Write the cover letter and FAQs outside the proposal tool, from a fact sheet that holds only the figures and assumptions the design software printed, then paste them into the tool's text blocks. Tell the AI to describe the system, cite modelled figures with their assumptions, and never state savings, payback or environmental benefits as outcomes.
The tool's own pages print modelled numbers with assumptions. The risk sits in the cover letter and FAQs, where AI adds warm, unsupported lines: "you will save thousands", "a greener future", "pays for itself". If you have customers in the EU, rules on environmental claims that national laws must apply from 27 September 2026 can turn that wording into a legal risk as well as a trust problem.
What the proposal tool already says, and where AI text fits
Solar proposal software produces the numbers: layout, system size, modelled generation, a bill-savings table, a quotation. OpenSolar's help pages describe the inputs behind its savings figures as the customer's annual consumption, usage pattern and tariff before and after solar; the design (equipment, roof slope and direction, inverter sizing, battery and shading); and financial settings such as system price, incentives, payment method, energy price inflation and the years simulated. Most proposal tools have a version of that list, and it is the list your assumptions page should print.
The tool leaves gaps that people fill with words. In OpenSolar's Scroll template, according to its help article, the sections include a Cover Message, About Us, How Your System Works, Environmental benefits, Bill savings, Next Steps, Case Studies, Terms and Conditions and three Content Blocks. The cover page message accepts tokens that insert organisation or project details, the disclaimers for hardware, system performance, bill savings, quotations and payment types are editable, and the OpenSolar disclaimer is required and cannot be removed. The article does not list a dedicated FAQ section, so a Content Block is the natural home for FAQs there. Other tools differ, so find the equivalent free-text blocks in yours before you draft anything.
Those free-text blocks are where AI helps. It is quick at turning a design summary into a warm letter and a set of plain-English answers. It is poor at knowing which sentence will cause trouble in a year, which is why the rest of this tutorial is about fences.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowThe fact sheet: only what your own tool printed
Build one sheet per proposal and fill it from the design software's output, not from memory. The sheet is what you paste into the prompt, so it is also the boundary of what the letter may say. Leave out the customer's address, phone number and email; use a first name placeholder and a property description. Names and contact details go in through the tool's tokens afterwards, not through a general AI chat.
The worked proposal below is illustrative: a 12-panel system on a single main roof face, no battery.
| Field | Illustrative entry |
|---|---|
| Customer | [first name]; detached home, one main roof face used |
| System | 12 panels at 420 W, which is 5.04 kWp; 5 kW inverter; no battery |
| Roof as surveyed | 35 degree pitch; direction 15 degrees off the ideal; shading allowance 6% for a tree near the boundary |
| Modelled year-one generation | 4,350 kWh, printed on page 4 of the proposal |
| Consumption used in the model | 3,900 kWh a year, from the customer's last 12 months of bills |
| Tariff and price settings | Import and export rates as entered from the customer's bill; energy price inflation set at 3% a year; settings listed on page 8 |
| Page map of the proposal | Generation page 4; monthly pattern page 5; bill effects page 6; quotation and warranties page 7; assumptions page 8, headed "How these figures were calculated" |
| Warranties | Copied from the quotation page: [panel warranty], [inverter warranty], [workmanship warranty] |
| Timings | Median 24 working days from signed proposal to installation over our last 20 jobs; scaffold and approval timing vary |
| Allowed to mention | Modelled generation with its assumptions. Not allowed: bill savings, payback or return figures in our own words, environmental benefit, guarantees of output |
Notice the last row. It is the policy in one line, and it goes into every prompt. The proposal's own pages may print bill savings with the tool's assumptions and disclaimers, and how much you show is your decision with your adviser. The AI-written text points to those pages. It does not repeat the numbers in warm words, because a letter is read in a different mood from a table.
The EU rules on environmental claims, in plain words
If you market to consumers in the EU, Directive (EU) 2024/825 is the one to know. The directive asks member states to apply their national measures from 27 September 2026, and it changes consumer law on four points that matter to a solar cover letter. This is a caution, not legal advice, and national wording differs, so have a qualified adviser check your template once.
