AI Prompt Library
for Industrial Efficiency

Role: I am a plant manager considering replacing a compressor and you are the engineer tasked with doing the market research. Instruction: Provide a market research report of the 5 best replacements on the market with links and data for our situation. Context: We have a [compressor type] compressor system running at [system specs, pressure, operation ours, duty factor, etc.] Examples: Find market research reports online to use as examples for the report you will give me. Constraint: Make sure all considered compressors can meet the current specifications. And keep the report to under 5 pages. Output: Give me a report with the 5 best replacements with links and data for each. Also provide your recommendation. List them in order of the best to the worst. Provide insights on the expected energy efficiency of each.

Role: Address me from the point of view of an energy consultant providing recommendations on energy efficiency in my facility. Instruction: Provide a report of prioritized energy efficiency recommendations for [system or area of interest]. Assume reasonable values for missing information and state your assumptions in the report. Context: I have [equipment specs, operation hours, and other relevant system information]. Examples: Find and use energy efficiency reports online as an example. Constraint: Limit the projects to those that have average payback of less than 3 years and limit the report length to one page. Ask me questions to fill in any missing information you need to perform the calculations and improve the report. Output: Generate the results in a report format with bullet points for each recommendation and prioritize them by payback period with expected annual savings when calculations are possible with the provided data.

Role: Act as an engineer at a company assigned to brainstorm possible solutions to the problem below. You are assigned to be creative and think outside the box. Instruction: Provide a report of 5 completely different ideas that could be used to approach the problem. State any caveats and list pros and cons of each. List all the specific components needed as well as a price estimate. Also briefly address installation and maintenance. Context:You are designing [description of your project]. It must be able to [constraints of your project] The data you need [the output of your project] Examples: Find available reports online that fill a similar function of reporting various solution ideas to a boss. Constraint:Limit the report to 5 pages. Be concise in your tone. If a solution doesn’t meet all the requirements address it. Output: Give me a report style output addressing clearly and concisely each of the different options with pros and cons.

Role: Act as an industrial energy engineer with 15+ years of experience conducting facility motor audits and reading equipment nameplates from field photos. Instruction: Process every uploaded photo individually without skipping, merging, or stopping early, and produce exactly one table row per photo. Flag any equipment lacking both HP and FLA as “Excluded — not a motor” with a one-line reason. Then identify duplicates sharing manufacturer, HP, Voltage, RPM, and Enclosure: keep one as “Qualifying motor (representative)” and mark the others as “Duplicate; combined with [Motor ID].” Context: I am uploading nameplate photos taken during a plant walkthrough. Each photo shows one equipment nameplate. [Add plant name, area surveyed, and date.] Examples: M-03 | Photo 3 · SN:8842-A1 | Baldor | 25 | 460 | 30.1 | 93.6 | 1,780 | TEFC | Qualifying motor (representative) Constraint: State all assumptions. Do not invent data — mark unreadable fields as [ASSUMED] or [ESTIMATED; reason], and if a serial number is not visible, write [ESTIMATED; not found] in the Source column. Before finalizing, verify the row count equals the photo count. Output:A table with columns: Motor ID, Source (Photo [#] · SN:[serial]), Manufacturer, HP, Voltage (V), FLA (A), Efficiency (%), RPM, Enclosure, Status – followed by an inline horizontal bar chart of HP by Motor ID for unique qualifying motors only, sorted largest to smallest.

Role: You are an AI prompt engineer with specific expertise building apps using [your AI app tool]. Instruction: I would like to create an app that [the purpose of your app]. Include the following features: [list desired features]. Context: This app will be used by [people who will be using your app]. It should be mobile friendly and easy to use in [your circumstances]. Examples: For any calculations you preform, use trusted references like [specific references]. Format the app layout and theme similarly to [your reference]. Constraint: Specify that the app should show all equations that are used in any background functions. Include industry best practices where relevant. Output: Output a prompt that can be directly copy and pasted into [your AI app tool] and used to build the first iteration of an app. Format the prompt in the most clear and efficient way possible for this tool.

