AI Post-Call Sales Coaching for Dog Trainers
Automate coaching feedback after every sales call. Turn your best sales methodology into an AI coach that scales with your team.
2 Weeks
AI Builder
Manual Upload
Manual coaching doesn't scale
Ariel reviews calls manually — this doesn't scale
Feedback is delayed, inconsistent, or never delivered
No structured way to enforce sales methodology
Trainers miss coaching moments that affect close rates
How Ask Moses Works
Upload Call
Admin uploads audio file or pastes transcript directly
Transcription
If audio, Whisper AI transcribes automatically
AI Analysis
LLM evaluates transcript against the sales rubric
Coaching Email
Personalized feedback sent to the dog trainer
Aggregate Summary
Ariel receives daily/weekly coaching digest
History Log
All analyses stored and searchable
What We're Building
Script / Rubric Manager
Admin interface for Ariel to create, edit, and version sales scripts and evaluation rubrics.
- Multi-script support
- Version history
- Markdown editor
- Rubric sections with weights
Manual Call Upload
Simple interface to upload call recordings or paste transcripts for analysis.
- Audio upload (.mp3, .wav, .m4a)
- Direct transcript paste
- Whisper AI transcription
- Trainer selection
AI Call Analysis
LLM-powered evaluation of each call against the active rubric.
- Per-section scoring
- Specific quote citations
- Improvement suggestions
- Overall rating
Post-Call Coaching Email
Automated email delivery of personalized coaching feedback to the trainer.
- Branded HTML template
- Score breakdown
- Actionable tips
- Encouragement tone
Aggregate Summary
Daily or weekly digest for Ariel with team-wide coaching insights.
- Average scores
- Top performers
- Common gaps
- Trend indicators
Simple History Page
Searchable log of all past call analyses for reference and debugging.
- Filter by trainer
- Filter by date
- View full analysis
- Basic search
Out of All Tiers
These features are not included in any of the three tiers. They represent future phases after MVP validation.
CRM Integration
- GoHighLevel (Pepper) sync
- Auto-pull trainer contacts
- Sync results to CRM
- CRM workflow triggers
Real-Time Features
- Live call coaching
- Real-time voice analysis
- Instant feedback during calls
- Tone detection
Platform Scale
- Multi-tenant architecture
- White-label solution
- Mobile app
- SSO / SAML authentication
Advanced Analytics
- Predictive close-rate modeling
- A/B testing of scripts
- Performance correlation dashboards
- Team leaderboards
How We Measure Success
Do trainers engage?
- Email open rate > 60%
- Trainers reference feedback in calls
- Ariel uses aggregate summaries weekly
Is feedback helpful?
- Qualitative survey: 4+ out of 5
- Trainers request more detail (good sign)
- Ariel saves 5+ hrs/week on manual review
Is AI accurate?
- Rubric scores align with Ariel's judgment 80%+
- No hallucinated quotes
- Actionable suggestions (not generic)
Does it impact outcomes?
- Close rate improvement tracked
- Time-to-close reduction
- Trainer confidence increases
Post-MVP Roadmap
After the MVP (Starter → Pro → Pro + RAG), here's how Ask Moses can continue to evolve.
MVP Build Timeline
Starter
Manual Upload + AI Core
Pro
+ Twilio Integration
Pro + RAG
+ Knowledge Base
CRM Integration
- GoHighLevel (Pepper) sync
- Auto-pull trainer contacts
- Sync coaching results to CRM
- Workflow triggers
Enhanced Coaching
- Interactive coach dashboard
- Trainer response tracking
- Script A/B testing
- Detailed analytics
Real-Time Features
- Live call guidance
- Voice tone analysis
- Real-time objection handling
- Instant coaching nudges
Platform Scale
- Multi-tenant SaaS
- White-label offering
- Enterprise SSO
- Public API
Built to Evolve
Every component we build in the MVP is designed to scale with your product. This is production-ready code, not a throwaway prototype.
| MVP Component | What We Build | Phase 1.5: Twilio | Phase 2: GHL | Phase 3: Enhanced | Reuse |
|---|---|---|---|---|---|
Script/Rubric Manager | Create & edit sales scripts | Same, no change needed | Per-team scripts via GHL | A/B testing scripts | 95% |
Manual Call Upload | Audio upload + paste transcript | Replaced by Twilio webhook | Auto-pull from GHL | Multi-source ingestion | 85% |
AI Call Analysis | Per-section scoring & citations | Real-time processing | Sync results to CRM | Detailed analytics | 90% |
Post-Call Coaching Email | Branded HTML + tips | Auto-triggered on call end | GHL workflow triggers | Trainer response tracking | 100% |
Aggregate Summary | Daily/weekly digest | Same, no change needed | Team comparisons | Advanced analytics dashboard | 90% |
Simple History Page | Search & filter calls | Real-time updates | CRM-linked records | Interactive coach dashboard | 95% |
Code vs No-Code Platforms
- ×Platform lock-in
- ×Limited AI customization
- ×Needs specialized "no-code dev"
- ×Rewrite needed to scale
- ×Monthly platform fees forever
- You own the code 100%
- Full AI SDK access
- Any React/Next.js dev can maintain
- Scale without rewriting
- Deploy anywhere (Vercel, AWS, etc.)
