AI's Impact on the Software Industry
with Des Traynor, founder of Intercom
AI can write codes. What does that mean for the future of the software industry?
Recently, I listened to an episode of the "Invest Like the Best" podcast featuring Des Traynor, the founder of Intercom, a customer service software provider.
Traynor shared profound insights into how AI is poised to influence the software industry, strategies to thrive in this fiercely competitive and rapidly evolving space, and his perspectives on investing.
Let’s dive into the takeaways.
On how AI impacts software development
The rise of AI capabilities, such as ChatGPT, is sparking significant questions regarding competition within the software industry.
The level of threat hinges on whether new AI completely invalidates existing codebases. If the core architecture of software remains sound, incumbents can integrate AI into their stack, while leveraging users’ familiarity and branding. However, if AI necessitates a ground-up rebuild, new entrants gain an advantage by architecting for AI natively.
Many AI startups are often hyped as full-fledged products when, in reality, they are just features. Winning software requires robust execution and a well-thought-out strategy, encompassing clean user interfaces, seamless integrations, and a clear route to market.
In this dynamic landscape, speed and decisiveness are critical for software founders. When core assumptions change, pivoting rapidly becomes essential, as exemplified by Intercom's response when ChatGPT was publicly released.
Regarding AI capabilities, differentiation lies beyond the large language model itself, encompassing the workflows and data stores built on top. Achieving commercial viability in mission-critical sectors like healthcare demands accuracy, as translating AI features into robust enterprise products that consistently provide credible answers remains a formidable challenge.
Traynor discussed the high cost of using GPT-4 on the backend, which has resulted in elevated pricing for Intercom's products. While OpenAI is the first-mover, competitors are steadily catching up, and pricing should trend downward over time as customers demand the best value.
Durable moats are a rarity in software. Consistent innovation and branding remain vital for staying ahead of the curve.
In summary, AI is poised to augment rather than replace human roles within the software industry. Achieving success in the AI-powered software era necessitates pragmatism regarding AI's current capabilities, coupled with a vision of its long-term potential.
On the importance of execution
"Strategy is for amateurs, execution is for professionals," says a veteran software founder. In a world where AI can write code, success in the software industry still boils down to meticulous execution.
The core idea behind AI-powered software may be brilliant, but building an exceptional product around it demands immense effort. Many hyped startups merely add an AI feature to lackluster software. Truly winning solutions require seamless integrations, an aesthetically pleasing user interface, and well-thought-out workflows that effectively utilize the AI's output.
The route to market and understanding user willingness to pay are also indispensable factors. Even a theoretically solid product can falter if positioning and distribution are flawed.
I found Traynor’s 4-point investing framework for startups to be insightful: Start-up founders should clearly articulate
What makes their solution unique (can others do it?)
Its value to users (do users give a f**k?)
Is there really a need for this product? (or not?)
Its simplicity (is the pitch too complex?)
If any of these pillars crumble (and the descriptions in parentheses hold true), the chance of achieving outsized success diminishes.
On SaaS competition
Once a software product gains traction, new competition becomes inevitable as everything eventually gets copied over time.
The most effective defense against this onslaught is relentless innovation—consistently rolling out features that users crave, rendering rival versions perpetually outdated. The substantial time investment required to replicate an established product acts as a formidable barrier to entry for disruptors.
A parallel can be drawn from what I read in "Chip War," where USSR spies stole American semiconductor technology. However, the challenge for the USSR was that once they learned how to manufacture the product, Americans were already advancing to the next technology node. This continual innovation left the USSR lagging, which had a direct impact on the sophistication of the chips used in Soviet weapons, perpetually putting them at a disadvantage on the battlefield.
When this technical lead is coupled with robust branding (think of examples like Slack, Stripe, Zoom, Figma) and active community engagement (integration with other software, fit in the customer's tech stack, data sharing agreements, robust developer community, etc.), it creates a significant competitive edge.
