The breakpoint
The moment most likely to have changed the direction of the call, with the prospect’s exact words and an explanation of why it mattered.
ONE CALL · EXACT EVIDENCE · NEXT-CALL PLAN
Stop guessing why the buyer said no.
Your rep remembers one version of the call. The transcript contains another. QC 4.1 Light finds the exact moment the conversation changed, shows the evidence and turns it into a replay plan for the next call.
OPERATOR KNOWLEDGE→AI SYSTEM→PROOF
THE SHORT ANSWER
AI call summaries tell you what was said. This tells you what to fix.
QC 4.1 Light is a free sales-call analysis skill for AI agents. Give it one transcript and it returns a structured diagnosis grounded in exact quotes. The goal is not to sound insightful. The goal is to show where the conversation broke, why that mattered and what the seller should say differently next time.
WHAT YOU GET
A review your team can challenge, not another confident AI opinion.
Every important finding carries the excerpt that supports it. If the call does not contain enough evidence, the report has to say so.
Three mistakes
Observable seller actions, not personality labels. Each mistake includes its consequence and a concrete correction.
Objection evidence
Price, trust, implementation or another concern only appears when the transcript supports it. Uncertainty remains visible.
A replay plan
One opening question, one diagnostic follow-up, one objection clarifier and one next-step line for deliberate practice.
SYNTHETIC EXAMPLE · SAFE DATA
The prospect named the real risk. The seller pitched past it.
This example comes from the synthetic transcript shipped with the skill. No customer or private call data is used.
“We tried an automation tool last year and the team stopped using it.”PROSPECT · 00:38
DIAGNOSIS
The seller answered a previous adoption failure with “our AI is much better” instead of finding out why the team stopped using the last tool. The implementation risk stayed unresolved before price appeared.
RECOVERY LINE“Before we talk about a discount, can we go back to what failed last time and define what your team would need to adopt this successfully?”
CHOOSE YOUR FASTEST PATH
Get the review. Keep the level of control you want.
You do not need to become a developer to inspect one call. Use the product for the shortest path, or install the open skill inside an agent you control.
Analyze the call in Closing Code AI.
Upload audio or paste a transcript. No card required. Use this path when you want the diagnosis without cloning a repository or maintaining an agent harness.
Audit my first call free →git clone https://github.com/mgallmur-glitch/qc41-light.git ~/.hermes/skills/qc41-lightRun it in Hermes, Claude Code, Codex or a generic agent harness. No QC account, uploader or hosted API is required.
Install from GitHub →WHEN TO USE IT
Useful for one call. Honest about where it stops.
Use it after discovery calls, demos, enrollment calls or high-ticket sales conversations when you have a transcript and need specific feedback.
Good fit
- You already record or transcribe sales calls.
- You want exact evidence behind coaching feedback.
- You use Hermes, Claude Code, Codex or another agent harness.
- You need a repeatable review in English or Spanish.
Not a fit
- You need live persuasion or autonomous outreach.
- You expect a universal closer score from one conversation.
- You do not have permission to analyze the transcript.
- You need longitudinal team analytics or managed coaching.
PRIVACY
The package does not send your transcript anywhere.
QC 4.1 Light is files, prompts, a schema and local validation scripts. It has no uploader, telemetry or API client. But the model you choose may process data remotely. Remove personal information and use only recordings you are authorized to analyze.
FAQ
Before you install it.
What does QC 4.1 Light analyze?
One sales-call transcript at a time. It reviews transcript quality, call stages, strengths, the most likely breakpoint, objections, observable mistakes, corrections and the next-call replay plan.
Does it upload my call?
No. The skill contains no uploader, telemetry or hosted API client. Your transcript is processed by the AI agent or model you choose, so its privacy policy still applies.
Which AI agents can use it?
The repository includes instructions for Hermes Agent, Claude Code, Codex and generic agent harnesses. The analysis contract works in English and Spanish.
Is this a call scorecard?
No. It does not produce a universal score or pretend to measure hidden buyer psychology. It makes bounded findings that must point to exact transcript evidence.
Can I use it with real customer calls?
Only with authorization and appropriate data handling. Remove names, emails, phone numbers and payment details before sending a transcript to a hosted model.
Review the call before the next coaching session.
Put the buyer's exact words, the breakpoint and the replay plan in front of the seller.