MSP Playbook: A Practical Guide to Starting Your Customers on AI (the Right Way)

1) Ground Rules: Automate Without Dehumanizing

What to automate: repetitive, predictable, information-extraction or drafting tasks.

What to protect for humans: judgment, context, empathy, negotiation, and exceptions.

Prompt the customer:

• “What work steals hours but doesn’t require a human relationship?”

• “If your team had 10% of their time back, where would you redeploy it?”

SMB starter examples with Hatz AI:

• Auto-summaries of meetings, calls, or weekly ops huddles → 10 bullet points + action items.

• First-drafts of routine emails/proposals → humans polish and send.

• Data cleanup (CSV/Excel) and simple analysis → quick insights without analyst cycles.

Thought-provoker: If you insist on humans doing low-value tasks, you’re choosing slower service and higher costs.

2) The Executive Litmus Test (Don’t Skip This)

Make leadership usage non-negotiable. Adoption follows the org chart.

Ask and confirm in writing:

• “Which two executives will personally use Hatz AI this month?”

• “What weekly rhythm will they adopt? (e.g., 3 prompts/day, weekly reflection email)”

• “Who is the named executive sponsor for this rollout?”

Quick “Exec Use” play:

• 15-minute install + 10 sample prompts aligned to each exec’s real work (board notes, pipeline questions, variance explanations, customer emails).

• A 3-week commitment: use daily, share one win per week.

• MSP runs a short read-out: “What the execs learned, and where we scale next.”

Thought-provoker: If leaders won’t use it, pause the project. It will become shelfware (and you’ll be blamed).

3) Workflow-First Integration (Find the 10% that saves 40%)

Map the top 5–10 workflows in each function (sales, service, finance, ops).

For each workflow, run this checklist:

1. Decompose: What are the 10–50 steps?

2. Classify: Which steps are AI-able vs. Human-critical?

3. Data Check: Does the step involve sensitive data? If yes, use guarded patterns (e.g., redact, private docs, role-based access).

4. Speed to Value: Which 3 steps, if automated, unlock the most time in 30 days?

5. Guardrails: What approval/QA is needed before output is customer-facing?

Hatz AI patterns you can deploy:

• Chat for on-demand drafting, summarizing, and analysis.

• Workflows to chain steps (ingest → transform → draft → route for review).

• Integrations to pull/push from tools like CRM, sheets, docs, or ticketing.

• Role-aware access so sensitive content stays in the right lanes.

Thought-provoker: Don’t start with “AI features.” Start with time thieves.

4) Crawl → Walk → Run (with Real SMB Use Cases)

Crawl (Weeks 1–4): Safe, obvious wins

• Sales & CS: Draft follow-ups from call notes; summarize deals; propose next steps.

• Ops: Clean CSVs; generate weekly ops summaries; create SOP first drafts.

• Finance/Admin: Draft variance commentary; summarize invoices; vendor email replies.

Hatz AI moves: Chat + simple Workflows. No sensitive data yet.

Metric: Hours saved; % of drafts accepted with minor edits.

Walk (Months 2–3): Team workflows & light integrations

• Marketing: Repurpose one long asset (webinar/podcast) into blog, email, social; weekly content calendar suggestions.

• Service Desk: Ticket triage summaries; knowledge snippets from resolved tickets.

• Sales Ops: Territory/update summaries; pipeline hygiene nudges; proposal templating.

Hatz AI moves: Workflows + Integrations (docs, CRM, ticketing).

Metric: Cycle time reduction; volume handled per FTE; SLA adherence.

Run (Month 4+): Cross-department automation & agents

• Revenue Ops: Lead routing explanations; QBR packet assembly from dispersed data.

• Finance: Month-end close narratives; recurring vendor communications.

• Ops/HR: Policy Q&A on internal documents; onboarding checklists.

Hatz AI moves: Multi-tool Workflows, role-aware access, specialized AI agents for repeatable tasks with human approval gates.

Metric: Cost-to-serve reduction; employee NPS; customer NPS.

‍

Thought-provoker: “Run” isn’t more prompts—it’s fewer steps.

5) 30/60/90 Adoption Plan (You Can Copy/Paste This)

Days 0–30 (Crawl)

• Name executive sponsor + two exec users (usage commitment in writing).

• Enable Hatz AI for pilot teams (sales, CS, ops).

• Ship 3 quick-win workflows; measure time saved.

• Weekly 30-minute stand-up: wins, blockers, next two automations.

Days 31–60 (Walk)

• Add light integrations (docs/CRM/ticketing).

• Expand to one cross-team workflow (e.g., QBR prep).

• Draft AI usage guardrails: data handling, approval steps, output QA.

• Publish a “What Works Here” internal guide (best prompts + patterns).

Days 61–90 (Run)

• Stand up 1–2 role-aware agents for repeatable tasks with human review.

• Roll out success metrics to leadership dashboards.

• Executive retro: what we keep, what we stop, where we invest.

6) Success Scorecard (Define “Good” Up Front)

Track five things every week:

1. Executive Usage: # of exec prompts, # of shared wins.

2. Time Saved: Hours recovered (self-reported + observed cycle times).

3. Quality & Trust: % of AI drafts accepted with minor edits.

4. Adoption Breadth: # of active users per function; repeat usage.

5. Business Outcomes: Faster response times, higher throughput, lower cost-to-serve.

Thresholds to watch: If drafts are heavily rewritten, either prompts are weak, data is missing, or the task shouldn’t be automated.

7) Governance That Doesn’t Kill Momentum

Practical guardrails:

• Redact PII/sensitive fields in early phases; use private docs and role-based access as you mature.

• Human review before anything customer-facing (proposal, legal, pricing).

• Log prompts/outputs for audit on sensitive workflows.

• Keep a “do not automate” list (e.g., layoffs communications, legal admissions).

‍

Thought-provoker: Governance should be a seatbelt, not a parking brake.

8) Anti-Patterns to Call Out (Gently but Firmly)

• “We’ll wait for the big platform rollout.” Translation: no value for 6–9 months. Start small now.

• “We’ll hire an AI team first.” Start by freeing time in current teams; specialists come later.

• “Let’s automate everything.” You’ll erode trust. Keep humans in the loop.

• “Leaders don’t need to use it.” They do. Or the project dies.

9) First Conversation Script (Use This on Your Next Call)

Openers

• “What are the 3 most annoying, repetitive tasks your team does weekly?”

• “If we gave you back 10% of team time in 30 days, where would you spend it?”

Executive asks

• “Which two executives will personally use Hatz AI for three weeks?”

• “Can we schedule a 20-minute exec ‘wins’ read-out at the end of each week?”

Pilot scope

• “Let’s pick 3 workflows we can fix in 14 days and agree on how we’ll measure success.”

10) How to Start (Hatz AI in Four Steps)

1. Turn it on: Create the tenant and invite pilot users.

2. Load context: Connect relevant docs/sheets (non-sensitive first).

3. Ship three wins: Meeting summaries, routine email drafts, CSV cleanup.

4. Review weekly: What saved time? What needs a guardrail? What’s next?

Your north star: Fewer steps, faster cycles, more human time where it matters.

Final Word

AI isn’t about replacing people; it’s about removing the work that keeps people from being great. As the MSP, your value is to set the bar high on executive engagement, pick winnable workflows, install lightweight guardrails, and prove value fast. Do that, and the rest takes care of itself.

‍