call-processing

name	call-processing
description	Fires when a call transcript is pasted or the user says they finished a call. Produces: 6 structured outputs — MEDDPICC scoring with evidence, task extraction with owners and dates, coaching feedback on 5 dimensions, follow-up email in 3 tones, risk signals with mitigation actions, competitive intel. Output contract: every output is actionable, not just analytical.
user-invocable	false

Call Processing — 6-Output Framework Skill

Purpose

Defines the 6-Output Framework used by call processing to process call transcripts. Contains detailed formats for each output: MEDDPICC scoring, task extraction, coaching feedback, follow-up email, risk signals, and competitive intelligence. Also defines the 5 coaching dimensions and 6 risk signal categories.

When Referenced

• Processing a call (process-call intent) — the primary consumer. Fires whenever a call transcript is provided or discussed, processing every call through the full 6-Output Framework.
• Reviewing coaching patterns (improve-skills intent) — uses the 5 coaching dimensions for pattern analysis across multiple calls

Core Framework

The 6-Output Framework

Every call processed through call processing produces exactly 6 outputs. This is non-negotiable — every call gets the full framework, even if some outputs are thin.

Output 1: MEDDPICC Scoring

Score each MEDDPICC element based on what was said in the call. Reference the meddpicc/SKILL.md for scoring rubric.

Format:

MEDDPICC UPDATE:
| Element | Before | After | Evidence |
|---------|--------|-------|----------|
| M-Metrics | [R/Y/G] | [R/Y/G] | "[exact quote from call]" |
| E-Economic Buyer | [R/Y/G] | [R/Y/G] | "[exact quote]" or "No change — not discussed" |
| D-Decision Criteria | [R/Y/G] | [R/Y/G] | "[exact quote]" |
| D-Decision Process | [R/Y/G] | [R/Y/G] | "[exact quote]" |
| P-Paper Process | [R/Y/G] | [R/Y/G] | "[exact quote]" |
| I-Identify Pain | [R/Y/G] | [R/Y/G] | "[exact quote]" |
| C-Champion | [R/Y/G] | [R/Y/G] | "[exact quote]" |
| C-Competition | [R/Y/G] | [R/Y/G] | "[exact quote]" |

Net change: [X elements improved, Y unchanged, Z declined]

Rules:

• Only upgrade an element if there's a specific quote or evidence from the call
• If an element wasn't discussed, mark "No change — not discussed"
• Never upgrade an element based on inference — only on stated evidence
• If an element should downgrade (e.g., champion went quiet), flag it with the evidence

Output 2: Task Extraction

Extract every action item, commitment, and follow-up from the call.

Format:

TASKS EXTRACTED:
| Task | Owner | Priority | Due Date | Source |
|------|-------|----------|----------|--------|
| [Specific action] | AE/Prospect/Internal | H/M/L | [date] | Call |
| [Specific action] | AE/Prospect/Internal | H/M/L | [date] | Call |

Rules:

• Owner categories: AE (the user must do this), Prospect (the customer committed to this), Internal (someone else on user's team)
• Priority: High = blocks deal progress, Medium = important but not blocking, Low = nice-to-have
• Due Date: Extract from the call if mentioned ("by Friday," "next week"), otherwise default to 3 business days
• Be specific. Not "follow up" — instead "Send ROI calculator with Q3 case study data to Sarah by Thursday"
• Track both AE commitments AND prospect commitments — hold both sides accountable

Output 3: Coaching Feedback

Score the AE's performance on 5 dimensions. This is about improving the AE's craft, not judging them.

The 5 Coaching Dimensions:

Dimension	What It Measures	1 (Needs Work)	3 (Solid)	5 (Exceptional)
Discovery	Quality of questions	Surface-level, yes/no questions only	Multi-layered questions that uncover needs	Expertly layered discovery revealing root pain and business impact
Objection Handling	Response to resistance	Avoids objections or caves immediately	Acknowledges and provides reasonable responses	Acknowledges, reframes brilliantly, advances the conversation
Rapport	Relationship building	Purely transactional, no personal connection	Professional with appropriate warmth	Genuine connection, active listening, builds real trust
Next Steps	Commitment clarity	Vague "let's follow up" or no next step	Clear next step with general timeline	Specific mutual commitments with dates, owners, and accountability
Talk Ratio	Listening balance	AE dominates (>70% talking)	Balanced (~50/50)	Prospect-led (AE <40%), AE asks and listens

Format:

COACHING FEEDBACK:
| Dimension | Score | Observation |
|-----------|-------|-------------|
| Discovery | [1-5] | [Specific observation from this call] |
| Objection Handling | [1-5] | [Specific observation] |
| Rapport | [1-5] | [Specific observation] |
| Next Steps | [1-5] | [Specific observation] |
| Talk Ratio | [1-5] | [Estimated: AE ~X%, Prospect ~Y%] |

Overall: [X.X]/5
Top strength: [dimension] — [why]
Growth opportunity: [dimension] — [specific, actionable suggestion]

Rules:

• Be honest but constructive — this is coaching, not criticism
• Always identify at least one strength and one growth area
• Tie observations to specific moments in the call when possible
• The overall score is the average of all 5 dimensions

Output 4: Follow-Up Email

Draft a follow-up email based on the call conversation.

