autonomously in the physical world. Companies like Tesla (Optimus) and Boston Dynamics (Atlas) lead innovations, enabling robots to adapt in real-time and handle complex environments. These developments parallel digital AI, edging closer to human-like robotic proxies in daily life.

    Let’s look at the state of automation systems today that go beyond traditional automation, introducing sophisticated tools capable of handling complex, nuanced tasks.

    Prominent AI Agent platforms and their approach:

  1. Microsoft: Through products like Microsoft Copilot and Dynamics 365, Microsoft has positioned itself as a go-to enterprise solution for integrating AI into everyday business tasks. With its OpenAI partnership, Microsoft offers users tools embedded directly in software like Microsoft 365, allowing employees to automate document generation, data analysis, and workflow management across productivity tools. The tech giant will release ten new autonomous agents designed to augment sales, service, finance, and supply chain teams.This seamless integration into widely used platforms is helping businesses streamline complex processes, making Microsoft a major player in AI-driven enterprise solutions.

  2. Anthropic and OpenAI: Both Anthropic’s Claude and OpenAI’s GPT models power a wide range of intelligent agents that automate knowledge-based tasks, from customer service responses to content generation and complex planning functions. OpenAI, with its GPT-based ChatGPT, has achieved mainstream adoption in customer support and business operations, enabling tasks like summarizing meetings, drafting responses, and interpreting vast data contexts. Anthropic’s Claude models emphasize explainability and user safety, making them popular in sensitive domains like healthcare and legal advisory.

  3. Skyvern: Skyvern employs Large Language Models (LLMs) and computer vision to automate tasks on complex, visually dense web interfaces. Skyvern’s approach uses contextual prompts and computer vision to interact with web elements dynamically, enabling it to adapt to website layout changes without relying on brittle rules like XPath. This makes it highly effective for web scraping, data collection, and e-commerce management across diverse platforms.

  4. Imbue: This company are tackling automation with an innovative approach known as “Mixture of Experts” (MoE) architecture. By segmenting tasks into sub-tasks handled by specialized agents, their frameworks can perform complex workflows with human-like efficiency. This allows for modular automation where tasks can be optimized and scaled according to needs, making MoE architectures suitable for dynamic, high-complexity environments, such as software development and project management.

We are in the era of agentive automation where AI not only follows commands but also reasons, adapts, and acts with contextual awareness. This shift brings a level of flexibility, scalability, and productivity that is transforming industries across sectors, from customer service to legal compliance, and advertising — making advanced automation a vital component of the modern enterprise ecosystem.

The Future of Agentive Automation: What Lies Ahead

  1. Increased Autonomy and Collaboration: Future systems will have even greater autonomy, capable of interacting with other systems and humans in ways that feel increasingly seamless. For example, we might see AI-driven project managers that can autonomously schedule and assign tasks based on resource availability, deadlines, and priority changes, all while coordinating with human team members for approval.

  2. Contextual Intelligence and Real-Time Adaptation: As systems improve, they will likely gain better contextual intelligence, understanding subtleties like intent and mood, even in ambiguous scenarios. For example, a customer service AI might adapt its response tone based on user frustration levels, offering escalated solutions or human intervention only when necessary.

  3. Interoperability and Hybrid Human-AI Workflows: Future agentive systems will be increasingly interoperable, allowing hybrid workflows where human intuition and AI efficiency blend seamlessly. A doctor, for instance, might rely on an AI assistant to analyze medical histories and suggest treatment options, all while retaining the final decision-making power. This hybrid model will enhance human work rather than replace it.

There’s also a segment of vertical models, which is the topic for a different post. Build on!

The rise of these companies signals a new era in agentive automation where AI not only follows commands but also reasons, adapts, and acts with contextual awareness.

Thought Exercise: Winning The Home With Opendoor

by Shadab Farooqui


Opendoor is a real-estate company that I have grown to admire because they are challenging the status quo of home buying and selling. Anyone can buy or sell their home using the platform. Following is a thought exercise for a new Opendoor product to help buyers make an offer that sellers will find attractive.