- Generic environmental claims. The directive's recitals list examples such as "environmentally friendly", "eco-friendly", "green", "climate friendly", "carbon friendly" and "energy efficient", and treat them as prohibited unless recognised excellent environmental performance can be shown. A claim is not generic when its specification is given in clear and prominent terms on the same medium. The recital's own contrast is "climate-friendly packaging" against "100% of energy used to produce this packaging comes from renewable sources".
- Offsetting. Claims that a product has a neutral, reduced or positive greenhouse gas impact because of offsetting are prohibited. The examples include "climate neutral", "carbon positive" and "climate net zero".
- Future promises. A claim about future environmental performance needs clear, objective, publicly available and verifiable commitments set out in a detailed and realistic implementation plan.
- Labels. Sustainability labels not based on a certification scheme or established by public authorities are prohibited.
The Commission's page for a separate Green Claims proposal listed it as pending when I read it in November 2026, so do not build a template around it either way. What helps in practice is a habit. Replace "green solar for a greener home" with something specific and checkable, such as "a 5.04 kWp system of 12 panels, modelled to generate 4,350 kWh in its first year under the assumptions on page 8". That sentence needs no environmental adjective, and it passes the same test in any market.
Prompt one: the cover letter that describes, not promises
Put the rules in the prompt, and put the fact sheet underneath. Name the pages the customer should read, because a cover letter that points to the assumptions page is doing a job the AI cannot do for you.
Write a cover letter of 180 to 220 words for a solar proposal,
using ONLY the fact sheet below.
Include: a greeting to [first name]; what we designed and why, in two
sentences, using the roof facts; the modelled generation figure
exactly as given, with the words "modelled" and "under the
assumptions on page 8"; a pointer to pages 4 and 8; what happens next;
one line inviting questions.
Never write: savings, save, payback, return, free, pays for itself,
guarantee, green, eco, clean, climate, carbon, planet, sustainable,
energy efficient, or any promise about bills or output.
Do not add numbers, dates or warranty periods that are not in the
fact sheet. If something is missing, write [MISSING].
British English. Plain, warm, no exclamation marks.
FACT SHEET:
[paste]A typical draft, illustrative, begins like this:
Dear [first name],
Thank you for having us survey your home. Your roof is a good fit for
a 5.04 kWp system, and you'll slash your bills while joining the green
energy revolution! Our design uses 12 panels on your main roof face,
with a 6% allowance for the tree near your boundary...The second sentence breaks three rules in one line, even though the prompt forbade all three. That is why the second pass exists.
| AI wording | Rule it breaks | Replacement |
|---|---|---|
| "you'll slash your bills" | Bill promise, not in the sheet | Delete. Point to page 8 for the assumptions behind any bill figures the tool prints |
| "joining the green energy revolution" | Generic environmental claim and hype | Delete. State the system and its modelled generation instead |
| "a good fit for a 5.04 kWp system" | "Good fit" is a judgement the survey supports, but the sheet does not say so | Keep only if the survey notes support it; otherwise "the survey found the main roof face suits 12 panels" |
| "a 6% allowance for the tree" | Supported by the sheet | Keep |
Before and after: the same paragraph, rewritten
This is the edit you will make most often. The AI's draft says what the customer would like to hear. The rewrite says what the design shows, and where to check it.
| First draft | Approved version |
|---|---|
| "This system will generate clean, green electricity for decades and cut your carbon footprint to near zero." | "The system uses 12 panels of 420 W (5.04 kWp). Our model puts its first-year generation at 4,350 kWh under the assumptions on page 8." |
| "Expect to see your bills fall from the first month." | "Page 8 lists the consumption, tariff and price settings behind the figures in this proposal. If your usage or tariff differs, the results will differ." |
| "Panels are guaranteed to perform for 25 years." | "The panel, inverter and workmanship warranties are set out on the quotation page, with their start dates and conditions." |
| "We're installing within two weeks of signing." | "Over our last 20 jobs, the median was 24 working days from signed proposal to installation. Scaffold and approval timing can move that." |
The third row deserves a note. AI loves a round-number warranty, and a 25-year figure in a letter can contradict the datasheet. Warranty lengths and conditions come from the paperwork and nowhere else.