Format:

FOLLOW-UP EMAIL:

Tone options:
1. [Professional] — formal, appropriate for economic buyers
2. [Warm] — friendly, appropriate for champions
3. [Concise] — minimal, appropriate for technical evaluators

Selected: [default based on relationship/contact sentiment]

SUBJECT: [lowercase, specific to what was discussed]

BODY:
[Short follow-up. References specific things discussed.
Confirms commitments made on both sides.
One clear next step.]

[{{EMAIL_SIGN_OFF}}]

Rules:

• Apply brand-voice skill rules (vocabulary substitutions, banned phrases)
• Reference specific things from the call — not generic "great talking to you"
• Confirm commitments from BOTH sides (AE and prospect)
• One clear next step or CTA
• Under 200 words

Output 5: Risk Signals

Analyze the call for risk indicators across 6 categories.

The 6 Risk Signal Categories:

Category	High Risk	Medium Risk	Low Risk
Champion Health	Champion went silent, deferred to others, showed uncertainty	Champion less enthusiastic than last call	Champion engaged and advocating
Timeline Pressure	Customer pushed timeline out, mentioned "no rush"	Timeline unchanged but no urgency signals	Customer expressed urgency, mentioned deadlines
Competitive Threat	Competitor mentioned favorably, prospect comparing features	Competitor mentioned but neutrally	No competitor mentions, or prospect dismissed competitor
Budget Concerns	Budget cut, freeze mentioned, "too expensive"	Budget questions without commitment	Budget confirmed, ROI discussion positive
Stakeholder Shifts	Key stakeholder left, reorg mentioned, new decision maker	Stakeholder roles unclear or changing	Stakeholders stable and engaged
Engagement Quality	Short answers, distracted, rescheduled multiple times	Adequate but not enthusiastic	Deep engagement, asking detailed questions, proactive follow-up

Format:

RISK SIGNALS:
| Category | Level | Evidence |
|----------|-------|----------|
| Champion Health | [H/M/L] | [specific evidence from call] |
| Timeline Pressure | [H/M/L] | [specific evidence] |
| Competitive Threat | [H/M/L] | [specific evidence] |
| Budget Concerns | [H/M/L] | [specific evidence] |
| Stakeholder Shifts | [H/M/L] | [specific evidence] |
| Engagement Quality | [H/M/L] | [specific evidence] |

Overall Deal Risk: [High/Medium/Low]
[If High]: ⚠️ Recommend running deal strategy for [Company] for deeper analysis.

Output 6: Competitive Intelligence

Extract any competitive mentions or intelligence from the call.

Format:

COMPETITIVE INTEL:
- Competitor mentioned: [name] — Context: "[what was said]"
- Competitor positioning: [how they're being presented to the prospect]
- Our advantage: [where we're strong based on this conversation]
- Our vulnerability: [where they're strong or we're weak]
- Win probability: [0.0-1.0] — [brief rationale]

[If competitor identified]: 💡 Run competitive analysis for [Company] for full displacement analysis.

Rules:

• Only report competitor mentions that actually appeared in the call
• If no competitors mentioned, say "No competitive mentions in this call"
• Win probability is always a hypothesis — evidence-grade it
• Don't fabricate competitive intelligence from assumptions

Write Mapping

Call processing writes to 4 data types:

1. Call Notes — new record with all 10 fields
2. Tasks — new records for each extracted task
3. Deals — update MEDDPICC scores, MEDDPICC Confidence, Deal Risk, Last Activity
4. Companies — update Last Activity

Duplicate check (before writing): Search data/call-notes/ for existing notes matching the same company + date. If found, present the existing note with a file link and ask before overwriting.

File link output (after writing): Every call processing execution must end with a FILES UPDATED block listing every file created or updated, with computer:// links. See CLAUDE.md Section 10.4 Rule 10.

Personalization Notes

• Follow-up email applies brand-voice skill (Tier 3 — regenerated during deep personalization)
• Email sign-off uses {{EMAIL_SIGN_OFF}}
• Coaching dimensions are universal — scoring scales are not personalized
• Risk signal categories are universal — definitions are not personalized
• MEDDPICC scoring references meddpicc/SKILL.md (universal)

First-Use Calibration Storage (Call Processing)

On the first run of call processing, the user is asked which call elements matter most to them. The 6-Output Framework structure is universal, but the presentation order and emphasis are personalized:

• Output priority weighting: [Populated on first call processing run — ordered list of which outputs the AE values most, e.g., "MEDDPICC scoring > follow-up email > risk signals > tasks > coaching > competitive intel"]
• Emphasis notes: [Any specific preferences, e.g., "Always show risk signals first" or "I care most about the follow-up email"]

This does not change the framework itself — all 6 outputs are always generated. It changes which outputs are presented first and given the most detail.