Why AI-native services are the next big bet

Most of the discussion in AI right now is aimed at the application layer. Will agents replace SaaS? How will incumbents defend their position? Will every workflow get rebuilt around a model? These are valid questions, but they assume the prize is the software budget.

Software is only a small part of what companies spend to get work done. The real money in any operationally heavy business is spent on the work itself—adjudicating claims, chasing renewals, assembling compliance files, reconciling records, and carrying information from one system to the next. For 30 years, software has made this work easier to track and route. The actual work, however, has largely remained with people.

From selling software to delivering the work

Traditional software companies sell tools that help customers do their jobs better, but someone still has to do the actual work. AI makes it possible to deliver the workflow, not just the tool. Instead of selling software to process insurance submissions, a company can receive submissions, extract information, check for missing documents, prepare the underwriting file, and send exceptions for human review.

This is the distinction between AI-enabled software and AI-native services. The former makes an existing team more productive. The latter uses AI to deliver an outcome, with humans providing oversight where necessary. And unlike a services business that grows by adding people, an AI-native company should deliver more work with progressively less human effort as the system learns.

Why now, and where is the opportunity?

Two shifts make this possible today. First, language models can read unstructured documents, extract meaning from emails, classify requests, and reason across incomplete information. Work that once required human interpretation can increasingly be handled by software, provided the workflow is sufficiently bounded and the quality bar is clear.

Second, buyers are already comfortable working with external delivery partners. A decade of offshore teams has made outsourcing familiar. Meanwhile, the software stack is crowded. A mid-market insurer, lender or accounting firm is not short of systems - it has systems of record, CRMs, document stores and point solutions. What it does not have is anyone to make all of it produce the outcome.

This creates opportunities across industries such as banking, education, insurance, healthcare, accounting and compliance. For example, in healthcare, AI-native services companies can handle administrative workflows such as revenue-cycle follow-ups and documentation. In accounting, they can reconcile transactions and resolve invoice exceptions. In compliance, they can collect evidence, review documents and prepare filings.

The common thread is not the industry. It is the nature of the work: repetitive enough to learn from, complex enough that traditional software could not fully automate it, and costly enough to justify a dedicated service.

Why India is well placed to build this

This category needs two capabilities at once: building software and running operations with discipline. India has spent decades mastering both.

India's SaaS success demonstrated that Indian founders can build technology products for global markets. Meanwhile, decades of delivering services have built deep expertise in process design, quality management and large-scale operations. AI-native services bring these two strengths together.

These companies will initially need people who understand the work, maintain quality and help the technology learn. That operational knowledge, fed back into engineering, is what eventually creates software leverage. Indian founders can serve global mid-market customers with an efficient delivery model from day one, then use AI to improve margins over time.

What founders need to get right

The first decision is choosing the right workflow. Existing outsourced work has already been through years of price discovery, making it difficult to compete with incumbents that have optimised delivery for decades. The strongest opportunities lie in operational tasks that companies still handle internally because the work is too complicated for traditional software and too costly to outsource economically.

But choosing the workflow is only the beginning. Founders need to understand the exceptions and build a feedback loop between delivery and engineering, so that human corrections improve the system rather than remain with the operations team. The test is whether the same unit of work becomes cheaper to deliver over time.

Finally, pricing must capture the value created. Charging per employee or seat risks passing efficiency gains to the customer as automation improves. Pricing against completed transactions, workflows or outcomes gives the company a better chance of retaining those gains.

The bet

AI-native services expand the venture opportunity beyond software budgets to the much larger cost of getting work done. The first wave of AI helped people perform their jobs more efficiently. The next will increasingly perform those jobs on their behalf.

The best companies in this category will deliver outcomes today and use every delivery to make the next one cheaper, faster, and more automated.

(The author is Partner, Stellaris Venture Partners. Views are personal.)