--- title: "Prompting Slate AI" slug: "prompting-slate-ai" updated: 2026-06-17T17:40:35Z published: 2026-06-17T17:40:35Z canonical: "knowledge.technolutions.net/prompting-slate-ai" --- > ## Documentation Index > Fetch the complete documentation index at: https://knowledge.technolutions.net/llms.txt > Use this file to discover all available pages before exploring further. # Prompting Slate AI [Slate AI](/v1/docs/slate-ai) is a Slate-savvy virtual assistant powered by a **large language model** (LLM). You can make requests of it on any page from the user menu sidebar. If you’re staring at the blinking cursor and wondering what to ask it, use this article as a starting point. **What’s an LLM? 💭** Large language models are generative AI constructs. They are trained on huge datasets that include the written word, images, code, and more. The *generative*part of “generative AI” refers to these models’ ability to produce a response to your natural-language input. Your input is called a **prompt**. An LLM’s response to a prompt is built out of **tokens***,*a unit of text around 4 characters long. Your input is similarly “tokenized” so the LLM can process it. LLMs pick the next token of their response based on the statistical likelihood that it follows the one before it. (To say this explanation is an over-simplification is an [understatement](https://arxiv.org/pdf/1706.03762).) Importantly, this output is not deterministic: you aren’t guaranteed the same output from identical prompts. Great results start with great prompts. Whether you're building a Form, reviewing data, or drafting a Message, Slate AI (and generally speaking, all AI) can help in meaningful ways, if you know how and what to ask. Think of prompting as a conversation with a capable colleague. Slate AI performs best when you provide clear context and a clear goal: - **Good prompts clarify:**They explain what you're doing, why you're doing it, and what you want to happen. - **Great prompts expand:** They invite Slate AI to identify blind spots, suggest improvements, and offer new directions. ## Example situations and prompting strategies Each example follows this structure: - **🧩 Situation** - **🎯 Goal** - **💬 Prompt** - **🤖 How AI can help** - **🔁 Follow-up prompt** ## Prompting for form building **🧩 Situation:** Request for Information form **🎯 Goal:** Improve completion rates **💬 Prompt:** > [!NOTE] > "I'm building a Request for Information form in Slate. What questions should I include to ensure high completion, without overwhelming users? Include Field suggestions and optional vs. required guidance." **🤖 How AI Can Help:** - Recommend ideal fields and logical order - Flag friction points - Suggest Conditional Logic or personalization ideas **🔁 Follow-Up Prompt:** > [!NOTE] > "Now suggest conditional logic based on the answer to 'Intended Start Term.'" ## Prompting for query analysis **🧩 Situation:** Campaign engagement data **🎯 Goal:** Spot trends or outliers **💬 Prompt:** > [!NOTE] > "Here are the results of a query showing engagement by campaign type. Help me identify any trends or outliers, and suggest 2–3 follow-up areas to explore further." **🤖 How AI Can Help:** - Highlight unusual patterns - Recommend filters or grouping strategies - Suggest action items based on behavior **🔁 Follow-Up Prompt:** > [!NOTE] > "What might explain the lower engagement in Segment C? How could I increase engagement?" 📖 **More ways to prompt:** - [Analyzing Query Results for Actionable Insights with Slate AI](/v1/docs/slate-ai-analyzing-query-results-for-actionable-insights) - [Predicting Melt Before It Happens with Slate AI](/v1/docs/slate-ai-predicing-summer-melt-before-it-happens) - [Reviewing Counselor Travel Outcomes with Slate AI](/v1/docs/slate-ai-reviewing-counselor-travel-outcomes) - [Understanding Admit-to-Deposit Conversion with Slate AI](/v1/docs/slate-ai-understanding-admit-to-deposit-conversion) ## Prompting for query-building help **🧩 Situation:** Building or troubleshooting subquery exports **🎯 Goal:** Get help choosing the right output type and configuring exports **💬 Prompts:** > [!NOTE] > - "How do I use the Rank output to always get the primary phone number when someone has multiple numbers listed?" > - "How do I configure a subquery export to display a student's TOEFL score if available, but their IELTS score if not, using Coalesce?" > - "How do I show the total number of forms submitted by each student using Aggregate in an independent subquery?" **🤖 How AI Can Help:** - Recommend the right subquery export output type for your use case - Walk through join, export, and sort configuration step by step - Troubleshoot unexpected results from subquery exports ## Prompting for email campaign strategy **🧩 Situation:** Nurture campaign for non-applicants **🎯 Goal:** Increase open and click rates **💬 Prompt:** > [!NOTE] > "I’m drafting a 3-email nurture campaign for prospects who haven’t started their Application. Suggest a subject line, CTA, and tone for each message, and how they should evolve over time." **🤖 How AI Can Help:** - Generate subject lines and tone variations - Recommend CTA phrasing - Suggest merge fields or personalization logic **🔁 Follow-Up Prompt:** > [!NOTE] > "Now revise that for first-gen students interested in Nursing programs." 