How artificial intelligence is rewriting the build-versus-buy equation, and why SaaS companies may need to rethink their entire approach to demonstrating value.
For the best part of two decades, the SaaS industry has operated on a relatively straightforward commercial assumption. Businesses need software, developing it internally is expensive and complicated, and buying access to an existing platform is almost always the more sensible option.
It was an assumption that helped create one of the most successful business models in modern technology.
Software companies could develop a product once, sell it repeatedly, charge customers annually and build substantial recurring revenue streams. Customers, meanwhile, benefited from avoiding the considerable expense, complexity and ongoing responsibility of developing and maintaining their own applications.
For the most part, it was a commercially attractive arrangement for everyone involved.
But something rather significant is happening.
The same customers who once needed software vendors to solve their problems are increasingly discovering they can solve some of those problems themselves. Not by hiring large teams of developers or investing millions in proprietary technology, but by using artificial intelligence to build applications that would have been prohibitively expensive, time-consuming or technically challenging just a few years ago.
And while the technology industry remains preoccupied with which AI company will disrupt which software company, a potentially more consequential competitive threat is emerging.
The customer.
This doesn’t mean Salesforce, Microsoft or SAP are about to disappear. Complex enterprise systems are considerably harder to replace than a simple reporting dashboard or workflow automation tool. Building software is also only one part of the equation. Maintaining, securing and supporting it represents an entirely different set of challenges.
But the question facing software companies is changing.
Historically, SaaS salespeople needed to convince customers that their product was better than the alternatives, or that the cost of doing nothing was greater than the cost of investing.
Increasingly, they may need to answer a rather more uncomfortable question.
Why should we pay you for something we could build ourselves?
And I’m not convinced the SaaS industry, or the sales organisations that support it, are particularly well prepared to answer.
To understand why this matters, it’s worth remembering why SaaS became such a compelling proposition in the first place.
For much of the history of enterprise software, organisations faced a fairly unattractive choice. They could invest significant capital in developing their own technology, with all the associated infrastructure, people and maintenance costs, or they could purchase an existing solution from a specialist provider.
Neither was particularly appealing. Traditional enterprise software was expensive to purchase, complicated to implement and frequently required considerable customisation.
The emergence of SaaS changed that calculation.
Rather than purchasing expensive licences and maintaining their own infrastructure, organisations could subscribe to applications hosted and managed by the vendor. Implementation became easier, updates were handled centrally and costs became more predictable.
Perhaps most importantly, businesses could access sophisticated capabilities that would have been impractical to develop themselves.
The commercial proposition was compelling. Why spend years and millions building your own CRM when Salesforce already existed? Why develop a proprietary HR platform when someone else had already solved the problem for thousands of organisations?
The answer was usually that you shouldn’t.
As the model matured, thousands of specialist SaaS companies emerged to address increasingly specific business requirements. Sales engagement, expense management, employee engagement, project tracking, contract management, customer success, sales enablement and countless other categories developed into substantial markets.
Entire organisations built their businesses around relatively narrow pieces of functionality.
And why wouldn’t they? If enough businesses shared the same problem, there was a reasonable argument for developing a product, distributing the development costs across thousands of customers and charging each an annual subscription.
Underpinning all of this was one rather important assumption.
It was considerably easier for the customer to buy the software than to build it.
That assumption is now being challenged.
For years, software development has been constrained by three things: technical expertise, time and cost.
Building even a relatively simple business application traditionally required some combination of developers, designers, infrastructure, project management and ongoing technical support.
Consequently, the idea that an operations manager might decide to replace a subscription-based reporting tool by developing their own alternative was, in most organisations, fairly unrealistic.
Artificial intelligence is changing the accessibility of that process.
AI-assisted development tools such as Cursor, Claude Code and GitHub Copilot allow users to translate natural-language instructions into working software. Low-code platforms are becoming increasingly capable, while AI agents can help with everything from generating code and creating interfaces to debugging and integrating applications.
None of this removes the need for technical judgement. But it changes who can participate in software development and what a relatively small team can realistically achieve.
A process that might previously have required a substantial development project can, in some circumstances, be reduced to a much smaller investment of time and resources.
The consequences are beginning to appear in the research.
