Enterprise Automation Pack

Hi — I'm Muhammad from consultance.ai. This guided setup gets you from "I just opened the pack" to a working first automation in one session. One step at a time.

First question before anything else:

**What type of automation work are you starting with today?**

(A) Full enterprise transformation — I want to map every workflow worth automating, then blueprint and build
(B) Single function — I want to automate one specific function: finance close, CRM/pipeline, IT service desk, or reporting (tell me which)
(C) Quick win — I want to find the 2-3 highest-ROI automations I can deploy this month
(D) Governance first — I have automations running but they are fragile or undocumented, I want a debt audit and approval framework

Reply A, B, C, or D and I will route you to the right starting prompt and walk you through it.

---

Once you reply, I will ask one more question about where your process documentation lives — SOPs, workflow descriptions, or process maps you can paste or upload. Then we start.

**Setup (do this once before we begin):**
Open Claude at claude.ai. Create a new Project and name it "Automation Desk." This is your private workspace. Everything you paste or upload stays in your own Claude tenant — nothing is seen by or stored by consultance.ai.

Pin Opus 4.8 as your model for the first session.

---

**Path A: Full transformation**

After your answer, we will run in this order:
1. Prompt 01 (Desk Router) — sets the scope, classifies your data source, confirms assumptions
2. Prompt 02 (Workflow Audit) — ranks every process worth automating by composite score
3. Prompt 03 (Bottleneck Miner) — finds the real constraint before you touch a tool
4. Prompt 12 (Pilot Design) — plans the 30-day first sprint before any building starts
5. ROI baseline (Prompt 11) — captures before-state numbers BEFORE anything goes live

Then: blueprinting and building for the pilot workflow.

First session target: a ranked list of your top 10 workflows with a named pilot candidate and a captured before-state baseline.

---

**Path B: Single function**

Tell me the function (finance close, CRM/pipeline, IT service desk, or reporting) and I will route you directly to the matching prompt with the right context questions. We will have a working blueprint in 45 minutes.

Function-specific prompts:
- Finance close: Prompt 06 (AP, reconciliation, journal entries, close checklist)
- CRM/pipeline: Prompt 07 (lead routing, stage gates, follow-up sequences)
- IT service desk: Prompt 08 (ticket triage, routing, SLA, resolution automation)
- Reporting: Prompt 09 (dashboards, knowledge base, stale article detection)

Before any function-specific prompt: run Prompt 10 (approval gate design) — every automation touching money, customers, or access needs a named human gate before it goes live.

---

**Path C: Quick win**

We run Prompt 02 (workflow audit) across the functions you describe, and I score every process you name on volume, time cost, error risk, and automation fit. Output: top 3 ranked with a rationale for each. Then Prompt 12 (pilot design) for the #1 pick.

First session target: a ranked quick-win list you can present to your team this week.

---

**Path D: Governance first**

We run Prompt 15 (quarterly debt audit) to surface what is broken or orphaned, then Prompt 10 (approval gate design) to formalize the control layer on everything that is live. Output: a debt register with severity ratings and a named fix owner for each item.

First session target: a debt register table and an approval matrix ready for controller review.

---

**Model guidance:**
- Opus 4.8: Prompts 02, 03, 04, 10, 11, 15, Bonus 01 (analysis and judgment-heavy)
- Sonnet 4.6: Prompts 05, 06, 07, 08, 09, 12, 13, 14 (structured output, lower complexity)
- Prompt 01: either model

**The approval gate rule:**
Every `<review_gate>` tag in the vault is a real control. A named human on your team must sign off before any automation goes live that touches money, sends external communications, or modifies customer or access data. The prompts design the gate; you enforce it.

**If a prompt output surprises you:**
Paste the output back and tell me what was unexpected. I will diagnose whether it is a data quality issue, a branching assumption that needs correction, or something the prompt genuinely cannot handle alone.

Use the live ROI Calculator at consultance.ai/library/enterprise-automation-pack to run your before and after numbers as you go.

Related: [[ai-discovery-playbook]] · [[customer-zero-finance-ops]]