The future of SaaS looks bleak. Recently, Klarna CEO Sebastian Siemiatkowski announced that the company will end its service provider relationships with Salesforce and Workday as part of a major internal overhaul driven by AI initiatives.
“This news from Klarna should have every enterprise SaaS company shaking in their boots. If an internal team using AI can replicate over 20 years of work and customisation from Salesforce and Workday, to the extent the company doesn’t feel the need to pay for these tools anymore, everything we know about the stickiness and durability of enterprise software needs to be rethought in the light of AI,” Gokul Rajaram, former Coinbase board member and early stage startup investor, said.
Klarna is not alone. In an exclusive interview with AIM, Harneet SN, founder of Rabbitt AI, revealed that he is partnering with the University of California, Berkeley, to create AI-powered tutors for their online learning programs. He said that the university has opted to move away from its previous SaaS providers in favour of building in-house solutions.
Harneet cited an example of another company that previously relied on voice bots and chatbots from Yellow.ai. However, as they seek to integrate generative AI capabilities into their voice bots, they are now opting to develop their own chatbots. “Yellow.ai’s SaaS-based chatbots are not working well for them in the GenAI era,” he added.
Many feel that with the advent of GenAI coding tools like GitHub Copilot and Anthropic’s Claude, one can expect software development to become cheaper and the job market for coders to evolve, creating a more accessible environment for talent, although at lower price points.
“You can rebuild most enterprise SaaS functionality, host for super cheap, and get basically [over] 90% functionality,” said Akber Khan, founder of Evolve Machine Learners. His company is currently building in-house SaaS solutions for large enterprises.
Building on this idea, a user posted on X, “My general belief is that the next decade is going to be B2C, not B2B. Generally, AGI/AI will increase people’s capabilities enough [so] that B2B SaaS won’t be a good purchase. You can just build it in-house.”
SaaS companies are not alone. Indian IT giants such as TCS, Infosys, Wipro, HCLTech, and others, which are arguably still testing the waters of GenAI, could soon face a critical challenge when spending on their services becomes redundant. With AI tools, anyone within enterprises may be able to build front-end, web, and other applications with minimal coding.
AI Coding Tools Are All You Need
“Generative AI can now take on substantial workloads in software development, particularly in bug fixing, vulnerability remediation, and even optimising code quality,” said Asankhaya Sharma, founder of Patched AI. He added that Patched’s work with LLMs has shown significant progress in reducing the developer workload on security fixes by automating patch generation for code vulnerabilities.
He explained that developing in-house AI involves higher initial costs due to infrastructure, talent acquisition, and maintenance expenses. However, in the long run, SaaS costs can add up as subscription fees scale with use.
Similarly, Harneet said that SaaS is comparable to an EMI rather than a one-time payment. He said, “While the upfront cost of in-house solutions may be 10-15 times higher over two to three years, the ROI (return on investment) will ultimately be positive.”
“The reality is that there’s no universal solution. The choice between in-house AI development and SaaS adoption depends heavily on an organisation’s specific situation, goals, and resources,” said Pradeep Sanyal, AI and data leader at a global tech consulting company.
Sanyal added that, when it comes to software development, AI can handle significant advancements in the workload. He pointed out that AI excels at automating repetitive tasks, suggesting code snippets, and assisting with basic debugging.
“In-house AI solutions can be scalable if they are well-architected. However, many AI solutions fail in production because they were initially built as proof-of-concept projects that didn’t account for the demands of production-scale environments. SaaS offerings, designed for enterprise requirements, may offer better scalability in many cases as they are thoroughly tested across various workloads,” said Pavan Nanjundaiah, head of Tredence’s studio innovation team.
Rise of Vertical SaaS
Harneet said AI has the capability to solve very niche problems. “The companies that focus on niche problems and solving them…will be really successful,” he said.
Along similar lines, A16z recently published a blog which said AI is unlocking a new era for vertical SaaS. In functions like marketing, sales, customer service, and finance, AI will augment, automate, or, in some cases, replace many of the rote tasks currently performed by people, allowing VSaaS companies to offer even more with their software.
“The early winners in LLM-based solutions might just be general-purpose platforms. Over time, vertical AI agents will emerge. It’s like how, in the box software world, the early vendors were just trying to convince people to use software…As the market matures, it will get more sophisticated, and vertical solutions will become dominant players,” said Jared Friedman, group partner at Y Combinator, in a recent podcast with YC president Gary Tan while discussing the future of SaaS.
