EST. 2022 — SEOUL · KIGALI
AI ENGINE · PRODUCT BUILDER · SINCE 2022
AI engine to product, direct.
A full-stack technology firm that designs, builds, and runs voice-analysis engines, LLM pipelines, multi-tenant SaaS, and mobile apps. 9 patents filed. 76+ repos.
- 9 patents filed (child-language / voice analysis)
- 76+ cumulative repositories
- 15+ B2B / public references
- SEOUL · KIGALI
- 9 patents filed — child-language development and voice analysis
- Dururbarun Social Cooperative · speech therapy engine live
- KAIST · biofoundry education platform build
- Band Foundation · Eunpyeong · Gong-gong Gong-gan · Gangwon Education
- 76+ cumulative repositories · 15+ B2B and public references
- Certified venture · Deep-tech Startup (2025)
- Seoul · Kigali engineering
The problems we solve
You bring the problem. We ship the engine.
Recurring problems from public sector, research, and clinical settings — solved by an in-house AI engine plus full-stack build. These are live today.
- 01
Language development assessments take 2–3 hours each, backing up the queue.
→ A 3-minute screening report from an audio recording.
SpeechMap · KAIST · Dururbarun Social Cooperative
- 02
Drafting specialist reports costs ~1 hour per session.
→ 3-pass LLM generates the draft report and supporting visuals.
Dururbarun Speech Therapy Center in production
- 03
Clinical data is scattered across spreadsheets and audio files.
→ Assess → store → analyze in one platform. Meets public-sector procurement requirements.
Gangwon Province Education live
- 04
RFP requirements are strict and typical outsourcing can’t clear them.
→ One team builds end-to-end against procurement, security, and accessibility bars.
15+ B2B and public-sector references
WHO WE ARE
Three sentences that define us.
01 — AI Engine Builder
We build the engine ourselves.
Not an agency-style delivery. We design, build, and patent voice-analysis, LLM-report, and RAG pipelines in-house.
02 — Full-stack Delivery
We run the product to the finish.
Design, web, mobile, admin, payments, deployment — closed inside a single team. Next.js 16 · React 19 · Supabase · Cloud Run.
- Next.js 16
- Supabase
03 — Trusted Partner
Public sector and B2B trust us.
KAIST · Gangwon Office of Education · Eunpyeong-gu Office · Band Foundation · Woori-hamkke Social Cooperative · Dururbarun Social Cooperative — references across public, research, and B2B.
- KAIST
- Public
- Dururbarun
Our AI
SpeechMap — a voice-analysis engine, 9 patents filed.
Linguistic analysis (post-STT) and acoustic/prosody analysis fused inside a single engine. We designed the engine ourselves and filed 9 in-house patents in child-language development and voice analysis. Deployed inside Dururbarun Co-op's speech therapy program today.
- in-house patents filed
- 9
- cumulative repositories
- 76+
- clinical R&D partner
- Dururbarun
음향
MFCC · Δ
운율
F0 · Jitter
언어학
STT 후처리
References
25+ references across public, research, and B2B.
Only institutions we have clearance to name. Public sector, universities, and research labs — plus B2B, commerce, and healthcare clients.
- KAIST
- Seoul Media Institute of Technology
- Band Foundation
- Eunpyeong-gu Office
- Gangwon Office of Education
- Woori-hamkke Social Cooperative
- Dururbarun Social Cooperative
- Gong-gong Gong-gan (WithGoods)
- Marea Holdings
- Kijamitable (Kijanify)
Why we build it ourselves.
The AI engine and the product — design, web, mobile, deploy — all live inside one team. Remove the middle layers and quality and speed rise together.
AI engine
- Typical setup
- Direct OpenAI / Gemini API calls
- Song & Stark
- Voice-analysis engine designed in-house · 9 patents filed (child-language / voice analysis)
Dev pipeline
- Typical setup
- Planning · design · dev subcontracted separately, manual deploy
- Song & Stark
- AI agents + automation fused into every step
Team structure
- Typical setup
- Sales PM → planner → subcontracted dev, 3-4 hops
- Song & Stark
- Design · web · mobile · payments · deploy inside one team
Academic grounding
- Typical setup
- Off-the-shelf tools, no professor on the invention
- Song & Stark
- 9 in-house patents filed; papers and IP as assets
Communication
- Typical setup
- Contractor → account PM → outsourced dev
- Song & Stark
- Direct Slack and email with the core engineer
Delivery evidence
- Typical setup
- Generic web / app portfolio
- Song & Stark
- 15+ B2B and public-sector references · 76+ repos
Letter from the founder
AI engine to product, direct. Voice-analysis engines, LLM pipelines, SaaS, and mobile — designed and deployed by our own hands.