AI does not become business value when a prompt is written.
It becomes business value when a proven AI capability reaches the right person or system, with the right context, permissions, version, safeguards and measurement.
The market already knows how to create prompts. What it has not yet mastered is how to distribute operational intelligence reliably.
That is the market we intend to build.
The market signal
Capital formation around AI is substantial, but deployment remains constrained. Global private investment in AI reached approximately $344.7 billion in 2025, with $170.9 billion going to generative-AI companies, according to the Stanford AI Index 2026, chapter 4. These are investment figures—not enterprise spending or our addressable market. They demonstrate the scale of ecosystem formation, not the size of our category.
Adoption, however, is not the same as transformation. In McKinsey’s 2025 survey, 88% of respondents reported regular AI use in at least one function, yet only about one-third said their organizations were scaling AI. Just 39% reported any enterprise-level EBIT impact, and most of those attributed less than 5% of EBIT to AI. McKinsey, State of AI 2025
The broader economy shows an even earlier market. The U.S. Census found that only 18% of employer firms used AI in a business function between November 2025 and January 2026—32% when weighted by employment. Among adopters, 57% used AI in no more than three functions. U.S. Census Business Trends and Outlook Survey
The McKinsey and Census numbers are not contradictory: they measure different populations. Our inference is that they reveal a two-speed market—intensive experimentation among leading organizations and shallow penetration across the wider economy.
Brazil presents the same opportunity. In 2025, 17% of Brazilian companies with more than ten employees reported using AI, rising to 50% among large companies. Among AI adopters, 80% acquired ready-to-use systems and 60% hired external suppliers to develop or adapt solutions. Cetic.br, TIC Empresas 2025
Our view: companies do not only want to invent AI capabilities internally. They need trusted capabilities they can acquire, adapt and deploy.
Our market theses
Thesis 01
AI creation is becoming abundant; reliable distribution remains scarce
GitHub reports that more than 1.13 million public repositories imported an LLM SDK by August 2025, representing 178% year-over-year growth. GitHub Octoverse 2025
This measure covers public repositories rather than the entire software economy. Even with that boundary, the signal is clear: the supply of AI applications, agents, prompts and tools is expanding quickly.
Our thesis: the next constraint is discovery, qualification, integration, permissioning and reuse.
Thesis 02
Adoption has outrun trust
In the Stack Overflow Developer Survey 2025, 84% of respondents were using or planning to use AI development tools. Yet 46% distrusted the accuracy of AI output, compared with only 33% who trusted it. Sixty-six percent cited outputs that were “almost right” as a major frustration.
The market does not simply need more AI output. It needs capabilities with evidence:
- What has been tested?
- For which tasks?
- With which models?
- Against which evaluation criteria?
- Under which policies?
- Who owns and supports it?
- What happens when it fails?
Trust cannot remain an informal promise. It must travel with the capability.
Thesis 03
A prompt is not the final product
AWS provides native prompt storage, testing and versioning. Google provides prompt storage, versioning and restoration. Microsoft explicitly recommends managing prompts as versioned code assets. AWS Bedrock Prompt Management · Google Prompt Management · Microsoft GenAIOps
OpenAI’s product evolution offers an even stronger signal: reusable API prompt objects were deprecated in June 2026, with shutdown scheduled for November 30, 2026. Its migration guidance treats production prompts as application code requiring review, testing, deployment and versioning. OpenAI deprecation notice · OpenAI migration guide
Our thesis: raw prompt storage is not a defensible business category.
The durable distributable unit is a certified AI capability containing:
- Instructions and structured inputs
- Model compatibility
- Context and knowledge sources
- Tools and integrations
- Evaluation results
- Safety and usage policies
- Ownership and licensing
- Version and rollback information
- Deployment interface
- Usage and outcome telemetry
“Prompt Distribution” is our market entry point. “AI Capability Distribution” is the long-term category.
Thesis 04
Governance must move with the capability
The NIST Generative AI Risk Management Profile recommends inventories, provenance, version control, supplier assessment, usage rights, monitoring, red-teaming, fallbacks and incident response.
The EU AI Act, generally applicable since August 2, 2026 with phased exceptions, is making transparency and value-chain accountability increasingly concrete.
A Prompt Distribution business unit is not automatically a legal “distributor” under the AI Act, and individual prompts are not automatically AI systems. Classification depends on the concrete role in the value chain and requires appropriate legal analysis. The strategic direction is nevertheless clear: AI capabilities cannot circulate without responsibility, documentation and control.
Governance must be part of distribution—not an audit performed after deployment.
Thesis 05
Interoperability enables movement, but does not create trust
MCP standardizes how applications discover and retrieve prompts and other capabilities. Its official Registry catalogs MCP servers and anticipates public server marketplaces and private enterprise subregistries. MCP prompt specification · MCP Registry preview
But MCP does not standardize prompt-level licensing, pricing, certification, ownership or quality scores.
Our thesis: protocols can transport a capability. Our business must explain whether that capability deserves to be used.
Thesis 06
The winning metric is not prompts published
A library can accumulate thousands of prompts while generating little business value.
Our North Star
Verified business outcomes successfully delivered through reusable AI capabilities.
We will measure:
- Time from approval to first production use
- Reuse across teams, channels and applications
- Accepted-output rate
- Human correction and escalation rate
- Cost per accepted outcome
- Model portability
- Rollbacks and incidents
- Active capability consumers
- Revenue and retention per capability
Distribution without outcome telemetry is publishing. Distribution with telemetry becomes an improvement system.
Our market position
Why
To transform fragmented AI knowledge into reusable, governed and measurable organizational intelligence.
How
By certifying, packaging, versioning, permissioning, delivering and continuously evaluating AI capabilities across models, applications and business channels.
What
A vendor-neutral distribution platform composed of:
- A private capability registry
- Evaluation and certification pipelines
- Versioning, approval and rollback workflows
- APIs, agent, MCP and embedded delivery channels
- Access, policy and licensing controls
- Usage, cost and outcome analytics
- Curated vertical capability collections
- A controlled marketplace for approved external capabilities
How we enter the market
We will begin with private enterprise distribution, where the pain is most concrete: duplicated work, uncontrolled prompts, inconsistent quality, shadow AI, weak traceability and slow movement from experiments into production.
Next, we will create certified vertical capability packs for repeatable workflows.
Only after establishing quality, demand and outcome evidence will we expand into an open or partner marketplace.
The defensible advantage will not be the number of prompts stored. It will be the combination of evaluation data, domain reputation, workflow integrations, cross-model compatibility, governance and outcome history.
The truth clause
We will not claim that all generative-AI spending is our addressable market.
For example, Gartner forecast approximately $644 billion in worldwide generative-AI spending for 2025, but roughly 80% of that forecast was hardware. It cannot credibly be presented as a prompt-software TAM. Gartner forecast
Our market must be sized bottom-up: target organizations × relevant workflows × annual platform value × distribution volume.
The evidence supports the existence of the problem. Customer adoption and economics must prove the category.
Our declaration
We believe prompts are becoming executable business intent.
We believe executable business intent must be tested, governed, versioned and observable.
We believe intelligence trapped in individual conversations creates no durable organizational advantage.
We believe the future belongs to companies capable of moving trusted intelligence across people, systems, models and markets.
We are not building a warehouse of clever sentences.
We are building the distribution infrastructure through which operational intelligence travels.