The key to winning and establishing a moat in the software industry boils down to several factors:
Building the best product and moving swiftly
Shifting from a technical positioning to a brand positioning
Expanding reach into the hearts and minds of customers and partners
Software products have a tendency to become obsolete quickly, requiring continuous reinvestment to maintain your position.
While SaaS operating leverage has its advantages, it also has limits. Trying to serve too many customer segments can lead to bloated code and a loss of ease of use. The most successful products strike a balance between broad appeal and simplicity in design, necessitating constant fine-tuning.
Building software can be challenging because the market can be segmented in many ways, each demanding different design and customization. As you expand into different segments, your workload grows, and the user experience may deteriorate if not managed carefully.
Bloomberg Terminal is an example
There is a Financial Times article that discusses the enduring success of the Bloomberg terminal, despite numerous attempts by competitors. There are discernible patterns at play here.
Bloomberg was the first mover, driven by Michael Bloomberg's visionary pursuit of creating a software product that could consolidate real-time market data into a single platform. One can only imagine the complexity of such an engineering feat.
While competitors try to replicate Bloomberg’s core functions, Bloomberg has evolved into more than just a data terminal. It has transformed into a sales and trading platform, a social network, and a centralized hub for various verticalized use cases.
However, I can imagine the challenge at first is to make the economics work because a Bloomberg terminal primarily caters to the financial services industry. It’s a smaller user count than if it can cater to diverse end markets.
Thankfully, as it continued to infiltrate its core market, the network effect grew stronger: Wall Street professionals found themselves connecting with peers, acquaintances, and even strangers over the terminal.
The product also offered tools such as an Excel plugin to pull market data into their spreadsheets, saving time. However, this integration significantly increased the switching cost. (Personally, I'd dread losing my BDH functions.)
Similarly, Bloomberg tackled a significant issue by bridging the gap between tools and data, offering robust quality assurance in a profession that places a premium on accuracy and timeliness. This further ingrained Bloomberg in the hearts of its customers and partners.
With a high switching cost and growing value-add, Bloomberg garnered significant pricing power. Charging $25,000 became feasible due to its clientele's high compensation and relentless focus on a go-to-market motion that centers around value.
As the adoption rate within finance grew, the Bloomberg terminal became a status symbol beyond a productivity tool. This shift exemplifies the transition from a technical positioning to a brand positioning.
Will AI ever fully replace humans?
True replacement is unlikely - AI will mainly play an augmenting role. In mission-critical fields like healthcare and transportation, human override and oversight remain indispensable.
Technology has its limits - no algorithm can replicate the nuance and adaptability of human cognition. AI may assist in diagnosis and treatment, but doctors will still lead in patient care. Autonomous vehicles require vigilant human monitoring.
Know when to fold
For startup founders, having ample funding runway shouldn't justify pursuing dead-end ideas indefinitely. Be realistic about your goals and set a specific timeframe to demonstrate traction. If success doesn't materialize within 6-12 months, consider working for someone else rather than wasting your best years.
As a first-time entrepreneur, thankfully I understood the value of personal capital - I gave myself one year, and now I am actively looking for a job to continue my career with a company.
Thanks for reading. I will talk to you next time.
If you want to advertise on my newsletter, contact me 👇
Resources for your public equity job search:
Research process and financial modeling (10% off using my code in link)
Check out my other published articles and resources:
📇 Connect with me: Instagram | Twitter | YouTube | LinkedIn
If you enjoyed this article, please subscribe and share it with your friends/colleagues. Sharing is what helps us grow! Thank you.
If you're interested in listening to the entire podcast, click on the link below:
Hi Richard, I hope you're doing well. Massive fan of your Substack.
I am a founder of Zeed (https://zeed.ai/). We're helping creators transform their written content (like this Substack) into dynamic video pieces using AI to capture new audiences.
Would love to show you an example and jump on a call if you're interested in hearing more! Best, Rohan.