In its August 2026 State of AI survey, McKinsey reported that 32% of respondents said their organisations had decided against purchasing at least one software product or feature because they could build the functionality internally using AI coding tools.
That’s an extraordinary finding.
Not because it suggests a third of SaaS products are about to disappear. It doesn’t. Rather, it indicates that organisations are already making purchasing decisions differently because of AI-enabled development.
The impact isn’t hypothetical. It’s influencing commercial decisions today.
Research published by Retool earlier in 2026 provides another indication of the shift. In a survey of 817 builders and customers, 35% reported having replaced at least one SaaS tool with a custom-built solution, while 78% expected to build more internal tools during the year.
The survey comes from a company that sells tools for building internal applications, so it would be unwise to treat the results as representative of every business. Nevertheless, the findings illustrate an emerging pattern.
Customers aren’t necessarily abandoning software altogether. They’re becoming more selective about which software they genuinely need to purchase.
And the more capable AI development tools become, the more attractive that alternative may appear.
There is another reason this shift could prove particularly uncomfortable for SaaS providers.
Over time, many SaaS applications have become extraordinarily complex.
What began as a relatively straightforward solution to a specific business problem has evolved into a platform containing hundreds of features, integrations, dashboards, automations and administrative capabilities.
Some of that complexity reflects genuine customer demand. Enterprise customers have complicated requirements, and vendors understandably develop additional functionality to remain competitive.
But feature expansion also creates an interesting commercial problem.
Customers frequently pay for considerably more software than they actually use.
Consider a hypothetical business paying £40,000 annually for a specialist software platform. Its employees might regularly use just four or five core capabilities, while the majority of the remaining functionality sits largely untouched.
Historically, that might have been accepted as part of the cost of doing business. Building an alternative was simply too expensive to consider.
But what happens when that same business discovers it can create an application covering its most important workflows without undertaking a major development project?
Suddenly, the economic comparison changes.
The relevant question is no longer whether an internally developed solution can replicate every feature of the purchased platform.
It’s whether it can deliver the outcomes the customer actually cares about.
And those are very different questions.
A SaaS provider might reasonably argue that its product offers hundreds of additional capabilities, sophisticated reporting, extensive integrations and years of accumulated development expertise.
But if a customer uses only a small proportion of those capabilities, their willingness to pay for the rest may be limited.
This creates an uncomfortable possibility for software vendors: the breadth of functionality that once supported premium pricing may become less persuasive when customers can develop something narrower and more closely aligned to their requirements.
In certain categories, purpose-built and comparatively simple software may become a credible alternative to feature-rich platforms.
Of course, the internal application must still be reliable, secure and maintainable. But the discussion has shifted from the availability of functionality to the commercial value of that functionality.
And that’s a rather different sales conversation.
Much of the discussion surrounding AI and SaaS has focused on whether entire software categories will disappear.
I suspect this overlooks a more immediate and commercially significant threat.
SaaS businesses don’t necessarily need to lose their customers outright to experience substantial disruption.
Consider a hypothetical customer paying £100,000 annually for a platform.
The customer might decide that the core application remains indispensable. Replacing it would be complicated, risky and economically unjustifiable.
However, they might also conclude that several premium modules, reporting capabilities or workflow extensions could be developed internally for less than the cost of maintaining those subscriptions.
At renewal, the customer decides to retain the core platform but reduce its annual commitment to £75,000.
The vendor hasn’t lost the customer. Its product remains embedded within the organisation, and the relationship continues.
But a quarter of the annual contract value has disappeared.
Multiply that behaviour across a substantial customer base and the implications become obvious.
For businesses valued heavily on recurring revenue, net revenue retention and predictable expansion, relatively modest changes in customer purchasing behaviour can have significant consequences.
The threat may not arrive as a dramatic wave of customers cancelling their subscriptions. It could emerge gradually through smaller contracts, lower adoption of premium functionality, longer purchasing cycles and customers becoming increasingly reluctant to expand their commitments.
There is also a potentially damaging effect on new customer acquisition.
A prospective customer doesn’t have to build a competing application to affect a SaaS company’s economics. They merely have to believe they could.
