Laravel Ai Core Laravel Package

Product Decisions This Supports

• Standardize how AI-generated outputs are assigned to domain entities in Laravel instead of building bespoke assignment, correlation, and reporting glue.
• Support roadmap items around AI-assisted workflows, classification, scoring, content generation, and human-review queues where outputs must be tied to a customer, order, document, or other subject via subject key stamping.
• Enable admin/QA workflows with rejected-answer reports, making it easier to review low-quality or rejected AI outputs and feed findings back into prompt/model improvements.
• Use assignable models and the row factory to accelerate prototypes, fixtures, and tests for AI assignment flows.
• Inform build-vs-buy: adopt where a Laravel-native assignment/evaluation layer is the main need; build custom orchestration if the roadmap requires multi-provider routing, streaming, tool calling, embeddings/RAG, or cost governance not evidenced in v0.4.0.
• Account for migration cost: teams extending ModelAssignments must replace assign() overrides with protected store() implementations because assign() is now final. Pin versions and budget an engineering pass if custom subclasses exist.

When to Consider This Package

• Consider it for Laravel/PHP applications that need a structured way to assign AI outputs to models, correlate results by subject key, report rejected answers, and create test data quickly.
• Good fit for internal tools, admin dashboards, pilot AI features, evaluation workflows, and multi-tenant or per-subject AI output tracking.
• Adopt with caution in production: this is a 0.x package with breaking API changes already present in v0.4.0. Evaluate documentation, test coverage, maintainer responsiveness, and upgrade frequency before committing.
• Look elsewhere if the product needs a mature AI gateway, multi-provider failover, streaming responses, prompt management UI, vector search/RAG, observability, cost controls, or strict enterprise support.
• Avoid or defer if the team relies on overriding assign() in ModelAssignments subclasses and cannot absorb the migration in the near term.

How to Pitch It (Stakeholders)

• Executives: v0.4.0 adds practical capabilities—subject key stamping, assignable models, rejected-answer reports, and a row factory—that can reduce custom glue for AI assignment and review workflows. It supports faster pilots and clearer QA reporting, but it remains early-stage and should be scoped to internal or non-critical features first.
• Engineering: plan for the breaking change by implementing protected store() in ModelAssignments subclasses; assign() is final. Use the row factory for tests, and validate the package against your Eloquent models and reporting needs before broad adoption.
• Product: prioritize use cases where outputs must be tracked against a subject and reviewed for rejection/quality. Do not position this as a full AI platform; treat it as a Laravel-native assignment/evaluation building block. Budget migration effort if custom subclasses exist, and pin versions until the API stabilizes.