Here’s a presentation to go with the writeup.

**Note: I don’t work at Opendoor. I just spent a few hours researching and writing this up**

Mission: Help buyers make an offer that sellers won’t refuse.

Goal: Design an MVP product to help buyers make attractive bids to sellers, especially in a competitive market.

North Star Metric: Bid Conversion Rate: % of offers accepted (# accepted bids / # bids).

Secondary Metric: Contract Conversion Rate: % of offers accepted that closed ( # contracts closed / # accepted bids).

My Approach
I applied first principles by talking to an Opendoor home buying advisor and loan officer, an independent real estate agent, and an in-market buyer. I then analyzed the current Opendoor buyer flow (iOS app), while researching real estate market trends, to investigate the home buyer and seller problem space. Finally, I hypothesized potential solutions for building a new product that could have an impact on Opendoor’s bid conversion rate.

What is Opendoor?
Opendoor is a web and app driven platform for home sellers and buyers. Sellers in Opendoor markets can get an instant offer (if the home meets the Opendoor ​criteria​) from Opendoor. Think of it like CarMax for houses, except you don’t have to leave your house. Buyers can make offers on an Opendoor-owned listing, or any listing that is not on Opendoor but is in an Opendoor ​market​. The Opendoor app helps you seamlessly navigate the buyer transaction, making the home buyer’s journey more predictable by providing help with:

  • Financing​: Work with an Opendoor partner institution or use Opendoor underwriting (conventional and FHA loans) that comes with perks such as a 1.5% discount on the sale price, $1000 closing credit, and savings from zero processing and underwriting fees.

  • Search​: Start your search in the Opendoor mobile or web application that allows virtual or self-driven in-person visits. Opendoor advisors can unlock some homes remotely for convenience. A buyer can also make an offer within the app should they choose to.

  • Agent​: Buyers can work with their own agent or choose an Opendoor agent, who could be an independent partner agent or employed by Opendoor.

  • Education​: While educating the buyer on the process, it also enables buyers to communicate with dedicated Opendoor advisors available to answer questions via in-app chat or a phone call, making the process as low-touch as preferable by the buyer.

Core Hypothesis
Home buyers and sellers would prefer a more seamless and touchless end-to-end experience driven by a real estate platform and partner they trust, without having to navigate across multiple platforms, vendors, and other real estate professionals.

Buyers will represent themselves as agents, while expert real-time support would become a value-add for education and peace of mind for the unsophisticated buyer. Gen Y, Millennials, and Gen X represent 61% of home buyers according to the 2020 National Association of Realtors ​report​. These generations are tech savvy and will change the status quo of multi-touch processes and gatekeepers.

Competitive Markets
Opendoor operates in select ​markets​ (21 cities in 7 states). For this product, we will analyze markets with the lowest bid conversion and offer conversion rates in addition to the volume of closings to gauge potential impact.

Buyer Segments
Home buyers can be segmented in various ways to identify their needs.

  • First-time or repeat buyers​: A repeat buyer may have set preferences about the agent or lending partner and is more educated about the process than an unsophisticated buyer.

  • Property Use​: Primary home, second/vacation home or investment.

  • Listing Type:​ Opendoor property or an independent listing that is in an Opendoor market.

  • Funding​: Cash or financed offer.

  • Agent​: Self-serviced (buyer is the agent), looking for an agent, has an agent.

    However, to better understand what forces allow one buyer group to have a winning bid versus another, we will dive deeper to consider the potential impact of the offer provided and ways in which it could have been enhanced. We will also identify opportunities for increasing the offer submission rate by examining buyers who started an offer process but did not proceed to submit their offers.

Buyer Needs

  • Bid​: Buyers in competitive markets lack visibility and clarity on offer dynamics. That need is currently facilitated and mediated by agents who are equipped with MLS data and provide recommendations for competitive bid packages.