Prompt two: FAQs the installer can stand behind
An FAQ is where a customer's real worries show up, and it is the part of a proposal most likely to be quoted back later. Give the AI the questions and let it answer only from the sheet. Where the sheet has no answer, the right output is "ASK INSTALLER", not a plausible guess.
Answer each question for a customer-facing solar proposal FAQ, in
two to four sentences, using ONLY the fact sheet. If the sheet does
not contain the answer, write ASK INSTALLER. No promises about bills,
output, payback or environmental benefit. Mention the page where
assumptions are printed when a question touches figures.
QUESTIONS: Will I save money? How much will it generate in winter?
What happens if electricity prices change? Does shade matter? What
do the warranties cover? Can I add a battery later? How long does
installation take? Who do I call if something stops working?
FACT SHEET: [paste]These are the answers that work, with the pattern each one follows. They are illustrative and carry the worked proposal's numbers.
| Question | Safe answer pattern | Illustrative answer |
|---|---|---|
| Will I save money? | Point to the modelled figures and the assumptions; name what changes the result | "The proposal models bill effects on page 6 using the consumption and tariff settings on page 8. Real results depend on how much electricity you use and when, and on your tariff, so we show the assumptions rather than a promise." |
| How much will it generate in winter? | Say it varies by season; give the annual modelled figure only | "Output is lower in winter and higher in summer. The model gives an annual figure of 4,350 kWh for the first year; the monthly pattern is on page 5." |
| What if electricity prices change? | State the assumption used and that it is an assumption | "We set energy price inflation at 3% a year in the model. That is an assumption, not a forecast, and you can see its effect by asking us to rerun the model with a different figure." |
| Does shade matter? | Cite the survey allowance | "Yes. We allowed 6% for the tree near your boundary, based on the survey. A taller tree or a new building would change it." |
| What do the warranties cover? | Refer to the paperwork; do not paraphrase periods | "The panel, inverter and workmanship warranties are on the quotation page, with their terms. We will go through them with you." |
| Can I add a battery later? | Say what depends on the inverter and wiring | "Often yes, but it depends on the inverter and the electrical set-up. We can check this at the survey and note it in your records." (ASK INSTALLER if the sheet is silent.) |
If your offer includes batteries, wording for backup and resilience needs even more care, and the tutorial on solar battery quote templates with options explained covers it. The same no-savings-promise discipline appears in insulation proposals that explain options without promising savings.
Prompt three: find promises, generic claims and invented numbers
A second chat, with no memory of the draft, is the best proofreader. Paste the finished letter and FAQs next to the fact sheet and ask for a list of sentences, not a rewrite.
Below are proposal text and the fact sheet it must match. List:
1) every sentence that mentions savings, bills, payback, returns,
"free", guarantees, output promises or the future;
2) every environmental statement (green, eco, clean, climate, carbon,
planet, sustainable, energy efficient, renewable, net zero);
3) every number, date, percentage or warranty period, with the
fact-sheet line that supports it, or NO SUPPORT.
Do not rewrite anything.
TEXT: [paste]
FACT SHEET: [paste]The output is a table you can work down, with a decision for each row.
| Flagged item | Support | Decision |
|---|---|---|
| "first-year generation of 4,350 kWh" | Fact sheet, modelled generation | Keep, check it says "modelled" |
| "the green choice for your roof" | NO SUPPORT; environmental statement | Delete |
| "3% a year price inflation" | Fact sheet, tariff settings | Keep, check it says "assumption" |
| "installation in 3 weeks" | NO SUPPORT (sheet says median 24 working days) | Replace with the sheet's sentence |
| "25-year warranty" | NO SUPPORT | Replace with a pointer to the quotation page |
Do not skip the manual search. Open the final text and use find for the risky words: save, free, guarantee, will, always, green, eco, clean, climate, carbon. Each hit either has a line in the sheet or it goes. A tool does the same job faster, but a person who knows the design should still read the result once. The tutorial on catching made-up figures in AI-drafted proposals goes deeper on the numbers side.
