DemandRamp · field notes · vol. 01

The Progression.

How technology categories grow up — and where AI is right now.

Every major software category follows the same arc. It starts with pro services. It matures into packaged products. It lands in a market where buyers can build, buy, or do both — pragmatically, based on what the problem actually calls for.

We've watched it happen twice. We're watching it happen again.

The Internet

Phase 1

Custom builds.

In the early days of the web, everything was bespoke. A website cost four or five million dollars. Every e-commerce experience was hand-rolled. The market was services-led because nothing else existed.

Phase 2

Productization.

Then came the packaged layer. CMSs. E-commerce platforms. Frameworks. The space gelled. What used to take a six-month custom build became a configuration exercise.

Phase 3

Build vs. buy.

The market matured into real optionality. Want a custom application? Build it. Want to move faster? Buy a product and configure it. Want both? Hybrid. The category stabilized once buyers had a mature menu of choices.

The pattern: services-led → packaged → buyer optionality.

Voice Recognition

Phase 1

Pro services.

Nuance and Speechworks pioneered the category with massive custom engagements for Fortune 500 call centers — American Airlines and others. Long, expensive, bespoke per customer. The technology was real, but bringing it to market meant building it from scratch every time.

Phase 2

Packaged applications.

Then came the packaging layer. Vocent — a company our founder co-founded — built a packaged voice authentication application that made voice biometrics viable for the enterprise. Configurable. Integration-ready. Not a custom build.

Phase 3

Productized integration.

Voice became something you could buy, configure, and integrate. The market matured because the ecosystem matured around it.

Same arc as the Internet. Different category.

AIWe are here.

Phase 1

Pro services. (We are here.)

The AI services market is exactly where voice was twenty-five years ago and where the Internet was thirty years ago. Pro services-led. OpenAI and Anthropic are aggressively building embedded enterprise services teams to bring AI to market through bespoke engagements at Fortune 500 companies. Seven-figure, eight-figure deals. Embedded experts. Custom everything.

This is the right starting point for a new category. It's also where every prior category got stuck before it grew up.

Phase 2

Packaging. (Just beginning.)

The packaged-product layer for AI is still work in progress. Tools companies are emerging. Frameworks are forming. But there's a real friction in the current market: the dominant rhetoric is custom, custom, custom — and custom-only is a posture that every prior category has proven doesn't scale.

If the model providers underinvest in the packaging ecosystem — the developers, the services partners, the packaged-product builders — the market stays chaotic instead of maturing. Custom alone has never been the path to a healthy category.

Phase 3

Build vs. buy.

It's coming. The packaged AI application layer will grow. Buyers will eventually have real optionality. The question isn't whether AI follows the same arc as the Internet and voice. It's how fast we get there, and who helps the market get there pragmatically.

the gap

Where DemandRamp fits.

The gap right now is between AI being exciting and AI being real for an actual business.

We don't take an ideological stance on custom-vs-packaged. We take a pragmatic one. Every project gets measured on its merits:

Is this genuinely a custom application? Then we build it, with the discipline of a software shop — discovery, definition, design, development, rollout.
Is there a packaged product that fits, configured and customized to the customer's environment? Then we recommend it and integrate it.
Is the right answer a hybrid? Most of the time, yes. That's how mature markets actually work.

This is the same posture we've taken across fifteen years of marketing operations work. We've seen what happens when categories grow up. We've helped customers navigate the transition before. And we've been on the inside of the packaging phase of a category that grew up exactly this way.

We're not waiting for the AI market to mature. We're helping our customers operate inside it right now — pragmatically, with the experience of two prior category evolutions and the operational backbone that makes new technology actually work in a real business.

That's the gap. That's where we fit. That's why we built the AI services practice we built.

what this means

For you.

We're not competing with the model providers' pro services teams. They're going after the Fortune 500 with seven- and eight-figure embedded engagements. We're bringing that same embedded-services depth to the mid-market — the Fortune 3000, companies with 200 to 1,500 employees who need this work done now and don't want to wait their turn.

But the size of the customer isn't the only thing that's different. So is the agenda.

We're LLM-independent.

We don't have a commercial incentive to put everything on a particular model, or to recommend custom-on-our-stack when a packaged product would do the job better. The model providers have an obvious bias — every custom engagement is also a model commitment. We don't carry that bias.

We bring the wisdom of two prior category evolutions.

We watched the Internet grow up. We helped build the packaging phase of voice recognition. We know what custom looks like, what packaged looks like, and how mature markets shake out. That experience shapes every recommendation we make.

We pick the right solution for the project, not the one that pays us best.

Custom when custom is right. Packaged-and-configured when that's the smarter call. Hybrid when the problem demands it. Our job is to be honest brokers — and that's what fifteen years of all-referral business is built on.

You get the embedded-services experience the Fortune 500 is getting from the model providers — at mid-market scale, on mid-market timelines, with a partner whose only agenda is making AI real in your business.

start the conversation

Want to talk about where AI fits in your operation?

Book a strategy call →