  • Experience​: Buyers of all levels of experience want the upsides of controlling the process but don’t want to deal with its complexities.

  • Savings​: Buyers would like to save on various fees in the process that they could instead invest in their new homes.

Dynamics of a Competitive Offer

In markets where there is more demand than inventory, sellers often receive multiple offers with bids that are at or above the listing price. Buyers prepare by optimizing every aspect of the offer to make bids more attractive to sellers.

  •  Offer Bid:​ Aligned with available inventory, demand, and comparable to recently closed transactions.

  • Offer Type:​ Sellers often prefer an all-cash bid due to the increased likelihood of closing compared to certain financed offers. Financed offers come with varying loan statuses such as pre-qualified, pre-approved, and underwritten pre-approved.

  • Contingencies:​ These are a variable for sellers in the closing contract. Being flexible with contingencies such as inspection, appraisal, and sale of current home makes offers more attractive to sellers.

  • Closing Time-frame:​ Sellers may prefer a flexible time frame for closing.

  • Personal Letter:​ Including a personal letter increases a buyer’s chances of winning a bid, especially when all other bids are equally competitive.

Hypotheses & Potential Solutions:

Hypothesis 1​: Buyers want to know how to evaluate their overall offer package based on MLS real-time data and other factors to gauge the likelihood of offer acceptance.

Hypothesis 2​: Buyers will improve their offers by following Opendoor’s recommendations.

Feature 1:

”Offer Simulator” - Opendoor shows an offer’s chances of acceptance on a scale of ‘Low-Average-High’ based on the bid, listing price, market conditions, and other factors.

When buyers enter details in the Offer Simulator,​ like offer amount, preferences for agent selection, how the purchase will be financed, contingencies, closing-time, and inclusion of a personal letter, they are provided with the following assessments:

(a) Chance of acceptance on a scale of Low-Average-High.
(b) List of recommendations to improve the offer’s chances of acceptance.

Hypothesis 3​: Buyers want to make a competitive offer within their budget. Showing a suggested bid amount in the offer flow or within Offer Simulator will guide buyers to submit a competitive offer.

Feature 2:

”SUggested BiD” shows the buyer the suggested $ bid in Offer Simulator with details about the factors affecting the bid.

  1. Budget

  2. Listing price

  3. Offer price

  4. Earnest money

  5. Type of transaction (cash or finance)

  6. Loan approval status (pre-qualified, or pre-approved, or underwritten)

  7. Lender type (Opendoor partner or others)

  8. Contingencies (inspection, appraisal, and existing home sale)

  9. Closing time-frame

  10. Timeline (flexible vs hard deadline for the closing)

Riskiest Assumptions​:

  • Opendoor is able to calculate an accurate suggested offer that does not have a negative impact such as overbidding, loss of trust in Opendoor when the offer is not accepted due to getting outbid because of other factors.

  • Suggested price will have a greater predictability and accuracy than ​Zestimate​

  • Suggested price won’t have a negative impact on the number of offers submitted if the suggested offer amount is more than the buyer’s budget or expectations.

  • Offer Simulator & Suggested Offer would provide more optimal and real-time projections than agents.

  • The average buyer’s offer-to-acceptance ratio is low.

Success Metrics:

Quantitative​: Increase offer acceptance by 20% points.

Qualitative​: Offer Simulator & Suggested Offer helped sellers submit a competitive bid.

Qualitative​: High NPS survey score of buyers who have used Offer Simulator to make a bid.

Mockup: offer simulator, suggested bid and buy now features with listing page as entry point


Why Fast? Why Now?

by Shadab Farooqui


What’s Fast?

Fast offers sellers a Fast Checkout button which can be implemented on an e-commerce product page or the shopping cart.

Shoppers with a Fast account can checkout with one click, without having to sign-in again at stores that accept Fast.

