Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

Syapa: The Sloth's Secret Weapon Against Freelance Scope Creep (Powered by Gemma & Local AI)

Описание и применение

Syapa автоматизирует пост‑встречные задачи, сокращая ручной труд примерно на 1,5–2 часа на клиента, и устраняет неотслеживаемые дополнительные работы через проверенное выявление «scope creep».

Задача

Freelancers spend excessive time post‑meeting reviewing recordings, spotting unbilled scope creep, and drafting pushback emails.

Результат

Syapa automates post‑meeting tasks, reduces manual effort by ~1.5–2 hours per client, and eliminates unbilled concessions through verified scope‑creep detection.

Рабочий процесс

  • 1. Click "Start Recording" in Chrome MV3 side panel during a call.
  • 2. Recording is streamed to MediaRecorder and sent to /api/meetings/transcribe.
  • 3. Server validates audio, calls ElevenLabs STT, performs speaker diarization, and stores transcript segments.
  • 4. Asynchronous pipeline runs Gemma locally via Ollama to extract summary, action items, scope creep, and unbilled promises.
  • 5. Deterministic string‑matching verifies quotes against transcript; unverifiable quotes are discarded.
  • 6. Generated pushback email and change‑order template are made copy‑pasteable.
  • 7. Optional practice mode uses ElevenLabs Conversational AI to spar against realistic client personas.

Доказательства

Every freelancer knows the dread of post‑meeting administrative chores.

Источник

Syapa lets freelancers channel their inner sloth:

Источник

Zero Post‑Meeting Effort: Once the call ends, you literally walk away. Syapa’s asynchronous background pipeline automatically transcribes the conversation with speaker diarization...

Источник

Deterministic Hallucination Filter: All extracted quotes are verified verbatim against the raw transcript using deterministic string validation.

Источник

Automated Boundary Coaching: Syapa generates copy‑pasteable negotiation pushbacks and formal change orders ready to send to the client.

Источник

Total Time Saved: ~1.5 to 2 hours per client

Источник

  • Дата записи: 2026-10-25T18:26:08.787Z