CRP-SPEC-041: Adoption & Ecosystem - Time-to-Value, Integrations, and Growth¶

Document: CRP-SPEC-041
Title: Context Relay Protocol (CRP) - Adoption & Ecosystem: Framework Adapters, Onboarding, Templates, and the Growth Engine
Version: 1.0.0
Status: Foundational - Adoption-Critical
Author: Constantinos Vidiniotis, AutoCyber AI Pty Ltd
Date: 2026-06-01
License: CC BY 4.0
Prerequisites: CRP-SPEC-032, CRP-SPEC-040

Abstract¶

A protocol's quality determines whether it deserves adoption; its ecosystem determines whether it gets it. CRP has the engine (40 specs) and the steering wheel (the SDK, SPEC-032), but adoption depends on factors no protocol spec has yet addressed: how a developer gets to value in the first five minutes, whether CRP works inside the frameworks they already use (LangChain, LlamaIndex, the Vercel AI SDK), whether there are templates and examples to start from, and how free usage converts to paid. This document specifies the adoption layer: framework adapters that make CRP one import inside existing stacks, a concrete onboarding flow with a quickstart that delivers value in minutes, a template/example gallery, the free-to-paid conversion mechanics, and the developer-facing growth surfaces (docs, showcase, community). The principle: meet developers exactly where they are, get them to a working governed-and-grounded result before they have to learn anything, and make the path from first call to production frictionless.

1. Framework Adapters - CRP Where Developers Already Are¶

Most developers building AI are already inside a framework. CRP must be one import in those frameworks, not a migration away from them.

1.1 The Adapters¶

Framework	Integration	Usage
OpenAI SDK	drop-in base_url (SPEC-032 §2)	change one line
Anthropic SDK	drop-in base_url	change one line
LangChain	`CRPChatModel` drop-in for `ChatOpenAI`	`llm = CRPChatModel(model="gpt-4o")`
LlamaIndex	CRP as the LLM + retriever	`Settings.llm = CRPLlamaLLM()`
Vercel AI SDK	CRP provider	`crp()` provider in `generateText`
LiteLLM	CRP as a provider target	route through CRP
Haystack / CrewAI / AutoGen	CRP-governed model wrapper	wrap the agent's LLM

1.2 The Adapter Guarantee¶

Each adapter preserves the framework's interface exactly - the developer's existing code keeps working - while routing every call through CRP governance. A LangChain user swaps ChatOpenAI for CRPChatModel and gets governance, grounding, and audit with zero other changes. The adapter is the on-ramp: no rewrite, no new mental model, immediate value inside the tools they already chose.

# LangChain - before
from langchain_openai import ChatOpenAI
llm = ChatOpenAI(model="gpt-4o")

# LangChain - after (governed + grounded, everything else unchanged)
from crp.langchain import CRPChatModel
llm = CRPChatModel(model="gpt-4o")

2. The Onboarding Flow - Value in Five Minutes¶

2.1 The Concrete Path¶

1. pip install crprotocol                              [30 sec]
2. crp init                                            [interactive, 1 min]
   → picks provider, writes crp.config.yaml, sets keys
3. First governed call works immediately               [zero-CKF mode]
   client.complete("hello")  → governance signals returned
4. crp ingest ./docs                                   [1-2 min]
5. client.ask("...")  → grounded, full-quality output  [value delivered]

Total: under five minutes from install to a governed, grounded result. The crp init command is critical - it removes the configuration cliff by generating the config interactively.

2.2 The Quickstart Repo¶

A public crp-quickstart repository the developer clones, runs, and sees working in one command - governance, ingestion, ask, a checkpoint, and the governance summary, all in a single runnable example. Seeing it work end-to- end in their own terminal is worth more than any documentation.

2.3 Time-to-Value Is the North-Star Metric¶

The adoption team's primary metric: median time from pip install to first successful grounded ask. Target: under 5 minutes. Everything in onboarding optimises this number. If it rises above 10 minutes, onboarding has a defect.

3. The Template & Example Gallery¶

Developers start faster from a working example than from a blank file. A gallery of runnable templates, each a complete governed application:

Template	Demonstrates
Governed chatbot	conversation (SPEC-028) + safety + checkpoints
RAG over your docs	ingest + CDR/CDGR grounded ask
Compliance-ready agent	tools + checkpoints + audit + evidence
Document generator	long-form quality-at-quantity (CDR)
Local-first deployment	BYO-storage (SPEC-038), no data egress
Code-governance CI	CRP Scan in a GitHub Action
Multi-agent system	safety budget + delegated agents

Each template is copy-runnable and maps to a real use case. The gallery is both onboarding and marketing - it shows what CRP makes easy.

4. Adoption Mechanics¶

Go-to-market detail is internal

The concrete free-to-paid funnel, conversion triggers, and expansion mechanics are documented internally because they describe commercial strategy rather than protocol behaviour. The public site describes the available plans on the Pricing page and the developer onboarding path in the Getting Started guide.

5. Developer Growth Surfaces¶

Surface	Purpose
docs.crprotocol.io	the progressive-disclosure documentation (SPEC-032 §8) - quickstart → guides → reference
Templates gallery	runnable starting points (§3)
Showcase	real systems built on CRP (social proof)
Community (Discord/GitHub Discussions)	support + ecosystem contributions (adapters, backends)
Changelog / roadmap	public progress; builds trust for a standards-track protocol
Standards page	the IETF/IANA story (credibility for enterprise)

5.1 Open Protocol as Adoption Engine¶

The protocol being open and standards-tracked (IETF, IANA) is itself an adoption driver: enterprises adopt standards more readily than proprietary lock-in, and the open spec invites community adapters and backends (SPEC-038) that expand the ecosystem without AutoCyber AI building every integration. Open drives adoption and invites community adapters and backends.

6. The Observability-as-Adoption Insight¶

A subtle but important point: CRP's value is mostly invisible (it prevents bad outcomes). The observability dashboard (SPEC-040 §5) makes the invisible visible - "47 fabrications caught this week," "quality trending up," "38% token savings." Making value visible is an adoption mechanism: developers who can see what CRP caught and saved stay and convert. Invisible value churns; visible value retains.

7. Honest Status & Limits¶

This is the least "protocol" and most "go-to-market" specification - it is product and growth engineering, not protocol design. It is included because adoption is what determines whether the 40 specs of engineering matter at all. The framework adapters are real engineering (each framework's interface must be tracked as it changes); the onboarding, templates, and docs are content and tooling work.

The conversion model is a hypothesis until validated with real funnel data. The triggers (§4.2) are reasoned but unproven; the adoption team must measure actual conversion and adjust. Time-to-value (§2.3) is the metric that will reveal whether onboarding actually works.

Framework adapters create a maintenance burden - each upstream framework changes independently. The adapter set should start with the highest-usage frameworks (OpenAI SDK, LangChain, LlamaIndex, Vercel AI SDK) and expand by demand, not cover everything upfront.

8. References¶

CRP-SPEC-032 - Developer Experience (the SDK adapters wrap)
CRP-SPEC-038 - Storage Backends (community-extensible ecosystem)
CRP-SPEC-040 - CRP Comply (the conversion destination)
CRP-SPEC-013/036 - Scan (the discovery funnel entry)