The UX feels minimalistic, limiting the number of fields during signup. The shipping address and payment method defaults to the information you enter when signing up for Fast.

Not having options to modify the payment method or shipping address might even pass as a feature and not a gap, though that functionality may get added in the future.

Fast co-founder Allison Barr also highlights not requiring users to fill out forms.

From e-commerce shopping cart to a healthcare provider payment flow, one common friction step is the sign-up and sign-in process required to complete a transaction. What if I could authorize access to my private data, be it payment info or health history, with a single tap?

It makes an argument for Fast being an identity company that does checkout even more intriguing than Fast as checkout company, that also does identity.

It makes an argument for Fast being an identity company that does checkout more intriguing, than Fast as checkout company that also does identity.

Fast Checkout.png

Spurred by the scarcity of available product information on Fast, this post aims to understand the problem space, and contextualize Fast's opportunity to evaluate its potential and possible strategies for success.

The Market

According to research by Digital Commerce, in 2019:

Retail Sales.png
  • 16% or $601 billion of $3.7 trillion retail-spend by U.S. consumers was online. 84% or $3,161 billion was spent in physical stores. 

  • The growth in e-commerce spend versus in-store was the biggest annual percentage point jump since 2000.

  • E-commerce accounted for more than half—56.9%—of all gains in retail last year — largest share of growth for online spending since 2008.

Adding to the momentum of e-commerce growth is COVID-19, which has confined consumers to their homes, pushing even the most reluctant consumers to shop online. 

It’s probably safe to hypothesize that 2020 will go down in history as the year for the highest recorded e-commerce growth, and the lowest ever for in-store spend.

“2020 will go down in history as the highest recorded year for e-commerce growth, and the lowest ever for in-store spend.”

The Problem

A checkout experience begins after a user has added items to their cart and explicitly clicks or taps to ‘checkout’. Consumers face checkout friction any time they have to fill a form to sign-up, sign-in or enter billing information.

A typical e-commerce checkout flow looks like this:

  1. User visits the product page and adds item to cart

  2. User visits cart on multiple devices (75% abandon cart) & clicks check-out (47.5% abandon checkout)

  3. User signs-in or signs-up to checkout OR uses Guest checkout OR uses PayPal

  4. User enters shipping address

  5. User enters credit card & submits payment

“Consumers face checkout friction any time they have to fill a form — to sign-up, sign-in or enter billing information.”

A study by Baymard Institute to understand “reasons for abandonment” found the following reasons given by users for checkout abandonment:

abandonment reasons.png

  1. 55%: Costs too high (including shipping, taxes etc.)

  2. 34%: Site requiring to create an account  

  3. 26%: Too long or complicated checkout process

  4. 21%: Could not see total order cost upfront

  5. 17%: Didn’t trust the site with credit card information

Opportunity

The changes in buyer and seller demographics could signal the opportunity for a renewed payments and checkout experience.

Millennials & GenZ: 

  • Today, over 50% of online shoppers are Millenials. However,  according to Accenture, GenZ would make up to 40% of US consumers by 2020, describing them as a mobile-native “see now, buy now” segment. Moreover, only 28% of GenZ trust their financial institutions to be fair and honest, making the majority more likely to try non-traditional ways of managing and spending their money.

  • Accenture further posits “They are likely to be the first generation to forgo the leather wallet for the digital wallet…they will also force traditional players to elevate mobile payments value as a matter of survival…to deliver, the industry must design payments experiences around human needs...delivering a unified mobile payments experience is ground zero in the battle for the consumer”  

Direct-to-Consumer:

  • According to eMarketer, D2C brands will account for $17.75 billion of total ecommerce sales in 2020, up 24.3% from the previous year, while 40% of US internet users expect D2C brands to account for at least 40% of their purchases within the next five years.

  • A differentiated payment experience that meets the needs of 87.3 million D2C ecommerce buyers could help carve out a niche to win early adopters.

