# imply+infer – Ideal customer profile (ICP)

Use this when reasoning about who imply+infer is for and what problems we solve.

---

## Core problem we solve

- **Hardware**: Best-in-class AI silicon exists (e.g. NVIDIA Jetson 67+–275 TOPS, Qualcomm).
- **Reality**: Most teams cannot use it. Setup takes 40–120+ hours: CUDA/TensorRT/driver issues, scattered docs, “evaluation” vs “production” gap of months.

imply+infer turns evaluation hardware into plug-and-play, production-ready systems.

---

## Primary ICP

- **Engineers and developers** building edge AI: prototypes, robots, drones, smart infrastructure, autonomous systems.
- **Researchers** running vision, LLMs, or multi-modal models at the edge.
- **Startups and small teams** who need to ship quickly without a dedicated integration team.
- **Enterprises** deploying fleets who want custom configurations and support.

---

## Jobs to be done

- Go from “eval board in a box” to “boots in <30s, stack works” in minutes, not months.
- Avoid dependency hell: CUDA, TensorRT, OpenCV, drivers, kernels—all verified together.
- Start building on day one with pre-installed apps (Ollama, YOLO, Roboflow, etc.) instead of spending weeks on environment setup.

---

## Fit signals

- Evaluating or already using Jetson, Qualcomm, or similar edge AI platforms.
- Has felt pain: “TensorRT won’t compile,” “camera needs a custom kernel,” “PyTorch wants a different NumPy.”
- Willing to pay a premium for a complete, pre-configured system (vs. DIY from components).
- Values time-to-prototype and time-to-production over minimal hardware cost.

---

## Not a fit (or secondary)

- Pure cloud-only AI; no edge or on-device requirement.
- Hobbyists who enjoy and have time for deep system integration.
- Buyers who only care about lowest component cost and will do all integration in-house.

---

## How we describe ourselves to the market

- “The Raspberry Pi moment for professional AI hardware.”
- “Making powerful AI hardware actually usable.”
- “Where implying capability meets inferring compatibility.”