Historically, software and services have operated in separate worlds, a line that is becoming blurry today. Global software spending has surged to nearly $900 billion. Software companies have high upfront investment and strong gross margins. In contrast, services businesses require less capital to start and scale linearly with leaner gross margins.
Over the past decade, software businesses have dominated the spotlight due to their repeatable revenue streams, fast cash collection, and attractive margins. With the right PMF and a scalable GTM strategy, they can grow rapidly, which explains why SaaS companies command high valuations despite their linear sales processes.
Services, on the other hand, often lack differentiation and predictability. Operating under a “we’ll do whatever you need” blanket, they struggle to achieve clear product-market fit till they reach a certain scale. For these businesses, their ability to grow rely on factors like sales capabilities, delivery scalability, cost management, and securing long-term contracts. As these companies mature, they often attract growth-stage or private equity investors.
The emergence of large language models (LLM) has dramatically altered the frontier of what is possible for machines to do, particularly for tasks that are sequential and involve both unstructured and structured data. As a result of this, there are two key trends we’re seeing:
- AI agents are taking over many tasks that are either being done in-house by companies or by services companies, cutting down on the need for human effort.
- As software becomes more AI native, its inherent inaccuracies will make a human-in-the-loop layer essential to keep everything running smoothly.
While #1 is widely understood today, we believe that there is a lack of appreciation for #2. AI software are fundamentally probabilistic, and are prone to hallucinations. Even 90% accuracy is considered high. Some of this may be acceptable for low criticality applications e.g. prediction of next movie a user may want to see, or autocomplete in a text box. However, for most business works, this is not acceptable. Take, for instance, a real estate software startup we recently spoke with. They’re focused on automating the closing process for residential real estate transactions in the US. This process involves multiple steps, including exchanging emails and scanned documents, extracting key information, performing actions based on that data, navigating and inputting information into packaged software, and handling payment transactions. In total, there are around 20-50 such micro-steps. Even if each individual step has an accuracy rate of over 95%, the likelihood of completing all 20 steps without any errors drops to just 35%. This highlights the complexity and potential for errors in automating such intricate workflows. For 50 steps, the probability drops to just 7.5%. You can’t afford to manage a $1M+ transaction with a 90% chance of error!
Today, most software companies offer AI solutions but place the responsibility for outcomes on the buyer. However, we believe that over the next decade, a growing number of companies will emerge that provide both software and services. These services won’t be limited to one-time implementation; they will also include ongoing, human-led support to manage edge cases and ensure oversight, guaranteeing the accuracy and reliability of AI-driven outputs.
The emergence of ‘Platform BPO’ companies
The nature of these emerging companies is neither purely software nor services, but a true hybrid of both. Traditional services companies typically offer a wide range of services, operating on a simple principle: once inside an account, expand as much as possible. The lines between “what we do” and “what we don’t” are often blurred, leading to continuous service expansion. As contracts grow larger and longer, sales costs as a percentage of revenue decrease, while resource utilisation – an essential driver of profitability – increases, thanks to more predictable, long-term engagements. For Indian services firms, an additional profit driver is the ability to offshore a significant portion of their workforce. This strategy has helped create global giants like TCS and Infosys, built on this very model.
It’s important to remember that even software companies often need services – sometimes seen as a necessary evil. When working with large enterprises, consulting, implementation, and ongoing support are crucial. While these services typically carry lower margins than software itself, they are essential for ensuring value realization and reducing churn, making them indispensable. Software companies also require significant upfront investment and continuous R&D to stay competitive. As a result, most remain highly focused on their core offerings. It’s usually only when they reach $100M+ in ARR that they start expanding their scope to boost ARPA (Average Revenue per Account), and even then, they carefully select adjacent areas for growth.
What we’re discussing here is something different – a new breed of companies that operate with the focus and efficiency of a software company but also provide services that help customers fully realize the value of the underlying software. These services, which might otherwise be handled by the customer or a third-party BPO, are managed directly by the company itself. The key distinction from traditional software companies is that these companies take full ownership of the process and outcomes. At the same time, they have no interest in expanding their service portfolio; they limit themselves to the specific services required to support their core product. We refer to these companies as ‘Platform BPO’ companies.
We’re witnessing a growing number of Platform BPO companies emerging from India. These companies benefit from a unique India advantage, as their services can be delivered remotely, leveraging labour arbitrage. In their respective sectors, we believe these Platform BPO companies offer a more compelling value proposition to customers compared to pure software companies. Unlike software companies, Platform BPOs guarantee outcomes, providing a level of assurance that software-only solutions cannot match.
This concept is also quite distinct from the current “Services as Software” narrative. Take Devin, for instance –it offers a software developer on demand, effectively allowing you to hire this “coding agent” as part of your software team. However, at its core, it remains just software. While it heavily leverages AI and performs tasks that were once unimaginable for software, it’s still software. Its pricing model may resemble services, but it’s important not to confuse it with true services – it’s simply packaged differently. Many AI-driven agentic applications today follow this model.
Economics of Platform BPO
We expect Platform BPO companies to deliver 60%+ gross margins, 20-25 percentage points higher than typical services companies. They will, however, be lower than classical software companies where gross margins can be 80%+ as well.
One of the key differences between Platform BPO companies and traditional services firms is their investment in R&D. While services companies typically allocate only 1-3% of their revenue to R&D (yes, they talk a lot more about it!), Platform BPO companies operate much like software companies in this regard. Even as they scale, we expect their R&D spend to be at least 10%, and potentially as high as 20%, depending on the industry they operate in.
At the same time, Platform BPO companies can command a higher ACV compared to traditional software companies. Why? Because they guarantee outcomes and take full ownership of the process. For instance, Spry, an India-based Platform BPO that helps physiotherapy clinics in the US collect payments from payors, charges a percentage of all the cash they collect. Similarly, the real estate company we discussed earlier charges per transaction closed. This is a classic pay-for-performance model, where customers only pay for results.
Founder and Team Archetypes
Balancing a product-first approach with a services-first approach can be challenging. We believe that founders who adopt a product-first mindset, while also having co-founders and/or founding team members with strong experience in services, can offer a fresh perspective. This combination allows for a more holistic approach, where the focus on building a product to solve specific pain points and bringing repeatability is complemented by the service delivery mindset and GTM acumen that experienced service professionals bring. Such a blend can lead to more effective solutions that address both product development and customer needs, ultimately driving better outcomes.
The convergence of software and services will create a whole new category of enterprise businesses and with services spending outpacing software spending, this is a golden opportunity waiting to be addressed.
If you’re building a company that leverages AI to automate a services segment we’d love to hear from you. Please reach out to us at sayantan@stellarisvp.com or nj@stellarisvp.com.