Casper Checkout.png

Creators: 

  • Creators are individuals who sell digital goods and services as one-off transactions or recurring subscriptions through marketplaces such as Patreon and Gumroad. The products are non-traditional, but the checkout flow and payment methods are often modeled after a retail e-commerce checkout flow

  • A frictionless, one-tap checkout experience could have a meaningful impact on conversion rates for creator marketplaces and data driven creators such as Daniel Vassallo.

Gumroad Checkout.png

Publishers: 

  • The shopping cart equivalent for a publisher such as the New York Times is the paywall - a pop-up window that requests the user to sign-in or subscribe.

  • NYT is the global leader in digital subscriptions according to assessments by Statistica. In 2019, it passed $420 million in revenue through its 5 million news subscribers - a new subscriber record.

  • NYT is experimenting with dynamic paywalls, including bundling, pricing and login/signup elements. However - a big piece of the paywall puzzle to unlock next 5 million users could be a payments experience that feels more native and quick (like Apple Pay but for ALL devices) and less of a process, like the “Checkout with PayPal or Card” flow (see screenshot below). 

NYT-Mobile-Subscription.png
public.jpeg

Conversational Commerce: 

  • Per Wikipedia, conversational commerce is e-commerce done via various means of conversation including live support, online chat, chat-bots and voice assistants.

  • According to  McKinsey, 31% of Chinese WeChat users made a purchase through WeChat in 2016 - double the rate of 2015. China is leading the way with the Chatbot revolution.

  • After adding a shopping feature to Instagram with a “checkout” & pay option within the app, Facebook launched a catalog feature for WhatsApp, letting users showcase their products along with pricing. Unlike Instagram, the checkout happens outside the app.  

  • A seamless mobile optimized checkout experience for social could open an untapped (no pun intended) shopping channel for the seller.

    Update 5/19: Facebook partners with Shopify to launch ‘Shops’, turning every Instagram business profile into a storefront.

Whatsapp Catalog.png

Update 12/8: WhatsApp launches shopping cart feature: https://blog.whatsapp.com/making-it-easier-to-shop-on-whatsapp-with-carts/

Charitable Giving:

  • Charitable giving is 2.1% of the GDP. In 2018, the largest source of US charitable giving came from individuals -- $292.09 billion, or 68% out of the $427.71 billion total charitable giving. The majority of it to the tune of 29% went to religion according to NPTrust.

  • While 49% of church giving transactions are made with a card, 60% are willing to give to their church digitally according to NPSource. A digital payment experience optimized for the live audience could increase giving while adding convenience for the attendees.

Competitive Landscape

Completing an e-commerce transaction with the lowest amount of effort is a delightful experience. Which is why, any payment processor or a marketplace that has traction with buyers and sellers creates its own “Pay with X” button.

Following is a matrix with payment incumbents who offer the distribution to sellers and convenience of a more seamless checkout to shoppers.

Matrix.png

PayPal is the most used alternative payment option, and third most used payment option after debit cards and credit cards used by 305 million users and 22 million merchants.

PayPal‘s One Touch product has 70 million registered users and 6 million merchants as of 2017 (source). It has also integrated Braintree into its onboarding flow to offer sellers payment processing services.

PayPal’s “One Touch” product has 70 million registered users and 6 million merchants as of 2017

Stripe has been used by over 50% of Americans who use it without ever noticing the underlying technology.

It flirted with the idea of a “Pay with Stripe” button, and instead settled on a Stripe hosted checkout experience that allows the shopper to save their billing information that is attached to their email address and phone number. 

In future visits to Stripe merchants, a repeat shopper receives an SMS to the phone number attached to the email provided at checkout. The customer can then input that code to checkout without having to re-enter their card information. 

One might notice that the “Pay with PayPal” option is missing from Stripe checkout integration, alongside Apple Pay, Google Pay and Microsoft Pay likely because PayPal acquired Braintree in 2013, becoming a direct competitor to Stripe.