That belief introduces another alternative into the purchasing conversation, potentially changing how much they’re willing to pay, how urgently they need to act and how much commercial value they expect the vendor to demonstrate.
In other words, the ability to build software can become a source of competitive pressure even when the customer ultimately decides to buy it.
Before declaring the death of SaaS, however, we should acknowledge that building software and operating software are two very different things.
The technology industry has never been short of people enthusiastically announcing that an established business model is about to disappear.
Most of these predictions tend to overlook the rather inconvenient complexities of operating an actual business.
AI might substantially reduce the effort required to create an application, but generating working code isn’t the same as developing secure, reliable, scalable enterprise software.
Someone still needs to understand the requirements, test the application, manage access permissions, protect sensitive information, maintain integrations, address vulnerabilities and ensure the software continues to operate when the organisation changes.
That responsibility doesn’t disappear simply because the original application was developed using an AI coding assistant.
Indeed, the apparent simplicity of AI-assisted development could introduce a different set of problems.
If individual departments can rapidly create their own applications, organisations may find themselves managing an expanding collection of internally developed tools with inconsistent security standards, limited documentation and unclear ownership.
And there’s a particular irony in that outcome.
Businesses that once complained about having too many disconnected SaaS applications could end up recreating precisely the same problem with software they’ve developed themselves.
There is also a danger in assuming that AI always makes development faster.
Research published by METR in July 2025 found that experienced open-source developers working on familiar codebases took 19% longer to complete assigned tasks when using the AI tools available at the time. Subsequent research suggested that newer tools may offer greater productivity benefits, although measuring those improvements reliably has proved difficult.
The lesson isn’t that AI coding tools are ineffective. Their capabilities are improving rapidly, and their usefulness varies considerably by task.
It’s that the relationship between generating code and delivering productive, dependable software is more complicated than some of the more enthusiastic commentary suggests.
A business might develop an internal application in a matter of days and subsequently spend years maintaining it.
Equally, an application that appears inexpensive to build could create substantial costs through security failures, unreliable integrations or the diversion of valuable technical resources.
This distinction becomes particularly important when considering core enterprise platforms.
According to Gartner’s October 2026 analysis, 51% of surveyed IT leaders were reassessing specialist point solutions, compared with just 22% reconsidering major systems such as ERP and CRM.
That distinction matters.
A relatively simple internal workflow tool is not the same proposition as an enterprise platform supporting financial reporting, regulatory compliance or thousands of employees across multiple countries.
For the foreseeable future, established SaaS vendors will retain substantial advantages in areas where security, reliability, scale, integration and accumulated domain expertise are difficult to reproduce.
But this doesn’t invalidate the build-versus-buy argument.
It makes the decision more nuanced.
Just because an organisation can build something doesn’t mean it should. Equally, just because a SaaS vendor has already built something doesn’t mean the customer should automatically buy it.
And it is precisely this new ambiguity that creates such an interesting commercial challenge.
There is a useful distinction that I think will become increasingly important as businesses reconsider their technology investments.
Some software capabilities are necessary for operating a business. Others have the potential to differentiate it.
Most organisations need systems for managing expenses, processing invoices, recording employee information and maintaining customer data.
But those capabilities rarely represent the reason a customer chooses to do business with them.
A hospitality business, for example, might reasonably question whether developing its own payroll system creates meaningful competitive advantage. Even if AI makes development relatively inexpensive, the organisation would still inherit considerable responsibility for something that is unlikely to differentiate its customer proposition.
Buying established software may remain the obvious decision.
But consider a different application, perhaps one that helps the same hospitality business dynamically allocate inventory, personalise guest experiences or manage a highly specific operational process.
If the business believes a bespoke application could deliver a meaningful competitive advantage, the case for developing it internally becomes more interesting.
This introduces a strategic dimension to the build-versus-buy conversation.
It’s no longer simply a question of which option is cheaper.
Businesses need to consider which capabilities genuinely differentiate them, where they want to retain control of intellectual property and where external expertise provides greater value than internal ownership.
In practice, many organisations will probably adopt a hybrid approach.
They’ll continue purchasing established platforms for complex, business-critical requirements while developing narrower applications, interfaces and workflows around those systems.
That creates both a threat and an opportunity for SaaS companies.
