Product direction · MVP assessment
A practical comparison of ready-made products, open-source foundations, and a fully custom build — assessed against the Oderland-inspired MVP brief.
Recommendation: custom editor experience + open-source AI foundation.
There is no clear off-the-shelf product that exactly delivers the required combination of selected-text editing, preview-before-apply, safe preservation of Gutenberg markup, Classic Editor support, and an Oderland-like UX.
Build the narrow product-specific plugin yourselves, but reuse open-source WordPress and AI provider infrastructure. This gives us control of the user experience and safety model without rebuilding model connectivity, credentials, and provider abstractions from zero.
The decision space
Install and configure a plugin such as AI Engine, GetGenie, or AI + Block Editor.
Build a custom Viva plugin on Gutenberg/WordPress APIs and reuse an AI client or provider layer.
Own the plugin, provider connectors, credentials, API layer, editor adapters, and operational tooling.
Existing products
| Product | What it offers | Fit for this MVP |
|---|---|---|
| AI Engine | Broad WordPress AI framework; multiple providers, content generation, APIs, function calling, and MCP-related capabilities. | Strong foundation Not a drop-in Oderland editor. |
| AI + Block Editor | Open-source Gutenberg features including tone changes, headlines, summaries, slugs, keywords, provider switching, and an AI sidebar. | Useful reference Gutenberg-focused; Classic Editor and safety workflow need work. |
| GetGenie | Ready-made AI content writing and SEO workflows inside WordPress. | Partial fit SEO/content product rather than a neutral custom editor. |
| WPVibe | Open-source MCP server for connecting Claude, ChatGPT, and agents to WordPress. | Later phase Better aligned with Step 2/3 CMS-agent work. |
Open source building blocks
A WordPress-native prompt and provider abstraction with PHP/JavaScript APIs, credential handling, model preferences, JSON output, capability checks, and WordPress-style error handling.
The repository is GPL-2.0. It currently notes a transition toward built-in WordPress 7.0 AI functionality, so compatibility with customer environments must be verified first.
Use the open-source block editor APIs for toolbar actions, sidebars, selection handling, notices, and targeted block updates.
The plugin must update only the selected content or target block — never replace the whole post casually.
Official open-source provider packages are available for OpenAI, Anthropic, and Google, enabling a provider-agnostic direction.
Open source does not remove model/API usage costs or the need for operational monitoring.
Recommended MVP shape
Technical spike
Scope boundary
Do not include RAG, vector databases, site-wide indexing, SEO automation, image generation, mass editing, complete page generation, WooCommerce agent actions, MCP orchestration, autonomous publishing, or advanced model routing in the first MVP.
WooCommerce remains a future integration target. Its open-source APIs can support later product and CMS operations, but they are not required for the initial text-editing proof.
Decision summary
Approve a focused technical spike using a custom Viva plugin UI and an open-source AI/provider layer. Benchmark AI Engine and AI + Block Editor during the spike, but do not commit the final UX to either product before verifying the exact Oderland workflow and content-preservation requirements.
If the spike passes in both Gutenberg and Classic Editor, continue into the MVP. If the shared AI layer is incompatible with customer WordPress versions, keep the custom plugin and swap only the provider adapter.
References