Harneet believes that in the era of AI, SaaS will become a sidekick. “It will be an enabler, but SaaS is not the main focus. In the internet world, SaaS was a game because the network was king, but in the AI era, the solution is king.”
Notably, all SaaS companies, including Salesforce and Oracle, are stepping up efforts to integrate AI solutions into their offerings. At Cloudworld 2024, Oracle announced that over 50 AI agents within its Fusion Cloud Applications Suite would automate a number of tasks that will help streamline business processes, deliver personalised insights, and boost productivity across various functions, including finance, supply chain, HR, and sales, among other tasks.
Similarly, at Salesforce’s biggest tech summit, Dreamforce 2024, the company unveiled Agentforce Partner Network, which brings together tech giants such as AWS, Google Cloud, IBM, and Workday to enhance the AI-powered Agentforce platform’s capabilities.
It would be naive to declare SaaS dead. Rather, existing SaaS companies are evolving into AI-first entities. “Legacy and new SaaS companies will truly become AI-first (not just marketing), abstract away the complexity of deploying LLMs,” said Matt Turck, VC at First Mark Cap, adding that Artificial Intelligence as a Service (AIaaS) has become the new SaaS.
SaaS companies such as Zoho, Freshworks, CleverTap, and Atlassian have added GenAI capabilities to their existing solutions. Meanwhile, numerous AI startups are developing new products based on these generative AI technologies. According to AIM, approximately 60 AI startups in India are building products using generative AI.
AIM got in touch with Priya Subramani, VP and GM of customer experience products at Freshworks, who said, “My view is that you should use your technology where it’s core to your business. For anything else, like customer service or other functions, you can adopt best practices from industry leaders.”
On the other hand, NVIDIA chief Jensen Huang believes that SaaS is sitting on a goldmine. “These platforms are sitting on a gold mine. There’s going to be a flourishing of agents specialised in Salesforce, SAP, and other platforms.”
Every AI startup that is not engaged in core research effectively becomes a SaaS company. For instance, several companies are developing AI-powered conversational platforms for customer care, such as Exotel, Freshworks, Gupshup, CoRover.ai, LimeChat and Yellow.ai.
However, not everyone believes the same. “Most SaaS companies rebranding themselves with AI are barely adding a limited chat functionality where users can ask a limited set of predefined questions and get directed to the dashboard for answers. There’s no AI here, just a bad UX that delivers little to no value to the end customer,” said Divyaanshu Makkar, co-founder of WizCommerce.
Companies like Ema and Alchemyst AI are working on developing ‘AI employees’ for businesses. In a recent blog titled ‘Why SaaS is Dead and the Future is Agentic’, Ema explains how Agentic AI overcomes the limitations of traditional SaaS. These AI agents can manage entire tasks on their own, make decisions, plan, and improve their performance based on feedback, similar to how humans learn and adapt over time.
Hybrid Approach
According to Sanyal, while the decision to choose between in-house AI and SaaS is complex, most enterprises are gravitating toward a hybrid model. “Companies are using SaaS to initiate quickly and fill gaps while building in-house capabilities for their most critical, differentiating AI needs. This pragmatic approach balances speed, cost, control, and long-term strategic value,” he said.
Echoing similar sentiments, Harneet said that SaaS is a proven model that will not disappear completely and will remain relevant for certain use cases. However, he added that enterprises focused on building core generative AI applications will prefer to develop their own solutions.
“B2B SaaS companies, which are taking a lot of data from businesses, will have a gloomy future,” he claimed.
Building in-house AI solutions also comes with its own challenges. “Building an in-house AI stack can take several weeks to a few months. Initial development may take anywhere from two to four weeks for a basic model, while fine-tuning, testing, and full deployment could add another six to 12 weeks. If rapid deployment is essential, a SaaS solution can be a faster option,” Sharma added.
Harneet explained that not every solution needs to be built in-house. “There are certain areas that enterprises or individuals should not attempt to build in-house, especially those that are not their core expertise.”
“For example, if I am not a finance company but an AI company, I might need a tool to automate the reconciliation of my finances. In that case, I can probably outsource that tool,” he concluded.