Closing Thoughts

PayPal found PMF with the Pay with PayPal button after an eBay user reached out to the PayPal team asking for permission to use the PayPal logo in its auction listing. PayPal challenged the status quo of credit card payments and had to deal with fraud and other operational issues along their journey to product-market fit. (PayPal Wars has a great account of the early days of PayPal).

Stripe’s Collison brothers challenged the status quo of a lengthy, multi-step process to accept credit card payments on the web. With no monthly fees or minimums (free to create an account), Stripe created a paradigm shift by enabling developers to accept payments with a single API call. Stripe also became a case study for developer adoption, helping it achieve a valuation of $36 billion in the most recent round of funding.

“In a great market — a market with lots of real potential customers — the market pulls product out of the startup. Conversely, in a terrible market, you can have the best product  in the world and an absolutely killer team, and it doesn’t matter — you’re going to fail.” - Marc Andreessen

Fast is essentially a a two-sided marketplace of buyers who seek a delightful checkout experience, and sellers who want to optimize their conversion rates.

The largest incumbent today is PayPal with over 10 billion transactions in 2019, and a competing product called One Touch. It has recognition and distribution with millions of buyers and sellers.

As Stripe led the Series A investment, it’s possible that Fast could get help by Stripe with merchant adoption, allowing merchants to dynamically display a Fast button to Fast shoppers.

Not having to fill out forms to pay is diffrentiated value that Fast needs to validate with scale — millions of buyers, sellers and transactions.

The initial feedback about Fast checkout from buyers using it on the Fast swag store has been positive. If this positive feedback can scale and Fast continues to innovate in making the payments journey more delightful, it has the potential to become an attractive offering to both the shopper and the merchant.


Adopting San Francisco

by Shadab Farooqui


One thing that I really admired about NYC was the street cleaning that happened on both a daily and weekly basis all over the city. The apartment buildings "adopted" the sidewalks and kept it clean by washing them with water, each morning. This is not to say that New York is the cleanest city in the world, but I admired the dedication and resolve with which the doormen washed the streets with a hose every morning.

San Francisco has a littering problem. It's everywhere, it is apparent and it is getting worse.

For every company or startup, being in San Francisco should be a privilege. Maintaining a street should be corporate responsibility.  Companies should more directly contribute toward keeping their streets clean, in addition to paying city taxes. If you took pride in your second home, i.e. your workplace, wouldn't you naturally care if the sidewalks on your street were clean? Why are clean sidewalks more of a luxury in San Francisco than in any other major US city?

Imagine the density of companies in the SoMa/South Beach area -- there are hundreds of companies and thousands of individuals per street block. If each individual could make a contribution in a small way, the systems to support the cleanliness initiatives would get funded. 

Ideas for starting some kind of an "adopt a street" program:

1) Fund more of those cleaning trucks or increase the frequency of cleaning. The costs should be covered by companies collectively (similar to an HOA)

2) Use private companies that maintain the sidewalk for you. Each company pays a monthly subscription fee. 

3) Push elected officials to take action

Will it make it more expensive to start and grow a company in SF? 

Absolutely, and why shouldn’t it?


5 Stages of Learning

by Shadab Farooqui


Learning Stages and Channels:

Stage 0: Unknown-Unknowns  <— Explore Twitter

Stage 1: Known-Unknown <— Explore Google

Stage 2: Familiarity with topic <— Explore and Exploit Google & other content Pipes

Stage 3: Comfortable with topic <— Practice via actual or simulated projects and situations

Stage 4: Advanced Knowledge <— Practice, Teach and share knowledge via various mediums and distribution channels. 

Stage 5: Expert knowledge <— Share what’s most important and what can be ignored via various mediums and distribution channels.    


98 Targeting Options In Facebook

by Shadab Farooqui


Facebook and Google are the most formidable threats to the entire ad industry. The depth of Facebook targeting is unprecedented and it will continue to add unique 3rd party data sources to go deeper. 

