TopPrism

Language / 语言: English primary · 中文使命说明 included below。

Software should adapt to people—not the other way around.

The agentic era gives us a chance to move beyond rigid forms, menus and workflows, and build software that understands human intent, collaborates with judgment, and helps people create more value.

让人摆脱僵硬软件流程的束缚，让 AI 与软件真正放大人的判断与创造。

Freedom from rigid software does not mean freedom from evidence, responsibility or governance. Human authority remains part of the system.

---

What we work on today

TopPrism provides AI-powered decision intelligence for consumer-goods businesses, turning physical-world data and operational constraints into explainable, reviewable decisions.

We work on turning operational reality, business knowledge, objectives and constraints into insights, recommendations and decisions that people can understand, review and act on.

Operational reality
        ↓
Understand what exists and what is happening
        ↓
Explain what it means—with evidence and limits
        ↓
Form alternatives, recommendations or decisions
        ↓
Human review, action and outcome
        ↓
Learn without silently rewriting organizational truth

Public evidence map

The repositories below answer different parts of that problem. The map describes their public questions and boundaries; it does not claim that every repository is one integrated product or has the same commercial maturity.

Understand the business world

• prism-ontology — governed semantic contracts and operational profiles.
• bge-entity-match — whether records from different systems refer to the same real-world entity.
• spatial-decision-intelligence — spatial-data trust, boundary generation and decision-readiness research; its strongest published operational validation currently centers on geofences.
• market-partition — deterministic geometry from human-defined market boundaries.
• themed-street-engine — commercially meaningful corridor structure from POI and road-network signals.

Explain the business world

• mousheng-outlet-insight — structured, traceable outlet insight that stops before execution.

Form and study decisions

• sales-resource-allocation-framework — dealer-territory decision design using a world model, governed knowledge and evidence-linked recommendations.
• visit-scheduling-optimizer — recurring field-sales visit planning under explicit operational constraints.
• open-dispatch and logistics-dispatch-clustering — explicit dispatch optimization and learned operational priors, currently published as separate capabilities.
• agentic-warehouse-engine and fashion-lifecycle-pricing — Decision Science research, not production-product claims.

---

How TopPrism learns

Our Native AI work explores how project experience can become reliable organizational capability without promoting every pattern, note or model output into truth.

Experience → Evidence → Candidate Capability
          → Evaluation & Governance
          → Reuse → New Experience

• agent-nurture-framework defines governance contracts; it is a protocol, not an agent runtime.
• cultivating-ml-agent records longitudinal ML capability formation across projects.
• notebook-knowledge-distillation, skill-tester and three-layer-wisdom-extraction cover candidate formation, evaluation and abstraction research.
• topprismwiki implements evidence-governed knowledge publication and agent-facing context.

Learning projects, employee utilities and upstream forks remain visible in the repository list as supporting evidence and reference assets. They are not presented as customer products.

The machine-readable repository map and relationship registry lives in TopPrism/.github. The shared company and capability vocabulary is maintained in TopPrism/.github/CANONICAL.md. Selected public evidence cases are summarized in TopPrism/.github/EVIDENCE.md.

---

How to read a TopPrism repository

We separate three questions:

Dimension	Meaning
Purpose	What question does this repository address?
Maturity	Is it applied, validated, experimental, internal or reference material?
Evidence	What data, study, benchmark or evaluation supports its claims?

A research result is not automatically a production deployment. Internal evaluation is not an external benchmark. A recommendation is not an approved decision. A decision is not execution.

What this GitHub is—and is not

This account is a public engineering evidence layer: reusable capabilities, applied systems, research, learning histories and clearly attributed upstream references.

It should not be read as a flat product catalog, a collection of unrelated demos, a spatial-only portfolio, a consulting brochure or a generic AI-agent toolkit.

Contact

TopPrism Data · topprismdata.com (in development)

Public engineering inquiries should use the company contact channel rather than personal addresses.

License

The profile content is released under the MIT License.