Build vs Buy: The Hybrid AI Strategy Reshaping CX

M&AAImarket
Written By Faraaz Khan

Since 2023, AI has changed the CX market faster than anything I’ve seen since SaaS went mainstream. Buyers are less patient, roadmaps are moving weekly not quarterly, and a lot of “pre-AI” platforms are now facing the same call:

Do we build the capability in-house, or do we catch up faster through M&A?

In the real world, both extremes typically fail. The teams moving fastest run a hybrid: build what matters, and use targeted M&A to catch up where they’re behind. For many legacy platforms, it’s one of the few practical ways to earn a different multiple. Not by rebranding, but by showing customer pull, better unit economics, and AI that shows up in the workflow.

Why building is harder than it sounds

In my experience, most teams underestimate what it actually takes to build AI that changes customer outcomes. It’s not just “hire a few ML engineers.” It touches everything: data, how fast you can ship, and leadership attention.

The teams that are getting traction usually have the same basics in place:

  • a clean, unified data foundation

  • a product culture that ships quickly

  • leadership that moves with conviction

  • funding to resource AI ahead of revenue

  • the ability to recruit and retain scarce AI talent

The bottlenecks show up immediately. Even strong engineering orgs get stuck on the same challenges: data readiness, governance, and prioritization. Talent is part of it. Across early- and growth-stage companies, demand for ML and AI engineers continues to outstrip supply, and multiple founder surveys from a16z and Air Street Capital show the same: AI engineering is one of the hardest functions to scale. Carta's latest compensation data showed AI and ML engineering salaries rose 5.4% to 9.1% between January 2024 and June 2025, with the fastest increases at younger companies. Even companies with strong engineering teams hit issues with data readiness, governance, and prioritization.

One of the clearest counterexamples has been Intercom. They aligned on AI early, had a unified data model, properly resourced the shift, and had the brand strength to attract talent. They moved quickly enough to build real capability in-house. It was an impressive demonstration of strength of conviction and speed of execution. Most businesses don’t have that combination in place.

Why buying isn’t the magic shortcut

Buying can absolutely compress time-to-market, especially when the bottleneck is team and know-how. But it’s not a cheat code.

The momentum in AI M&A has picked up quickly. Q1 2025 saw AI and ML startups raise a record $73.6 billion across 1,603 deals. This was the highest quarterly total on record by deal value. Globally, AI and ML startup investment increased more than 50% in 2024 to $131.5 billion, accounting for 35.7% of all VC global deal value. In Q1 2025 alone, 57.9% of global VC dollars went to AI and ML startups, with 70.2% of North American deal value flowing to AI/ML companies. Much of this activity is concentrated in smaller, specialized acquisitions under $100M that compress development cycles and focus on acquiring niche teams rather than products or revenue. That tells you where value is flowing: speed and skill, not revenue.

Buying only works if your product can actually absorb what you bought. If your data is fragmented and the product already feels bolted together, you won’t get end-to-end AI. You’ll ship another standalone feature. That’s how suites get heavier without getting smarter. A unified data foundation makes integrations cheaper and faster. When the plumbing is solid, you can bring in a team and ship. When it isn’t, integrations take forever, and you end up with more disconnected features.

At InsiderOne, MindBehind was a good example of the right kind of buy. The team had real conversational AI experience, and because our data layer and AI foundation were already in place, we could integrate tightly and ship, not just announce.

I think of buying as an accelerator, not a substitute for building. If the underlying engine isn’t working, M&A just helps you get to the wrong place faster.

Why a hybrid strategy is becoming the default

For companies that were built pre-AI, two things are becoming clear:

  1. Building everything yourself is usually too slow.

  2. Buying your way to “AI” usually doesn’t create differentiation. You end up with features, not a system.

A hybrid strategy is emerging because, if done correctly, it blends the strengths of both paths. It positions pre-AI SaaS platforms to re-rate into categories where AI-driven peers command meaningfully higher valuation multiples.

1. Build what’s closest to your core

Your competitive advantage is typically a combination of how your data model, workflow engine, and customer insight meet. If it sits in that core, you generally want to build it. That’s the part that improves with use and is hardest for competitors to copy.

Twilio Segment's 2024 report shows unified data platforms outperform fragmented ones across personalization and customer engagement metrics.

2. Buy for acceleration and specialized talent

This is where M&A plays a real role. The surge in AI investment, particularly in smaller transactions, highlights a noticeable increase in deals aimed at acquiring specialized teams.

The high-impact areas tend to be:

  • conversational AI teams

  • workflow-specific agents

  • vertical AI capabilities

  • model ops and evaluation expertise

These deals can save you 18–24 months. You’re saving time, not outsourcing ownership. You still need to integrate it into the product and make it part of the core.

3. Integrate into one coherent platform

Most of the fragmentation in CX comes from weak integration. ChiefMartec's 2025 marketing technology landscape now includes 15,384 solutions, up 9% from the previous year's 14,106, where only a small minority reach true scale. Fragmentation is the rule, not the exception.

Customers don’t care how you got there. They care whether it:

  • works across their workflows

  • uses their data in a unified way

  • drives outcomes without extra work

This is where internal AI and acquired talent reinforce one another. When integration is real, the product feels simpler and smarter. When it isn’t, you’ve just added another place to click.

A simple framework for build vs buy decisions

Here’s the decision lens I’ve found most useful:

Step 1: Map the impact on your core

High impact on your core data model or workflow engine → Build
Low impact on the core engine → Buy or Partner

Step 2: Map the time sensitivity

Need it inside 12 months? → Buy
Need it in 12–24 months? → Either
Need it beyond 24 months? → Build

In simplest terms: Build the core. Buy the accelerators.

What seems likely in the next 12–24 months

Looking at what teams are actually doing (not just saying), a few patterns are showing up.

1. The strongest brands will widen the gap

Companies that already had strong operating discipline, like Intercom, are pulling further ahead.

2. Talent-heavy M&A will increase

PitchBook already shows the trend accelerating. These are “team buys,” not revenue or product plays.

3. Pricing power will shift toward AI-native platforms

Gartner notes that AI ROI drops sharply when the underlying data is fragmented.

4. Data architecture will become the constraint

Teams with fragmented data will struggle no matter how good the UI looks. AI ROI is increasingly determined by the quality of the underlying data infrastructure, not model choice.

5. CX suites will reorganize around agents

Not chatbots and templates, but workflow-specific agents embedded across the lifecycle. The winners will be the teams that invest early in clean data architecture and workflow orchestration.

If you’re building, buying, or operating in this space, I’d love to compare notes.

You can reach me at faraaz@inorganicedge.com or on LinkedIn.

Sources

AI Talent, Hiring & Compensation

AI Adoption & Market Acceleration

AI M&A Trends

Data Fragmentation & Personalization

AI Productivity & Development Cycle Time

AI Integration & Organizational Readiness

Faraaz Khan
Previous

How to Re-Rate Your Pre-AI SaaS Assets into AI Valuations