Choose Your Starting Point
All tiers include production-ready code you own. ~90% reusable as you scale.
Starter
Week 1-2: Validate the core AI coaching value with manual workflow
Not included:
Pro
Week 3: Automated call ingestion for hands-off operation
Not included:
Pro + RAG
Week 4: Context-aware coaching powered by your training materials
Upgrade Path
Start with Starter to validate fast. Upgrade anytime — your codebase is built to evolve.
Ongoing Costs (Post-MVP)
Per Call Analysis Cost
Based on average call duration: 12.5 minutes (range: 5-20 min)
| Stack | Transcription | Analysis | Cost/Call |
|---|---|---|---|
| Premium | Whisper ($0.075) | GPT-4o ($0.10) | ~$0.18 |
| Recommended | Whisper ($0.075) | Gemini 1.5 Flash ($0.01) | ~$0.09 |
| Budget | Deepgram ($0.054) | Gemini 1.5 Flash ($0.01) | ~$0.07 |
| Ultra Budget | Groq Whisper ($0.023) | Gemini 1.5 Flash ($0.01) | ~$0.03 |
*Based on 12.5-minute average call. Includes email delivery (~$0.001).
RAG Knowledge Base Costs (Pro + RAG tier)
Audio files require transcription before indexing
| Operation | Model | Cost |
|---|---|---|
| Audio transcription (12.5 min) | Whisper | ~$0.075 |
| Text embedding | OpenAI ada-002 | ~$0.0001 |
| Total per audio file indexed | ~$0.08 |
Index 50 calls
~$4.00
Index 100 calls
~$8.00
*One-time cost per document. Text documents (PDFs, docs) only require embedding (~$0.0001 each).
Per Call
$0.03 - $0.18
Depends on stack choice
Hosting
~$20/mo
Vercel + Supabase
100 calls/day
$90 - $540/mo
Estimated API costs
Team & Timeline
Team Evolution
AI Builder
Solo execution with v0
Full Stack Developer
Integration & scaling
Senior AI PM
Product & strategy
2-4 Weeks
Depending on tier
1 → 3
Team growth
AI-First
Modern stack
MVP Tech Stack
Appendix
Detailed Deliverables
Comprehensive breakdown of what is included in each tier. This serves as the scope contract for the engagement.
Starter
DELIVERED ✓Script & Rubric Manager
- Create/edit scoring rubric with criteria(Persisted in Supabase)
- Define pass/fail thresholds per criterion
- Single active script at a time
- Customize system prompt for AI analysis
Manual Call Upload
- Upload audio files (.mp3, .wav, .m4a)
- Paste transcript text directly
- Input trainer name and email manually
- Upload progress indicator
AI Call Analysis
- Whisper transcription (if audio)
- GPT-4o analysis against rubric
- Fetches criteria + system prompt from Supabase
- Score per criterion (pass/fail)
- Overall call score
- Specific feedback per criterion
- Full transcript display for admin review
Post-Call Coaching Email
- Manual send coaching email via Resend
- Personalized coaching feedback
- Actionable improvement tips
- Score summary included
- Auto-save call to database after email send
Simple History Page
- List of all processed calls
- Search by trainer name or email
- View individual call details with full analysis
- Display transcript and scores
Aggregate Summary
- Total calls processed dashboard
- Average score trend visualization (line chart)
- Top improvement areas aggregated (bar chart)
- Smart insights & performance summaries
- Achievement badges & trainer recognition
- Real-time analytics dashboard
Pro
Twilio Webhook Integration
- •Webhook endpoint for call events
- •Automatic call ingestion when call ends
- •Twilio signature validation (security)
- •Recording URL extraction
- •Contact metadata parsing
Database Enhancements
- •Call queue for processing
- •Retry logic for failed processing
- •Webhook event logging
- •Idempotency handling (no duplicates)
Admin Notifications
- •Error alerts for failed processing
- •Daily summary email (optional)
- •Processing status dashboard
Pro + RAG
RAG Knowledge Base
- •Vector database setup (pgvector)
- •Document upload interface
- •Automatic chunking & embedding
- •Support for PDFs, docs, text files
Contextual AI Analysis
- •Semantic search across knowledge base
- •Context-aware coaching feedback
- •Reference specific training materials
- •Quote relevant best practices
Knowledge Management
- •View/delete uploaded documents
- •Re-index documents on demand
- •Document usage analytics
- •Version tracking for updates
Note: Any features, integrations, or modifications not explicitly listed above are considered out of scope and will require a separate agreement. Changes to scope during development may affect timeline and pricing.