Role: You are an industrial energy analyst specializing in electric utility bill analysis. Instruction: Review all uploaded utility bills and populate the attached Excel template with the required data. Read every page of each bill before extracting information. Extract values exactly as shown, preserving units, decimal precision, and signs. For every cost item, provide a brief explanation and classify it as demand-dependent (kW or kVA), consumption-dependent (kWh), or independent of both. Context: The uploaded files may contain multiple utility bills, and individual bills may span multiple pages. Utility tariffs vary by provider, so use the terminology and structure presented in each bill. Constraint: Review every page of every bill; do not skip any content. Do not infer or estimate missing information. If a value is unavailable, leave the corresponding field blank or mark it as “N/A”. Preserve the utility’s original terminology for all cost items. Verify that extracted totals match the source bill and report any discrepancies or assumptions. Output: Complete the attached Excel template for each bill. For every cost item, include its name, amount, unit (if applicable), dependency classification (Demand/Consumption/Independent), and a brief explanation. Ensure all extracted totals reconcile with the original utility bill. [describe any other specific fields associated with the excel template you are using]

Role: Act as an industrial energy analyst experienced in utility bill analysis and in preparing data imports for ORNL’s VERIFI tool. Instruction: Step 0 — Verify the template is current. Before extracting anything, check the latest VERIFI import template format: download or fetch the current template from https://verifi.ornl.gov/assets/csv_templates/VERIFI-Import-Data.xlsx (or check the ORNL-AMO/VERIFI GitHub repository for recent template changes). Compare its Electricity Meter Setup and Electricity Meter eadings column headers against the reference headers listed below. If they differ, use the newer template’s headers and explicitly report every difference found. If you cannot access the web, state this clearly and proceed with the reference headers below, flagging the output as “based on template version as of [[date you saved these headers]] — verify before import.” Step 1 – Process every page of the uploaded utility bill PDF for the primary electric account only — the largest-dollar account identified by the “EG ELECTRIC” meter line (or equivalent primary electric meter line if labeled differently). Ignore all water, drainage, and secondary accounts. Step 2 – Build the import files using the verified headers: (a) Electricity Meter Setup — one row for the meter. Reference Headers: Facility Name | Meter Number (UNIQUE) | Meter Name (DISPLAY) | Meter Group (ANALYSIS) | Calendarize Readings? | Unit (USAGE) | Unit (DEMAND) | S2S Factor | Agreement Type | Retain RECs? | Account # | Supplier | Location | Notes Fill Meter Number, Account #, Supplier, and Location from the bill header if present on the bill. Set Unit (USAGE) = kWh and Unit (DEMAND) = kW. Leave analysis- preference columns (Meter Group, Calendarize Readings, S2S Factor, Agreement Type, Retain RECs?) blank for manual entry. (b) Electricity Meter Readings — one row per billing period, with these exact columns: Meter Number | Read Date | Total Usage | Actual Demand | Power Factor | Total Cost ($) Map: Read Date = service period end date (current read date); Total Usage = kWh from the base usage / EG ELEC ENERGY line; Actual Demand billed kW/kVA demand if shown on the bill (use the billed column, not the actual/coincident column, when both appear); Power Factor as shown, converted from percent to decimal (e.g. 76.70% → 0.767); Total Cost = total amount due for that period (sum of all charge sections). Meter Number must match the Setup sheet exactly. (3) Supplementary extraction table (not for import) for audit trail purposes, list per billing period: service start date, service end date, previous/current meter readings, units consumed, and fuel adjustment value. Context: The PDF contains monthly electric bills for one facility. [Add facility name — this fills the Facility Name column and must match the facility name already created in VERIFI. If not provided, leave blank and flag it in the completion report.] Examples: Readings row: MTR-4471 | 02/02/2025 | 400080 | 1120 | 0.94 | 38412.55 Constraint: do not invent data, but do not block on missing data either: – Extract every field the bill contains. If a field the template calls for (Meter Number, Account #, Location, Facility Name, Actual Demand, Power Factor, fuel adjustment, etc.) does not appear anywhere on the bill or in the provided context, do not guess, estimate, or infer it leave that cell blank, note in-line in the completion report that it was “not found on bill,” and proceed to generate all three outputs anyway. Do not pause the task or ask the user to supply missing bill data before delivering the files to deliver the best-supported output first, then list exactly what’s missing and where it should come from. – Only ask a clarifying question before generating output if the request itself is ambiguous (e.g., which of two accounts is primary, or which column represents billed vs. actual demand when the bill’s layout is genuinely unclear) not when the bill simply lacks a field the template wants. – Dates in MM/DD/YYYY, numbers without $ signs or thousands of separators. Column headers must match the template character-for-character. Verify the Readings row count equals the number of billing periods in the PDF. – If Meter Number is not found and is therefore blank on both sheets, say explicitly in the report that this blocks the Setup Readings join in VERIFI until the user fills it in on both files. Output: Two downloadable CSV files (VERIFI-Setup.csv, VERIFI-Readings.csv) plus the supplementary table (as a third CSV), followed by a completion report with two lists: – Filled from bills — each column and its source line/label on the bill – Needs manual input — each blank column, why it’s blank (not on the bill vs. a user preference/analysis setting), and where to find or decide it