Vendors whose value propositions depend heavily on relatively simple, replicable functionality may find themselves under increasing pressure.
But those providing reliable infrastructure, proprietary data, sophisticated integrations, genuine domain expertise or difficult-to-reproduce business capabilities may become more valuable.
The question is whether SaaS companies can clearly articulate why their particular software belongs in the second category.
And that brings us to sales.
Having spent much of my career in B2B sales and sales enablement, I find the commercial implications of this shift particularly interesting.
For years, SaaS sales teams have been trained to compete against a relatively familiar set of alternatives.
There are competing vendors, existing providers, internal priorities and the ever-present possibility that a customer simply decides to do nothing.
Sales methodologies have evolved around understanding these alternatives, identifying business challenges, quantifying their impact and developing a compelling case for change.
That thinking remains entirely relevant.
But what happens when the customer’s internal technology team becomes another potential provider?
Consider a salesperson working with a prospective customer on a £60,000 annual software subscription.
The customer agrees that the business problem exists. They’ve identified the cost of the current situation, established the desired outcomes and accepted that the proposed solution could help.
The salesperson has conducted a credible discovery process, engaged the relevant stakeholders and developed a reasonable business case.
Then someone from the customer’s technology function joins the discussion.
“We’ve been looking at this internally. We think we could build something ourselves.”
Suddenly, the competitive landscape changes.
The salesperson isn’t competing against another vendor with a recognisable product, a published pricing model or a list of differentiating features.
They’re competing against the customer’s confidence in its own capabilities.
That creates a different challenge.
The internal team may believe it can build a suitable application quickly and inexpensively. It may have a working prototype. It might even be able to demonstrate functionality that closely resembles the vendor’s product.
What it may not have is a fully developed understanding of the ongoing costs, operational risks and long-term responsibilities associated with maintaining that application.
Equally, the vendor may be overestimating the complexity of what the customer actually requires.
Neither position should be accepted without scrutiny.
And this is where the quality of the sales conversation becomes particularly important.
A salesperson who responds by demonstrating another twenty features is unlikely to make much progress.
Neither is someone who dismisses internal development as unrealistic. In many cases, it won’t be.
Instead, the conversation needs to move towards the genuine economics of the two alternatives.
What will the internal solution cost to develop, deploy and maintain? How quickly will it deliver the required outcomes? What happens when requirements change? Who will own it? What other work would the internal development team need to deprioritise? What are the security, compliance and business continuity implications?
And, crucially, what commercial advantage does the purchased solution deliver that the internally developed alternative cannot?
These aren’t questions designed to frighten a customer away from building software.
They’re questions intended to help the customer make a better decision.
That distinction matters.
Because if a customer can genuinely achieve the same commercial outcomes by developing a secure, reliable application internally at a lower total cost, building may well be the right answer.
The role of the salesperson isn’t to win the argument at any cost.
It’s to establish whether buying represents the stronger commercial decision.
The potential implications extend beyond how salespeople handle competitive conversations.
They reach into the way SaaS businesses position and sell their products in the first place.
Historically, vendors have often differentiated themselves through functionality.
One platform offers more comprehensive reporting. Another provides more sophisticated automation. A third has a more intuitive user interface or an additional integration.
These distinctions can be valuable, but their importance depends on how difficult the capabilities are to reproduce and how directly they contribute to the customer’s objectives.
If AI continues to reduce the cost and complexity of developing software, functionality alone may become an increasingly fragile source of competitive advantage.
A feature that once required considerable engineering investment might eventually become something a capable internal team can reproduce in a relatively short period.
This doesn’t mean every feature is easy to replicate, or that the value of a product is simply the sum of its individual capabilities.
It does mean SaaS providers may need to become considerably more disciplined about demonstrating the business outcomes their products deliver.
Not just what the platform does, but what changes for the customer because it exists.
Does it enable the organisation to generate more revenue? Reduce operational costs? Lower risk? Improve customer retention? Accelerate a strategically important initiative? Deliver results that would be difficult, expensive or slower to achieve internally?
And can the vendor substantiate those claims?
These questions have always been central to effective B2B selling.
The difference is that answering them may become considerably more important when customers have a credible alternative to purchasing.
