Composable Enterprise AI Building Blocks
Combine reusable AI primitives into role-specific solutions
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 97%
This product-design framework treats enterprise AI capabilities as composable Lego blocks. Horizontal primitives may include knowledge ingestion, prompting or output augmentation, and the form factor of the resulting work product. Customers, however, rarely want a box of generic pieces; they want a recognizable solution for content generation, software engineering, sales, finance, or another concrete job. Product teams therefore combine reusable primitives into role-specific starting points and add specialized pieces, such as a GitHub integration for engineers, when generic components are insufficient. Direct customer feedback then determines which templates, integrations, and bundles deserve priority. The result preserves broad platform flexibility while reducing the cold-start problem for buyers who need to understand immediately how the product applies to their work.
Origin
Extracted from Marketing Against the Grain, where Scott White compared Claude's horizontal capabilities with Lego pieces that enterprises assemble into recognizable kits.
Core principles
- 01Broad use cases often share foundational primitives
- 02Enterprises adopt complete solutions rather than abstract flexibility
- 03Role-specific components supplement horizontal capabilities
- 04Customer feedback determines which bundles deserve investment
- 05Composable foundations make specialization economical
How to run it
- 1
Map customer outcomes
Collect the concrete jobs enterprise customers want to complete and group use cases with similar underlying shapes.
Pro tip Describe each use case in terms of inputs, transformation, and output.
Watch out Do not organize solely by job title when several roles perform the same information workflow.
- 2
Extract horizontal primitives
Identify reusable capabilities shared across the groups, such as knowledge ingestion, analysis, prompt augmentation, or artifact creation.
Pro tip Prefer components that can serve several validated use cases.
Watch out An overly generic primitive may be technically reusable but operationally meaningless.
- 3
Identify missing specialist pieces
Determine which roles require domain-specific integrations, permissions, formats, or workflows beyond the horizontal foundation.
Pro tip Look for components that are necessary but insufficient when used alone.
Watch out Do not force a generic connector into a workflow that requires deep domain context.
- 4
Assemble recognizable kits
Bundle the primitives and specialist pieces into starting solutions that map to customer language and desired outcomes.
Pro tip Provide templates that let users reach a first useful result quickly.
Watch out Offering raw flexibility without a starting point intensifies the cold-start problem.
- 5
Validate through direct feedback
Observe how enterprise customers use each bundle and where they become blocked. Use those signals to revise templates, integrations, and component boundaries.
Pro tip Maintain short feedback channels with active customers.
Watch out Do not prioritize bundles based only on internal speculation.
- 6
Promote proven components
When a specialist component proves useful across multiple kits, elevate it into the horizontal foundation and simplify future bundles.
Pro tip Track reuse and outcome improvement rather than feature adoption alone.
Watch out Prematurely generalizing a niche component can burden the entire platform.
In the wild
A generic document repository gives Claude organizational context but is insufficient for code work. The product team combines knowledge ingestion with a GitHub integration and engineering-specific starting workflows, producing a kit that engineers immediately recognize as relevant.
→ Horizontal context capabilities become a practical software-engineering solution.
A team combines style guidance, audience personas, product documentation, channel templates, and internationalization into a content-generation starting point. Each element remains reusable, but the bundle presents a complete outcome to marketers.
→ Marketers can begin with a concrete workflow instead of assembling every component themselves.
Common mistakes
Selling only the primitives
Customers may understand that the platform is flexible but still fail to see how to accomplish a specific job.
Building every vertical separately
Duplicating common capabilities across role-specific products wastes effort and makes the platform harder to evolve.
Guessing the priority bundles
Without close customer feedback, teams can invest in specialized combinations that do not solve urgent enterprise needs.
Is it for you?
Best for
AI product teams serving multiple enterprise roles, workflows, and information environments.
Not ideal for
Narrow products with one stable workflow and little need for reusable platform components.
From the transcript
“ultimately a lot of them share some fundamental building blocks and you can sort of think about this as like we are building a Lego…”
“enterprises though are ultimately looking for the Millennium Falcon”
“making sure that we're building the right Primitives that we think are horizontally applicable that can be applied to a lot of use cases but…”
From the episode
The Ultimate Guide to Using Claude AI for Work