Replacing a SaaS with a custom build one thanks AI. Yes or no?
Why are companies considering this option?
I have been reading about The end of SaaS more often, and the next post just landed into my inbox.
It refers to the Klaran’s CEO declarations that they are replacing some SaaS, Salesforce and Workday, with custom build solutions thanks to AI.
Klarna's (KLAR) CEO Sebastian Siemiatkowski said that the company is shutting down its software as a service provider (SaaS) Salesforce (NYSE:CRM) and within a few weeks will shut down Workday (NASDAQ:WDAY).
There are large ongoing internal initiatives that are a combination of AI, standardization, and simplification. As an example, we just shut down Salesforce. Within a few weeks, we will shut down Workday. We are shutting down a lot of our SaaS providers, as we are able to consolidate," said Siemiatkowski on a conference call.
Siemiatkowski added that with the help of AI, the company is able to standardize and create a more lightweight tech stack to operate more effectively with higher quality.
When I read these claims, I am skeptical at first, but it is worth exploring in which condition this can be an option.
I have been reflecting about how AI impacts organizations, you can see it in the following link.
In the current post, I will explore two key areas:
Budget and laws. How the laws and budget distribution will change how what you build and what you buy.
Value chain. What to start custom-building, and what to keep externalizing.
It’s about the budget and laws
Let’s simplify the context by assuming we have 3 main budgets; people, infra, and SaaS. By adopting AI, the organization become a 20% more productive.
In this scenario, we might want to either:
Increase our efforts into new business initiatives.
Become more efficient and spend less.
What it might happen is that the market is not expanding, quite the contrary, it is contracting. Therefore, you might want to reduce the spend, which it is the only thing a company is in control of. So, you have to reduce the spend in some budgets.
After reducing contracts with consultancies, and external people, we have seen several layoffs already happening in the industry, and more in countries that their laws are more friendly to those actions.
But, what happens in those countries that firing isn’t that easy, and they already reduced the external contractors for example? Senior leadership has to look at some other budget to reduce. In this case, external expenses for products like SaaS.
Either you create value or reduce cost
So, you have people that are more productive, you have the creation of value at full speed, and you don’t know where to keep investing. The ROI for new initiatives aren’t matching the expectation.
Then, you realize that you have a couple of bills a year that are above a 1M EUR, and they might be great to just spend less on those services.
So, you decide to start investing to reduce the cost on external services, and achieve a better impact per person.
Value chain
A company have a lot of users they serve, but let’s classify them into two main groups.
External. Those that you build the product for.
Internal. The people that helps you build the product. Developers, HR, Product, ...
When you consider where to invest in the initiative of reducing costs, you want to impact the customer the less you can. So, you can prove the initiative can be successful and then replicate the success to other riskier areas.
That’s why you start looking for replacing external SaaS, you look for those that helps your internal users (employees) do their work. Plus, you don’t want to impact heavily your people, that’s why you also look for operational work that wouldn’t impact your customers.
So, you find for SaaS that you can have some redundancy, and won’t impact the business.
What to replace first
The SaaS that you want to replace are the commodity ones for the next reasons:
They are well understood. Therefore, the Copilot, LLMs, and other services have data to support the tool creation/migration.
In case of needing to learn about a new domain, data is available.
Employees are their users, and they could understand their needs, and what’s more important based on the internal user flows, and data on how they are using the external systems.
Yet, nothing gives you the certainty that you will be able to replace a commoditized product with a custom one. A lot of companies have shown different failures on those initiatives already. Yet, worth to try with AI now and learn from the new (?) mistakes.
Is it a good decision to replace a SaaS if it isn’t core to your business?
My intuition says no.
You might have a momentary lower cost for certain features. But either you achieve feature parity for the use cases you require to keep business as usual, and you keep up on the constant evolution of that space, or you will lack behind competition.
By creating a custom build solution, you have a lot of consequences:
Custom-building subcomponents that you need to maintain.
Adopting new SaaS for those features that aren’t critical to the tool per se, repeating the initial scenario that you started all the initiative with.
You might decide to reuse some product components for this new internal tool. That will add pressure for Platform Teams for example, forcing them to understand and prioritize internal customers (devs) with two competing backlogs (product and internal supporting tools), by adding more work to them.
My take is that those initiatives will be something that large companies will try, and we will see the results in the near-mid term.
Let’s hope they share the learnings after the initiative succeeds or fails.
My expectation on those initiatives
My expectation is that the time to first use case will be lowered.
The time to a CRUD operation will provide a false sense of what a team can accomplish by themselves, when teams have to lead with complicated and edge use cases, I think companies will roll back to previous SaaS solutions.
You can see some points I did on AI and domain complexity here.
When they don’t roll back, pressure to deliver those supporting tools without an enough quality will raise, creating technical debt, lack of good practices, and making the progress for those critical tooling for organizational operations less and less effective.
At some point, for semi-working internal tools, either:
Someone will make the decision to go back to SaaS, losing the investment.
People will keep investing on internal tooling, which will cause a bad experience for internal users, like Sales teams, and burnout teams that are unable to meet internal expectations.
P.S. I’m sure that at some point, those initiatives will be tried to be commercialized as a SaaS by either the same company, or by a spinoff founded by the leading team.
What to do instead?
Because the cost of producing SaaS has been impacted by AI, I would look for more cost-efficient products, or those who have pricing alternatives to per seat, for example. Those products/companies are appearing, and it is worth to consider a migration of tooling instead of custom-building one.
With some NoCode applications, some nice integrations with products like Make, AppSmith, and Claude, you could leverage other SaaS, and add the productive boost for your scenario without replacing the whole value chain.