📖 **More ways to prompt:** - [Drafting High-Engagement Emails with Slate AI](/v1/docs/slate-ai-drafting-high-engagement-emails) - [Re-Engaging Incomplete Applicants with Slate AI](/v1/docs/slate-ai-re-engaging-incomplete-applicants) ## Prompting for data visualization **🧩 Situation:** Yield dashboard **🎯 Goal:** Highlight key funnel metrics **💬 Prompt:** > [!NOTE] > "I'm creating a dashboard to track yield. What are the most impactful metrics to include, and how should I group them for storytelling?" **🤖 How AI Can Help:** - Suggest high-impact KPIs - Recommend groupings by audience or behavior - Offer layout ideas (e.g., side-by-side comparisons) **🔁 Follow-Up Prompt:** > [!NOTE] > "Now adapt this for our transfer Population. What changes?" **📖 More ways to prompt:** - [Preparing a Board-Ready Enrollment Report with Slate AI](/v1/docs/slate-ai-preparing-a-board-ready-enrollment-report) - [Designing a Smarter Counselor Dashboard with Slate AI](/v1/docs/slate-ai-designing-a-smarter-counselor-dashboard) ## Power prompts for strategic leverage Not all prompts are about getting tasks done, some are about unlocking insight. > [!NOTE] > - “We’re overloaded during yield season. Based on what Slate can automate, where should we begin offloading work?” > - “We want to improve student retention post-admit. What data should we start tracking now to help?” > - “Here’s our current communication flow. Where might we be over- or under-engaging students?” **📖 More ways to prompt:** - [Forecasting Enrollment Health for Budget Planning with Slate AI](/v1/docs/slate-ai-forecasting-enrollment-health-for-budget-planning) - [Prompts for Institutional Leadership](/v1/docs/slate-ai-prompts-for-institutional-leadership) ## Prompting mental models Use these models to sharpen how you think about prompting: ### The builder model 1. **Structure it**. Draft your first version. 2. **Improve it**. Ask AI to critique or refine the draft. 3. **Stress-test it**. Ask where it could fail. 4. **Scale it**. Adapt for new audiences or needs. ### The lens model Use AI to review your work through multiple lenses: - **Technical**: “Is this query efficient?” - **User Experience**: “Is this form too long or confusing?” - **Strategic**: “Does this dashboard inform action or just show data?” You may need to provide an external link to the item you’re reviewing. ## Prompting to stress-test your work Slate AI can not only build, it can also uncover potential pitfalls with properly-formed prompts. > [!NOTE] > - “Where might this form design confuse applicants?” > - “How could this drip campaign go wrong?” > - “What assumptions am I making in this segmentation logic?” > - “What blind spots does this rule setup create?” Prompting Slate AI to challenge your work can catch potential issues avoid problems for your users. While the specifics of your conversation with Slate AI depend on what you’re working on, there are some general guidelines that apply in all circumstances: ## More prompting tips #### Give Slate AI structure While you might be tempted to ask Slate AI open-ended requests, like “Build me an admitted students portal” or “Write a donor thank-you email,” and while these requests will almost always create something to work from, you’ll find the AI is even better at following instruction than it is at coming up with stuff on its own. Providing a framework, like: ```plaintext Write a welcome email to a prospective student. The email should include:    - A personalized introduction based on the prospect's interests.    - A schedule of upcoming campus events with links to register    - Links to campus resources The tone of the email should be professional, in keeping with our style guide. ``` The more structure you give Slate AI, the easier it’ll be to iterate through drafts. #### Be aware of system instructions Your Slate Administrator may have imposed certain constraints on Slate AI [at a system level](/v1/docs/slate-ai-snippets). As with most AI models, when you ask Slate AI about its system prompt, it generally won’t tell you, but the nature of its responses may offer a hint: ![](https://cdn.us.document360.io/cd8ea7a6-07f3-4846-a554-627ac016d3e3/Images/Documentation/Asking about system instructions.png) Forms in Slate are used for a number of tasks, from collecting data from prospective students to administratively updating specific data points for student records. A form is quite often an integral piece for many of the other Slate modules. A well-developed form can play a crucial role in marketing efforts, record management, reader review, events, and interviews. A single communication in Deliver, including emails, texts, postcards, etc. Can be used interchangeably with "Mailing." Alternately, a "Campaign" is a series of messages. A comprehensive constituent relationship management system created specifically to meet the needs of higher education institutions, such as admissions, student success, and advancement offices, all from within a single unified interface. A field is part of a record and contains a single piece of data (a value). Similar to Liquid Markup, Conditional Logic allows text or images to be displayed or hidden based on defined criteria. **Deliver:**A series of messages in Slate most often used in reference to Drip Marketing. A campaign does not refer to a single message. **Slate for Advancement:**A broad fundraising initiative or goal represented in the *Campaigns* dataset. A set of data which pertains to an individual applying to your institution. A Population is a group of records with like attributes that are assigned a common label in Slate via population rules in the Rules Editor. Populations can include both prospect records and application records, and records can be enrolled in multiple populations at the same time.