From half-baked ideas to production-ready reality—without the translation tax
Intentive is designed to be a standard way to remove friction in an AI-native system. Tell the system what you want and watch it turn those words into code, workflows, and an audit trail you can actually trust.
Why Intentive Was Started
- Lost in Interpretation – Bold OKRs shrink into bite-sized tickets, specs drift, and the 2 a.m. fire-drill never quite goes away.
- Automation Islands – RPA here, CI/CD there, a chatbot somewhere else. None agree on a common “language of getting work done.”
- Speed vs. Compliance – Shipping fast is easy … until the first SOC-2 questionnaire hits your inbox.
What Makes Intentive Different
1. Intent as Data
Intentive captures a request - say “Run May payroll for APAC with quarterly bonuses” - as a node in a Canonical Intent Graph. Every downstream action (code generation, workflow orchestration, approvals) hangs off that node. No translation loss, no hidden side-effects.
2. Opinionated, Swappable Layers
Rather than a single monolith, Intentive is composed of clear layers—each replaceable if your stack demands it:
- Intent Mesh Gateway – Classifies and routes new intents (Fastify + embeddings).
- Temporal Workflows – Handles long-running jobs.
- Knowledge Graph – Anchors domain context (Neo4j, type-safe).
- Guardrails – Enforces budget, policy, and security (Policy DSL + OpenTelemetry).
- GraphQL Fallback – Provides a read-only safety net when custom handlers are offline.
If GitHub Actions, Zapier, and your risk team threw a hack-day together, this is what they’d ship.
AskQL vs. IntentQL — A Living Comparison
- Purpose
- AskQL focuses on safe, read-only queries.
- IntentQL (in active development) will drive full lifecycle actions—reads, writes, and triggers.
- Mental Model
- AskQL feels like a “smart SELECT.”
- IntentQL aims to feel like “git commit && deploy” for real-world operations.
- Example Usage
- AskQL → ask { orders(where:{status:"OPEN"}) { id } }
- IntentQL (road-map) → intent { launchPromo("SPRING25") region:"APAC" }
AskQL lets teams interrogate data safely. IntentQL will let them change the world behind that data—using the same mental model.
Intentive - For the Builders
- No Lock-In – Swap any layer; the core is open.
- TypeScript First – Strict types from HTTP edge to graph node.
- Plug-in Friendly - Designed with a plugin vision in mind
Roadmap Highlights
- Q3 ’25 – IntentQL alpha + multi-endpoint GraphQL fallback
- Q4 ’25 – Live RBAC guardrails + Slack approvals
- Q1 ’26 – Visual Intent Designer (drag-and-drop graph editor)
Try It
- Docs – intentive.dev for a five-minute quick-start (docs will be updated shortly on site).
- GitHub – https://github.com/agenticaivc/intentive Star the repo.
Intentive turns the sentence you just spoke into an auditable, observable, production-grade outcome—quicker than you can schedule the kickoff. Ready to let intent become your primary interface?