# Composability will define the next SaaS era ## Current Form The SaaS ecosystem is folding in on itself. As foundational tooling and AI agents become more accessible, **monetizable value is abstracting toward composability**—tools that enable users to manage, move, and manipulate data into custom-built outcomes. In this future, **apps become raw material**—not end products. Users define protocols, deploy agents, and compose bespoke workflows and software locally. The era of monolithic SaaS is giving way to decentralized, interoperable, and agent-driven experiences. The “interface” becomes ephemeral, ephemeralized by agent-native UX that adapts to user intent in real time. This raises a powerful tension: the more **personalized** and **fluid** the user interface becomes, the more **stable and constrained** the underlying components must be. **Agent-native UX depends on protocol-layer thinking**—shared schemas, defined affordances, and interoperable primitives. Without these constraints, agents have nothing to compose. You can define the primary flavors of defensible IP—the things that can be composed—through three buckets: data, protocol, and [[community as IP|community]]. In this kind of environment I think some seismic shifts start to happen, namely, consolidation in regulated spaces; explosion of DIY composable stacks elsewhere. Big fish can find safe harbors where there is heavy privacy, IP protection, or regulatory pressure, particularly at the enterprise level. Even still, I think we'll see further consolidation. A large majority of pure service applications —interfaces overlaid on databases—will be rendered obsolete as AI technology moves to the device level. [[Benoît De Nayer]] talks about these as "terminals" in his theory of "DeepSaaS." In this way the monolithic "app experience" fractures into a million tiny, highly unique and individual experiences composed from raw materials, which will have huge knock-on effects to the pillars of SaaS: large network-effect userbases, and recurring revenue. Similarly, the basic building blocks of SaaS investment (recurring revenue metrics and multiples) are likely to decompose. [[Kyle Poyar]] has been tracking evolving pricing models in great detail, and we're seeing a shift toward outcome and success-based pricing. At an enterprise level, you can hybridize recurring and non-recurring dollars, but I see more downward pressure applied here in many traditional SaaS use cases. An inevitable march toward non-recurring engagements will challenge SaaS teams and investors in ways they haven't yet imagined. For reference, it took Snowflake (a company that deals exclusively in complex data management, processing and storage) *multiple years* to accurately forecast revenue behavior after introducing [[Snowflake’s Forecast Transformation With Brad Floering - Clips|consumption-based pricing levers]]. As an example of this in action, tools like [Scrapeli.io](https://scrapeli.io) show where we’re headed: they extract raw data and let users _compose outcomes_ using agents, no-code platforms, and local automation. Services like Apollo, Clearbit, and ZoomInfo will struggle to succeed in this kind of composable outcome world, **who would pay for *access to* to a walled garden if they could simply scrape it as needed?** All of this underwrites the predominant value proposition pattern of "[X] platform that enables you to [Y]" and points to some more emergent thinking in [[consumption-based pricing and changing value roadmaps]]. #### Some Key Concepts I'm Exploring - **Composability**: The ability to assemble software functionality from interoperable components—scrapers, agents, protocols—rather than relying on rigid, vertically integrated tools. - **Protocol over Product**: What matters isn’t the app itself, but the standard or method for accessing, organizing, and reusing data. - **User as System Integrator**: In this paradigm, the user becomes the architect of their digital environment, assembling apps and automations from composable primitives. - **Data Autonomy**: The real value isn’t "being healthy" or "getting insights"—it’s owning, shaping, and embedding your own data for custom utility. - **Service App Death Spiral**: Apps that offer generic services or non-proprietary data will die off quickly as agents + scraping + AI reduce costs to near-zero. ### Spark I first started thinking about this when I heard [[Satya Nadella]] make the infamous "SaaS is dead" proclamation on BG2. ![[Satya Nadella BG2 W Bill Gurley & Brad Gerstner - Clips#^02229e]] I started to think about it a little more critically when I read [[resources/media/Full Document Contents/Articles/Benoît De Nayer’s Post re composability|Benoît De Nayer’s Post re composability]] on "DeepSaaS." All of it sort of points to the collapse of the user-interface, and if the majority of software development is UI layered on top of database, that introduces the obvious places where moats will dry up. The concept of [[composability]] takes a more forward-looking approach and tries to model where to go if you want to win in the next era. ### Tensions or Unknowns Still very prognostic-y. Not particularly unique (see Satya & Benoit), and too high-level to feel like a defensible, fully-formed opinion. - What gets lost when everything becomes composable? Are there tradeoffs in meaning, coherence, or friction that we actually need? - Is there a tension between user-level fluidity (agent-assembled experiences) and system-level rigidity (protocol constraints)? What happens when these layers fall out of sync? ### Next Evolution As an idea it's been a useful heuristic for first principles thinking about things I think are worth working on and which aren't, which has been a boon for [[Greenhaus]] clients, particularly the early/seed stage orgs. I'd like to continue to follow this thread a little farther before taking a stand, in particular I'm interested in continued development in enterprise pricing models and how AI-native companies are building value propositions and messaging. ### Wayfinding %% DATAVIEW_PUBLISHER: start ```dataview TABLE WITHOUT ID questions AS "Responds to", origins AS "Informed by", ideas AS "Developed alongside", concepts AS "Builds on" WHERE file.name = this.file.name ``` %% | Responds to | Informed by | Developed alongside | Builds on | | ----------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | <ul><li>[[thinking/questions/the future of saas.md\|the future of saas]]</li></ul> | <ul><li>[[resources/clippings/Benoît De Nayer’s Post re composability - Clips.md\|Benoît De Nayer’s Post re composability - Clips]]</li><li>[[resources/clippings/Satya Nadella BG2 W Bill Gurley & Brad Gerstner - Clips.md\|Satya Nadella BG2 W Bill Gurley & Brad Gerstner - Clips]]</li></ul> | <ul><li>[[thinking/saplings/consumption-based pricing and changing value roadmaps.md\|consumption-based pricing and changing value roadmaps]]</li><li>[[thinking/seeds/community as IP.md\|community as IP]]</li></ul> | <ul><li>[[thinking/concepts/emergence.md\|emergence]]</li><li>[[thinking/concepts/hyperobjects.md\|hyperobjects]]</li><li>[[composability\|composability]]</li></ul> | %% DATAVIEW_PUBLISHER: end %% > [!info]- Changelog > - [[2025-04-16-Wednesday]] - fleshed out some forward thinking for next evolutions and cleaned up some loose ends in my current thinking.