딥리서치 활용하기

소개

커리어 탐색을 위해 딥리서치를 활용해보았습니다.

진행 방법

지피티와의 일반 대화를 통해 향후 하고 싶은 일이 무엇인지 설명하고 저와 비슷한 생각을 가지고 있는 사람이나 회사, 서비스가 있는지 알아봄으로써 목표를 구체화하고자 하였습니다.

[DeepResearchPlan v1 — Screening Map for Special Populations (KOR 60 / Global 40; 2019–2025)]

Title

Screening Map of Practical Solutions for Aging & Neurodiverse Populations (Low–Mid–High Tech; Korea 60% / Global 40%)

Background (for context)

User: biomechanics/movement/cognition MS student; human-centered, field-based; wants small-scale, high-impact solutions and hybrid career (problem solver + maker + solution designer).

Use-case environments: home, community, school, workplace (daily living), not hospital-centric.

Research Objectives

1. Build a screening-level survey of real, market-facing solutions across Low/Mid/High tech for: older adults, MCI, ASD, ADHD.


2. Identify 30–50 reference entities (KOR+Global), with URLs and business model snapshots; mark a “Top 10 closest match” set.


3. Map low-cost/high-impact opportunity zones suited for a student-to-early-career creator.


4. Extract career path patterns, skills to build, and concrete next-step projects.



Scope & Constraints

Timeframe: last 5 years (2019–2025).

Geography mix: Korea ~60%, Global ~40%.

Tech ratio target: Low 40 / Mid 40 / High 20.

Evidence threshold: commercialized or actively commercializing; pilots with paying customers or credible B2B/B2G traction preferred.

Include both physical tools and digital (apps, platforms, DTx, wearables, AI); prioritize field-tested, user-involved design cases.

Language: no restriction; include KR/EN sources.

Sensitive issues (regulatory/ethics/accessibility) only as brief flags where relevant.

Methodology (how to search & judge)

Sources: company sites, app stores, accelerator/healthtech directories, assistive tech catalogs, university labs/tech transfer pages, rehab conferences (programs/papers), maker/assistive communities.

Verification: confirm active website/social update within 12–24 months; note regulatory status (MFDS/FDA/CE) if claimed; avoid defunct or purely conceptual.

De-duplication: cluster by company/product; keep the most representative entry.

Scoring (0–5 each; compute weighted score):
• Fit-to-user-philosophy (human-centered, field-based) – 25%
• Impact on target population/problem clarity – 25%
• Evidence & traction (pilots/sales/approvals) – 20%
• Feasibility for student entry (cost, buildability, openness) – 20%
• Korea relevance (for KOR quota) – 10%

Output style: table-first, bullets minimal, 1–2 sentence descriptions, working URLs only.

Deliverables (tables first, brief prose only where needed)

1. Low-tech Solutions (40%)
Columns:
| Category(센서리/균형/인지-운동/일상ADL/치매친화) | Target(User) | Problem | Solution/Product | Region | Price band | Evidence(usage/traction) | Biz model(B2C/B2B/B2G/Rx/OTC) | Design philosophy & co-design notes | Strengths | Limits | URL |


2. Mid-tech Solutions (40%)
Columns:
| Category(스마트폰/간단센서/저가웨어러블/커뮤니티헬스) | Target | Problem | Product/App | Core tech | Regulatory/claims | Biz model + GTM | Data/privacy notes | Strengths | Risks | URL |


3. High-tech Solutions (20%)
Columns:
| Category(AI gait/balance, exosuit/robotics, advanced rehab, DTx, neurotech) | Target | Clinical/Use setting | Product | Evidence(regulatory/clinical) | Price/CapEx/Opex | Biz model | Student entry points (open SDK/clinical collaborators/similar DIY) | URL |


4. Reference Companies & Labs (30–50 entities; KOR 60% target)
Columns:
| Name | Country | Tech level(L/M/H) | Population | What they do (1–2 lines) | Why it matches user’s philosophy | Traction signal | URL | Top-10? (★) |


5. Creator / Maker Examples
Columns:
| Individual/Team/Channel | Focus (assistive/maker/therapist-inventor) | Representative builds | Audience served | Openness(OSS/files) | Contact/URL | Why it’s a model for user |


6. Market Landscape & Opportunity Map



Produce a 2x2 matrix image or ASCII: (Cost/Complexity) × (Impact/Unmet need).

Bullet list: saturated vs. underserved niches; aging-in-place; community mobility; low-friction screening/monitoring; adherence/behavioral layer; caregiver tools.
Columns (summary table):
| Problem area | Current solutions snapshot | Gaps/Barriers | Low-cost high-impact ideas (student-buildable) | KR-specific angle | Notes/URL |

7. Career Path Insights for the User



Output 3–5 archetypes with “why fit” and “how to start in 6–12 months.”
Columns:
| Archetype | Core activities | Toolchain/skills | Typical employers/partners | Starter portfolio project | Long-term trajectory(5–10y) |

8. Actionable Next Steps
Columns:
| Skill to build | How to learn fast | 1–2 weekend prototypes | KR labs/companies to contact | Global references | Success metric(what to validate) |



Inclusion/Exclusion Rules

Include only entities with shipping product, pilots, or active deployments; exclude pure concepts/indefinite “coming soon”.

Favor user-tested/field-evaluated tools; mark co-design with special populations where found.

Ensure at least 60% of entries are Korea-based or KR-active; still include global exemplars for contrast.

Partition by User’s Populations & Settings

Tag each entry by {Older adult, MCI/dementia, ASD, ADHD} and {Home, Community, School, Workplace}. Prefer entries usable outside hospitals.

Final Packaging

One master spreadsheet (tabs for sections 1–8) + a 6–8 page overview with: Top-10 matches, opportunity 2x2, and career archetypes.

Use consistent tags: [pop] [setting] [tech] [bizmodel] [evidence] to enable filtering.

Quality/Style Notes

KR/EN mixed output ok; keep KR first for Korea entities. Use concise phrases, avoid marketing fluff.

Every row MUST include a working URL; if multiple, choose the clearest product page.

결과와 배운 점

생각이 구체화되지 않은 상태에서는 gpts를 활용한다고 해도 한계가 있다고 느꼈습니다. 특히 한번에 많은 양을 조사하고자 할 때 정보 탐색이 제대로 되지 않는 경우가 있었고, 적당한 분량을 나누어 질문하는 것이 중요한 것 같습니다.

도움 받은 글

딥리서치 프롬프트 작성 도우미

https://chatgpt.com/g/g-67e2161dd7b88191bfc4e6b56b66ce90-dibriseoci-peurompeuteu-jagseong-doumi

1개의 답글

뉴스레터 무료 구독

👉 이 게시글도 읽어보세요