Build an AI Custom Automation - Meeting prep account summary

Meeting prep shouldn’t be guesswork. This automation eliminates the scramble for context by automatically analyzing account data, conversations, and stakeholder insights. The result is a concise, AI-powered summary that shows what’s happening in the account, who matters most, and where to focus next — helping CSMs engage with precision and impact ahead of customer meetings.

20 seconds

Planhat

This guide is interactive, designed for you to follow along and build automations step-by-step. As we go through each section, we recommend recreating the setup in your own Planhat environment simultaneously. Each organisation will have their own specific needs, so use the steps as inspiration to create an automation that works for your business.

Summary: Build an AI meeting prep account summary

**Step 1: Decide the outcome and goal** - **Outcome:** Enable consistent, insight-led customer engagements that strengthen relationships and position CSMs as strategic partners — driving higher retention, expansion, and satisfaction - **Goal:** Build an AI-powered system of record for account intelligence, giving every team member unified context and insights to scale growth efficiently **Step 2: Identify the inputs** - Trigger: manual trigger on Company level - Get: Conversations, End User, Licence, Issues and NPS **Step 3: Design the AI step** - Use Planhat's Managed LLM and GPT-4 - Build a prompt that acts and reasons as an analyst for the entire account **Step 4: Decide the actions** - Create a Conversation called "Meeting prep" available on the Company profile **Step 5: Test and Iterate** - Test until you are happy with the output

Step 1: Decide the outcome and goal

CSMs spend valuable time piecing together fragmented information before customer meetings. Without a clear view of the customer’s strategy, stakeholders, and overall health, conversations remain tactical rather than value-driven. **Outcome:** Consistent, insight-led customer engagements that strengthen relationships - leading to higher customer satisfaction. It also frees up your team's time to position them as strategic partners - hopefully leading to higher retention and expansion. **Goal:** Establish an AI-powered system of record for account intelligence, ensuring every team member has the same context, priorities, and insights to drive growth at scale.

Step 2: Identify the inputs

Now that you know what you want to achieve, let's think about **what data the AI needs**. The goal is to give the AI enough context to summarize the account accurately and highlight what the CSM needs to know ahead of a meeting. For this use case, good inputs could be: - **Account data:** Core company information such as name, size, industry, products in use, NPS, usage metrics, - **Conversations:** Recent conversations and support - **Stakeholder data:** Known contacts, their roles, engagement history, and any notes on sentiment or influence (champion, blocker, etc.) - **Internal interactions:** Notes or transcripts from recent CSM or Sales conversations to capture tone, priorities, and ongoing issues - **Opportunity & renewal details:** Current contract, license type, ARR, renewal date, and success criteria defined during sales or onboarding - **Product signals:** Health scores, feature adoption data, or product usage trends that indicate risk or growth potential The automation surfaces key account health, stakeholder dynamics and engagement recommendations—helping CSMs prepare faster, prioritize effectively and engage with greater strategic impact.

**Triggering** this automation is best activated **manually** — allowing the CSM to generate fresh insights directly within the Company Profile right before a customer meeting.

Now let's get some data into the automation! Here, we want to bring in quite a lot of different data points, based on what data you decided you want to bring in. Whatever you feel is vital for your CSM to consider, you can add as input.\ \ Here's an example.

Each of these steps, will need to be set up & filtered as follows: **Object - Multiple** This collects multiple objects, which is good in this case, as we want a summary of everything in the account, not just a single object. **Filter: Match all - Company ID Equal To <<object.\_id>>** This filter ensures you collect data only from the specific company you are working on.

For each **Get step,** you can set **parameters** to, for example, retrieve only the latest 10 conversations or only a specific type of Conversation. This helps avoid overwhelming the AI with too much data.

Step 3: Design the AI step

Next up is the heart of the automation - it's time to configure the **AI step**! In this example, we’re using **Planhat’s Managed LLM**, but you can also select your own model from the **“Provider”** dropdown if you’ve connected an external LLM. The model choice is up to you. For tasks like summarizing lots of data, models such as **ChatGPT**, **Claude**, or **Gemini** all perform well. We recommend experimenting with a few models to find the one that delivers the most accurate and relevant summaries for your meeting prep notes. It's worth noting they all have different usage of AI credits / tokens, so choose the best one for your use case!

