Satya Nadella ended his 2024 preaching about how AI agents will take over SaaS.

This validation from one of the biggest gen AI investors means a lot for a thesis Insignia Business Review has been covering the whole 2024. Check out the following deep dives on AI agents from August through September last year:

1. Autonomous AI Agents introduction
2. AgentOps enabling the Autonomous AI Agent workflow
3. Autonomous AI as Apps of the Future
4. Navigating the Long Road to Autonomous AI Agents in Southeast Asia

Insignia Business Review also covered the opportunity for generative AI adoption in enterprise specifically, through this playbook-type article on how startups can approach gen AI adoption for enterprise in Southeast Asia, and more recently this commentary on the shifts in enterprise spending and investment into gen AI tooling.

I scraped through the latest F24 Y Combinator batch and I found out that out of the 93 companies in the batch, 80 are tagged AI/ML. I went through each of these 80 companies and around 45 companies are either Agentic AI or related to Agentic AI.

The most popular verticals that these agentic AI companies tackle are Sales/GTM, compliance (marketing, bug finding), and risk analysis (fintech).

A significant pipeline is emerging from Silicon Valley and the US of agentic AI companies, but what does this really mean for Southeast Asia?

Agentic AI vs Generative AI

But first, let’s align on definitions: what is agentic AI and how is it different from Generative AI?

In simple terms, generative AI is focused on creating while agentic AI is focused on doing.

In an article written by Deon Nicholas, he mentions that most AI agents today are, “dressed-up” and are actually just extensions of Generative AI called retrieval-augmented-generation (RAG). “RAG enhances chatbot responses by pulling from external sources beyond it’s LLM… but RAG is limited to answering questions.” 

On the other hand, agentic AI requires a multi-step decision system that identifies the problem, decides whether the inputs are enough, how best to solve it, and how to know if the problem is solved or not.

How agentic AI overcomes the product/technical limitations of more traditional SaaS implementation

This emergence of agentic AI was discussed in a video by the four partners at Y Combinator with which they even drew parallels to the history of SaaS.

They brought up the following points:

• Every SaaS unicorn, there can be an agentic AI counterpart
  • Before the emergence of these SaaS unicorns, there was the same box-software solution that was doing the same thing. This might also be the case for SaaS and its agentic AI counterparts.
  • When you think about the history of SaaS, the consumer applications worked first such as email and chat which made it much easier for adoption.
• The case of why agentic AI can get bigger
  • SaaS, you still need a set of people to get the workflows done. With agentic AI, not only will you be able to replace the software, you will also be able to reduce the manpower associated with running the software.
• There will be a lot of friction when it comes to adoption
  • The sales of these agentic solutions cannot be sold to company departments unlike SaaS solutions because these AI agents will ultimately replace most of them. Hence, adoption will be very much top-down.

AI agents will take over SaaS BUT what does this really mean for Southeast Asia?

As exciting as this seems, there’s a big BUT (as usual) when taking the agentic AI development into the context of Southeast Asia.

SaaS companies in the region have traditionally faced customer acquisition and revenue growth challenges in markets apart from Singapore. This has led to a number of SEA-based SaaS companies to either find early exits in global acquirers or rapidly expanding their target customer pool to tier 1 markets in east Asia and even across the Pacific to balance out the margins of operating in Southeast Asia.

In the same YC cohort that has been dominated by agentic AI, there was only one company from Southeast Asia (not classified as AI/ML).

While agentic AI can reduce implementation costs of software, there are other factors that play into driving adoption of AI agents in Southeast Asia’s emerging markets (assuming the product and technical risks are covered already).

Founders venturing into this space must ask themselves the following questions:

Why would it make sense for an agentic AI solution to come from SEA?

A key consideration for agentic AI startups in SEA is how to demonstrate big, tangible outcomes for startups originating from the region, rather than being overshadowed by their US counterparts much like the challenge of SEA SaaS startups.

Hence, having a moat around proprietary and localized data is key for these types of startups to thrive in this type of environment.

Will AI agents make sense economically, especially in the SEA landscape?

Labor is very cheap in the region and businesses, particularly SMEs, are highly cost-conscious. While SaaS adoption already faces challenges due to affordability, agentic AI could be perceived as too costly relative to the alternative which is the labor that these businesses already employ.

Founders must be able to show clear ROI both in the short-run and in the long run for businesses to make the shift to AI agents.

Are they ready to face cultural pushback?

In the US, businesses prioritize automation to reduce costs and reallocate human capital to higher-value tasks. This is supported by a general culture of risk-taking and strong investment in upskilling the workforce.

In Southeast Asia, where economies rely heavily on labor-intensive roles, automation is often viewed as a threat to job security, leading to bottom-up resistance and slower adoption. Couple this with the varying cultures, regulations, and market behavior across SEA and you get a landscape that is very challenging to go to market in. Mirroring the insights expressed by the partners at Y Combinator, the importance of going top-down is much more emphasized in this type of environment.

Despite these challenges, we see startups like WIZ.AI emerge as leaders, spearheading the innovation of AI Agents in the region. By efficiently automating complex workflows and effectively reducing the need for human intervention, the conversational agents of WIZ.AI can take the business function of customer service to the next level.

As businesses seek to replace manual, labor-intensive processes with intelligent agents, WIZ.AI positions itself as a key player in this transformative wave, offering scalable, agentic solutions that deliver measurable results.

Learn more how WIZ.AI views AI agents through their CRO Alex Song in this “AI Transformation Panel“: