EnergyPal Inbound AI Retell Optimization & Consultancy Proposal
Prepared for: EnergyPal
Prepared by: AYB AI
(Automate Your Biz)
Date: 05/08/2026
Executive Summary
This updated proposal reflects the revised approach discussed after EnergyPal showed AYB AI the existing Retell inbound AI agent that has already been partially built internally.
Instead of building from scratch, AYB AI will jump directly into the existing Retell setup, tighten the agent for production use, and help bring the system to a stronger live-ready state.

EnergyPal has already completed a meaningful amount of the work internally - the Retell agent is approximately 70–80% ready.
However, several production-readiness areas still need refinement before the agent is suitable for live inbound call traffic. This proposal is structured in two phases:
Phase 1 - AYB AI Retell Production Readiness Sprint

Phase 2 - Ongoing Consultancy & Performance Optimization
This approach allows EnergyPal to keep ownership of the CRM/API work internally, while leveraging AYB AI's experience with production-ready voice agents, call review, prompt refinement, and performance optimization.
Return On Investment
Reduced Dev Time
Reduced manual development time for EnergyPal by leveraging AYB AI's production experience.
Faster Go-Live
Faster route to go live because AYB AI is optimizing the existing Retell build rather than starting from scratch.
Improved Performance
Improved long-term AI performance through manual call review and continuous recommendation cycles.
Goals & Current Context
EnergyPal's Goals
  • Start with inbound call triage as this is an operational gap
  • Capture missed and overflow inbound calls
  • Route callers to the correct next step based on caller type and intent
  • Improve appointment booking, callback handling, support triage, and customer experience
  • Use the existing Retell agent as the starting point rather than rebuilding from scratch
  • Keep CRM/API work handled internally by EnergyPal
  • Use AYB AI's voice-agent production experience to improve agent performance before and after launch
Current Context
  • EnergyPal has already built a Retell inbound AI agent that is approximately 70–80% ready
  • The agent still requires production-level refinement before live call traffic
  • EnergyPal has strong internal technical capability and can handle the CRM/API integration work
  • AYB AI's role will be to optimize the Retell agent, improve voice-agent logic, and post-launch provide performance recommendations based on manually reviewing call recordings
Revised Engagement Model
The original proposal focused on AYB AI building the inbound triage agent from scratch.
After reviewing EnergyPal's existing Retell build, the recommended approach is now a two-phase optimization and consultancy model.
1
Phase 1
AYB AI helps finish the existing Retell setup and bring it to a production-ready standard based on prior industry experience handling tens of thousands of production calls.
2
Phase 2
AYB AI transitions into a monthly consultancy role, manually reviewing calls and providing performance recommendations based on production data.
Why This Approach Makes Sense
EnergyPal has already made a strong start and is best positioned to handle CRM/API work internally.
AYB AI can focus on the voice-agent layer where production experience matters most.
EnergyPal gets a clearer template for how production-ready Retell agents should be structured.
Future recommendations become easier to implement because the initial Retell setup will already be tightened.
Phase 1 - Retell Production Readiness Sprint
Timeline: 1–3 Days
AYB AI will jump into EnergyPal's existing Retell setup and help get the inbound AI agent production-ready. The scope below covers all key areas of the Retell voice-agent setup.
1
Agent Structure Review
Review the existing Retell agent structure and tighten the global prompt to align with EnergyPal's tone and call strategy.
2
Logic Improvements
Improve transition logic between sections and sub-flows. Improve fallback handling when the caller gives unclear, unexpected, or incomplete answers.
3
Sub-Agent & Extraction
Strengthen sub-agent prompts. Refine qualification, booking, callback, and support triage logic. Set up or improve key extraction fields.
4
Knowledge-Base & Tracking
Review the current knowledge-base approach and recommend a better structure if possible. Add performance tracking logic and improve call outcome/disposition capture.
5
Voice & Conversation Quality
Improve prompt quality around objectives, answer length, interruption handling, and human-like pacing. Ensure the Retell-side setup is ready for EnergyPal's CRM/API calls.

Important: EnergyPal will continue handling CRM/API work internally. AYB AI's Phase 1 work is focused exclusively on the Retell voice-agent setup, prompts, flows, extraction fields, routing logic, and performance-readiness.
Phase 2 - Monthly Consultancy & Call Review
Once the agent is live, AYB AI will move into the agreed consultancy model, providing ongoing manual call review and performance optimization recommendations.
Call Recording Review
Manually reviewing inbound AI call recordings and transcripts. Identification of failure points, caller friction, and missed opportunities.
Routing & Flow Optimization
Recommendations to improve agent structure, call routing, qualification, booking, and fallback logic. Recommendations for improving sub-agent flows.
Caller Experience
Review of caller objections, confusion points, and drop-off moments. Recommendations for improving voice, tone, pacing, and interruption sensitivity.
Performance Summaries
Review of extraction quality and CRM-update quality where call data is available. Performance summaries with recommended actions for EnergyPal's internal team.
Retell Optimization Areas
The following areas were identified as important for getting the current agent production-ready:
Global Prompt
  • Improve the master prompt
  • Ensure the agent follows EnergyPal's preferred tone and call strategy
  • Reduce unnecessary long explanations
Transition Logic
  • Improve handoffs between sections
  • Prevent loops and awkward or repetitive transitions
  • Ensure the agent knows when to move forward vs ask a clarifying question
Fallback Logic
  • Improve how the agent handles unclear answers
  • Improve recovery when a caller interrupts or changes direction
  • Improve behavior when required information is missing
Sub-Agent Prompts
  • Strengthen prompts for each triage path
  • Ensure each sub-agent has a specific purpose and does not over-question the caller
  • Keep flows tight and outcome-focused
Extraction Fields
Set up key fields for reporting, CRM/API handoff, and post-call analysis. Capture structured data such as caller type, intent, qualification outcome, booking status, and next step.
Knowledge-Base Approach
Review whether Retell's built-in knowledge base is the right approach. Recommend a better alternative. Keep company-specific details inside controlled flow logic where appropriate.
Performance Tracking
Add clear tracking fields and disposition logic. Build toward dashboard-style reporting similar to the Billing AI performance dashboard example shared. Track all performance outcomes.
Voice & Conversation Quality
Tune response length and interruption sensitivity. Improve pacing. Ensure the experience feels fast, useful, and aligned with EnergyPal's brand.
Integrations & Data Ownership
EnergyPal will continue to manage CRM/API-side development internally. This clear division of responsibility allows both teams to focus on their strongest areas.
EnergyPal-Owned Scope
  • CRM/API calls
  • Lead lookups
  • Appointment booking APIs
  • CRM updates
  • Internal disposition mapping
  • Data routing and backend logic
AYB AI-Owned Scope
  • Retell agent review and optimization
  • Prompt structure
  • Sub-agent logic
  • Key extraction fields
  • Voice-agent behavior
  • Performance tracking
  • Manually reviewing calls
  • Recommendations based on production performance
Performance Tracking & Reporting
AYB AI recommends that EnergyPal tracks performance from the start so optimization decisions are based on real call data and trends.
The goal is to create a performance feedback loop where real calls are reviewed, trends are identified, and the agent is improved continuously.
Volume Metrics
  • Total inbound calls handled by AI
  • Caller type
  • Call outcome
Outcome Metrics
  • Booked consultation count
  • Disqualified count
  • Callback request count
  • Support case count
  • Transfer request count
Failure Metrics
  • Caller hung up before outcome
  • Caller asked for human
  • No CRM match found
  • Booking conversion rate
  • Common failure reasons
Implementation Plan & Timeline
1
Access & Review
EnergyPal provides AYB AI access to the Retell agent, relevant context around current build structure, and confirms CRM/API responsibilities remain internal while sharing CRM access for testing.
2
AYB AI Retell Production Readiness Sprint (1–3 days)
Review and tighten global prompt. Improve sub-agent prompts. Improve transition/fallback logic. Review knowledge-base structure. Set up or refine key extraction fields. Improve tracking/disposition logic. Test the Retell agent with EnergyPal's CRM/API connections.
3
EnergyPal Internal Testing
EnergyPal tests the updated Retell agent internally. AYB AI adjusts based on test outcomes. EnergyPal confirms readiness for live inbound call routing.
4
Go-Live
EnergyPal routes live inbound calls to the agent. AYB AI begins monthly consultancy and manual call review.
5
Monthly Optimization
AYB AI manually reviews all calls. AYB AI sends recommendations based on call performance findings. EnergyPal implements relevant changes internally.
Phase 1 Deliverables
1
  • Review of existing Retell agent
  • Production-ready global prompt improvements
  • Updated transition and fallback logic
  • Strengthened sub-agent prompts
  • Key extraction field recommendations/setup
  • Knowledge-base approach review
  • Performance tracking improvements
  • Retell-side production readiness improvements
  • Short handover summary outlining what was changed and why
2

Phase 2 Deliverables
  1. Manually reviewing all calls
  1. Performance recommendations
  1. Failure-point analysis
  1. Prompt and flow improvement recommendations
  1. Monthly optimization notes
  1. Suggested changes for EnergyPal's internal technical team
  1. Strategic guidance based on AYB AI's industry experience
Pricing & Terms
Phase 1 - Retell Production Readiness Sprint
$1,500
One-time fee - Completion target of 1–3 days, depending on access and complexity.
  • Review of EnergyPal's existing Retell setup
  • Direct Retell-side optimization
  • Production-readiness improvements
  • Prompts, flows, fallbacks, extractions, and tracking
Phase 2 - Monthly Consultancy
$1,500/mo
Monthly consultancy fee - Completion target of 3–6 months, depending on time to optimal performance.
  • Manual reviewing all calls
  • Performance analysis
  • Recommendations to improve AI call handling
  • Reporting and support via agreed communication channels

CRM/API work: Handled by EnergyPal internally and excluded from AYB AI's Phase 1 fixed fee unless separately scoped.
AI Usage & Billing
EnergyPal's Usage
EnergyPal will continue using its own Retell environment and related accounts. Retell usage, phone usage, and any related platform/vendor costs are not included in AYB AI's fees and will be managed by EnergyPal directly.
What AYB AI's Fees Cover
AYB AI's fees cover Retell optimization, production-readiness work, consultancy, call review, and performance recommendations - not platform or vendor related costs.
Support & Communication
Support Hours
Monday - Friday
9am–5pm EST
Zoom
Video calls for sprint reviews, testing sessions, and consultancy check-ins.
Slack
Faster day-to-day communication via the AYB AI Slack channel.
Assumptions, Risks & Mitigations
Assumptions
  • EnergyPal provides Retell access and necessary permissions
  • EnergyPal continues to own CRM/API development
  • EnergyPal provides further context when asked about current Retell setup and desired behavior
  • English-only scope
Next Steps
Approve the Updated Proposal
Approve the updated proposal internally.
Finalize Invoice
Finalize invoice for the phase 1 build fee and first month of consultancy ($3,000).
Provide Retell Access
Provide AYB AI admin access to EnergyPal's Retell environment.
AYB AI Completes Production Readiness Sprint
We complete the Retell Production Readiness Sprint within 1–3 days. EnergyPal tests internally and prepares for live inbound routing.
Begin Monthly Consultancy
Once live, AYB AI begins the consultancy model for manual call review and ongoing performance recommendations.