Now to the **prompt**! There are many ways to set up your prompt, but as mentioned before, the more instructions, the better! It's worth spending time testing different prompts until you are satisfied with the output. Treat the AI as a new colleague who needs all the context to do its job. Below you can see the prompt we used for this specific use case, as a starting point: ### **Role & Objective:** Act as an analyst building an Account & Stakeholder Insights Report for **<<object.\_id>>.** \ Your objective is to synthesize internal data to produce actionable insights that help Customer Success Managers (CSMs) prioritize engagement, strengthen stakeholder relationships, and identify growth and risk signals across the account. ### **Data Set:** End User: <<s-Wn4>>\ License: <<s-Wn3>>\ Conversations: <<s-v7V>>\ NPS: <<s-8DK>>\ Issues: <<s-bIX>> ### **Reasoning Steps:** 1. **Clarify Inputs:** Review all provided data objects and confirm what’s missing 2. **Account Overview:** - Summarize key strategic initiatives, licenses, NPS trends, issues, and recent conversations 3) **Account Landscape & Opportunities:** - Highlight operational, financial, or technical challenges - Map opportunities where our solution could add measurable value or mitigate risk 4) **Stakeholder Mapping:** - Identify stakeholders from the data, their titles, and roles Categorize them as Decision-makers, Influencers, Champions, or Blockers - Summarize engagement history and assess sentiment (only point out positive or negative, ignore neutral) 5) **Engagement Insights:** - Recommend tailored engagement strategies per stakeholder group - Suggest key questions or topics to explore in upcoming meetings - Identify relationship risks, missing champions, or coverage gaps 6) **Action Plan:** - Recommend concrete next steps to improve engagement and relationship health - Prioritize which stakeholders or initiatives require immediate attention - Define 3–5 measurable indicators of account health and stakeholder alignment ### **Instructions:** - Always separate data-derived insights from web-research insights - Write in a professional, concise, and analytical tone - Use structured formatting: clear headings, bullet points, and short paragraphs - Avoid filler language or assumptions — only interpret what’s explicitly supported by the data - Do not include greetings (“Sure, here’s your answer...”) or closing remarks (“Let me know if you need more details”) ### **Output Format:** Structure the output as a professional account report with the following sections: 1. Account Overview 2. Account Landscape & Opportunities 3. Stakeholder Mapping 4. Engagement Insights 5. Action Plan Use clear headings, tables, and bullet points where appropriate. Maintain a logical flow from account-level insights to stakeholder-level recommendations. ### **Context:** This prompt is designed to create a meeting preparation report for CSMs.\ The output provides a single, comprehensive snapshot of the account — to help CSMs prepare for meetings, identify growth and risk signals, and plan effective engagement strategies.\ It supports proactive customer management, consistency across teams, and data-driven decision-making. ### **Final Instructions:** - Only generate insights where data exists; if insufficient information is found, clearly state “Not enough data.” - Do not make assumptions or fabricate missing information. - Focus on actionable insights that inform CSM prioritization and engagement strategy. - Ensure the final output reads as a strategic, customer-facing summary, not an internal system log or data dump.

Step 4: Decide the actions

Next, pick where you want the output saved. You could **Create** or **Update** conversations, documents or even **send them to yourself on Slack**. For this use case, we have decided to **create** a Conversation of type "AI Summary and Recommendations" so you can have it available from the Company Profile ahead of the meeting.

Step 5: Test and iterate

So there you have your AI automation! Now it's time to put it to the test. We recommend testing the automation on a Test Company to ensure it behaves as you expect. Trigger the automation, check the output, tweak the prompt, run again - until you are satisfied with the output!

Here you can see the automation in action!

This automation transforms how teams prepare for customer meetings — replacing hours of manual research with instant, insight-rich analysis. With every report, CSMs gain clarity on who to engage, what to focus on, and how to drive impact. The result? **Smarter meetings, faster decisions, and more strategic partnerships** — powered by AI and fueled by customer understanding. Go try it yourself!