"When combined (3rd party data from Axiom, Epsilon, Experian) with the information you’ve already given Facebook, you end up with what is arguably the most complete consumer profile on earth: a snapshot not only of your Facebook activity, but your behaviors elsewhere in the online (and offline!) worlds."

Orignal article: https://www.washingtonpost.com/news/the-intersect/wp/2016/08/19/98-personal-data-points-that-facebook-uses-to-target-ads-to-you/

Targeting options for Facebook advertisers*
1. Location
2. Age
3. Generation
4. Gender
5. Language
6. Education level
7. Field of study
8. School
9. Ethnic affinity
10. Income and net worth
11. Home ownership and type
12. Home value
13. Property size
14. Square footage of home
15. Year home was built
16. Household composition
17. Users who have an anniversary within 30 days
18. Users who are away from family or hometown
19. Users who are friends with someone who has an anniversary, is newly married or engaged, recently moved, or has an upcoming birthday
20. Users in long-distance relationships
21. Users in new relationships
22. Users who have new jobs
23. Users who are newly engaged
24. Users who are newly married
25. Users who have recently moved
26. Users who have birthdays soon
27. Parents
28. Expectant parents
29. Mothers, divided by “type” (soccer, trendy, etc.)
30. Users who are likely to engage in politics
31. Conservatives and liberals
32. Relationship status
33. Employer
34. Industry
35. Job title
36. Office type
37. Interests
38. Users who own motorcycles
39. Users who plan to buy a car (and what kind/brand of car, and how soon)
40. Users who bought auto parts or accessories recently
41. Users who are likely to need auto parts or services
42. Style and brand of car you drive
43. Year car was bought
44. Age of car
45. How much money user is likely to spend on next car
46. Where user is likely to buy next car
47. How many employees your company has
48. Users who own small businesses
49. Users who work in management or are executives
50. Users who have donated to charity (divided by type)
51. Operating system
52. Users who play canvas games
53. Users who own a gaming console
54. Users who have created a Facebook event
55. Users who have used Facebook Payments
56. Users who have spent more than average on Facebook Payments
57. Users who administer a Facebook page
58. Users who have recently uploaded photos to Facebook
59. Internet browser
60. Email service
61. Early/late adopters of technology
62. Expats (divided by what country they are from originally)
63. Users who belong to a credit union, national bank or regional bank
64. Users who investor (divided by investment type)
65. Number of credit lines
66. Users who are active credit card users
67. Credit card type
68. Users who have a debit card
69. Users who carry a balance on their credit card
70. Users who listen to the radio
71. Preference in TV shows
72. Users who use a mobile device (divided by what brand they use)
73. Internet connection type
74. Users who recently acquired a smartphone or tablet
75. Users who access the Internet through a smartphone or tablet
76. Users who use coupons
77. Types of clothing user’s household buys
78. Time of year user’s household shops most
79. Users who are “heavy” buyers of beer, wine or spirits
80. Users who buy groceries (and what kinds)
81. Users who buy beauty products
82. Users who buy allergy medications, cough/cold medications, pain relief products, and over-the-counter meds
83. Users who spend money on household products
84. Users who spend money on products for kids or pets, and what kinds of pets
85. Users whose household makes more purchases than is average
86. Users who tend to shop online (or off)
87. Types of restaurants user eats at
88. Kinds of stores user shops at
89. Users who are “receptive” to offers from companies offering online auto insurance, higher education or mortgages, and prepaid debit cards/satellite TV
90. Length of time user has lived in house
91. Users who are likely to move soon
92. Users who are interested in the Olympics, fall football, cricket or Ramadan
93. Users who travel frequently, for work or pleasure
94. Users who commute to work
95. Types of vacations user tends to go on
96. Users who recently returned from a trip
97. Users who recently used a travel app
98. Users who participate in a timeshare