A salesperson who can only explain functionality is vulnerable when functionality becomes easier to reproduce.
A salesperson who understands the customer’s business, can quantify the commercial impact of a problem and can objectively compare different ways of solving it is in a much stronger position.
There is a certain irony here.
As artificial intelligence makes it easier for businesses to develop their own software, it may simultaneously increase the importance of genuinely skilled, consultative selling.
Not because salespeople need to become more persuasive.
But because they need to become more commercially valuable.
For SaaS executives, this creates questions that extend beyond sales methodology.
Product leaders may need to reconsider whether adding features is always the best way to defend their market position. A larger platform isn’t automatically a more valuable one, particularly if customers are paying for functionality they rarely use.
Pricing strategies may also face pressure.
Seat-based subscriptions, premium feature bundles and annual price increases become harder to justify when customers can develop alternatives for specific use cases.
Customer success teams may find themselves having to demonstrate value more consistently throughout the relationship, rather than assuming that implementation and adoption will naturally translate into renewal.
And sales leaders may need to reconsider how their teams are equipped to understand and respond to internal development as a competitive alternative.
This is particularly relevant to account management and expansion.
A customer who has already adopted a platform may remain perfectly happy with its core functionality while becoming increasingly reluctant to purchase additional modules.
The commercial conversation shifts from whether the vendor offers a capability to whether purchasing that capability creates greater value than developing it internally.
That is a considerably more demanding standard.
It also raises a question about how SaaS companies themselves use AI.
If customers can use artificial intelligence to develop software more efficiently, vendors have access to many of the same capabilities.
They may be able to develop new features faster, reduce costs, deliver more tailored customer experiences and improve the economics of their own products.
In other words, AI doesn’t automatically favour the customer.
It changes the capabilities available to both sides.
The companies that adapt successfully may be those that use these advances to deepen their competitive advantages rather than merely increasing the volume of features they release.
I don’t believe we’re witnessing the end of SaaS.
The industry solves too many complicated, commercially important problems for that conclusion to be remotely convincing.
But I do think we’re witnessing the beginning of a meaningful shift in the relationship between software vendors and their customers.
For much of the past twenty years, SaaS companies have benefited from a significant imbalance.
They possessed the expertise, resources and technology required to develop sophisticated applications. Their customers, in most cases, did not.
That imbalance made buying software the logical choice for an enormous range of business requirements.
Artificial intelligence is beginning to narrow the gap.
Not evenly, not universally and certainly not without introducing complications of its own.
But enough to make internal development a credible alternative in situations where it previously wouldn’t have received serious consideration.
For some SaaS businesses, particularly those providing relatively simple and easily reproduced functionality, this could become a substantial competitive threat.
For others, it may provide an opportunity to demonstrate the depth of value that established, well-supported software can deliver.
What seems increasingly difficult to defend is the assumption that customers will continue buying software simply because developing it themselves is too complicated.
That argument is becoming less reliable.
And as it does, the commercial burden on SaaS providers will increase.
They’ll need to demonstrate why their software is worth purchasing, why the ongoing subscription represents good value and why the customer is better served by investing in an external solution than developing an internal alternative.
They’ll need to make that case against competitors who may not have products, sales teams, marketing budgets or even a formal software development function.
Because the biggest competitive threat facing a SaaS company may no longer be another technology vendor developing a superior product.
It may be a customer looking at their annual subscription, opening an AI coding tool and asking a question that would have seemed rather unreasonable just a few years ago.
“Why don’t we just build this ourselves?”
And for an industry built on the assumption that buying software is easier than building it, that may prove to be the most uncomfortable question of all.
Quick Summary• The best sales coaches specialise in different areas, from cold calling and prospecting…
In 1964, the filmmaker Stanley Kubrick wrote to the celebrated British science fiction author Arthur…
Quick Summary• Most negotiation myths have conditions: A tactic can work in one situation and…
Quick Summary• The choice between SPIN Selling vs Challenger is about capability. Challenger needs commercial…
Quick SummaryGap Selling is an approach that focuses on the “gap” between a buyer's current…
In 1891, Arthur Conan Doyle published A Scandal in Bohemia, a short story featuring his…