Paste it into the tool: where each block goes
Placing the text matters, because a good paragraph in the wrong block can sit next to a number it contradicts. In OpenSolar's Scroll template, the pattern is straightforward.
- Cover Page Message: the cover letter, kept short because customers may see it before they select Continue to Proposal. Use the tokens for organisation and project details instead of typing names and addresses into the AI.
- About Us and How Your System Works: the explanation text, drafted from the equipment list and checked for technical accuracy.
- A Content Block: the FAQs, in the order customers ask them.
- Disclaimers for system performance and bill savings: leave the tool's wording in place and add your own assumptions note under it. Never write a letter that sounds more certain than the disclaimer printed two pages later. The article says some sections exist only in the online proposal or only in the PDF, so open both versions and read the letter in each.
- Environmental benefits section: if your template shows it, read its wording against the rules above before sending, and hide the section if you cannot stand behind every line.
The tool's savings-calculation article is useful reading when you set the assumptions note, because it shows which inputs the tool treats as the customer's, the design's and the financial settings'. The tutorial comparing AI solar design software by cost per proposal shows how tools differ, and what solar design software does and does not do explains why modelled output still needs a surveyor's eye.
A five-minute read-aloud before anything is sent
Run this check on every proposal, however short the letter. It takes about five minutes, and it catches what the prompts miss.
- Read the letter aloud next to the PDF. Any number you say must be on a page you can point to.
- Search the text for the risky words listed above and clear each hit against the fact sheet.
- Compare every figure in the letter with the tool's figure, character for character. "About 4,500" is a different number from 4,350.
- Confirm the assumptions page matches the settings in the design software today, not the settings from last quarter.
- Check the warranty sentence points to the paperwork, and that the paperwork is attached.
- Have a second person who did not draft it read for tone and for anything that sounds like a promise.
- Save the approved text with the date and the name of the approver, so you can show what was sent.
When a customer asks outright, "How much will I save?", have the answer ready: "The model's bill effects are on page 6, with the assumptions on page 8. They change with your usage and your tariff, so we treat them as an estimate and not a promise." Practise it, because a verbal reassurance can undo careful writing.
Three slips that reach customers when nobody checks
- The stale setting. The tool's tariff or inflation setting changed after the letter was drafted, so the letter quotes 4,350 kWh while page 4 now says 4,290. A customer notices two different numbers. Draft the letter last, after the model is final, and compare character for character.
- The invented date. The AI writes "installation will be completed within two weeks" because proposals usually promise something. Timings come from the diary's median, with the factors that move it, or they stay out.
- The carbon line in an FAQ. Asked "is solar good for the environment?", an AI answers with confident statements about carbon payback and lifetime emissions. None come from your tool, and some would fall inside the offsetting and generic-claim rules above. Either answer with a specific statement from the sheet or write "ASK INSTALLER" and leave the question out until someone with sources has written it.
Keep an approved FAQ library and review it each quarter
After a few proposals, the answers repeat. Move the approved ones into a shared library with a review date beside each, and let the AI do the part it is good at: reordering, shortening, and swapping in the figures from a new sheet. Updating the library each quarter, or whenever tariffs, incentives or the design software's defaults change, keeps old assumptions from drifting into new proposals.
The library is also where a lesson from a complaint lands. If a customer misread a sentence, rewrite it once, in the library, and the correction reaches every future proposal. Compare that with fixing each letter by hand, and the case for the fact sheet, the prompts and the second-chat check makes itself. For how these pieces sit alongside a general-purpose assistant and proposal software, see AI proposal software against a general AI assistant.
Questions about AI text in solar proposals
Can a solar proposal show bill savings at all?
Most proposal tools print modelled bill savings with their assumptions and disclaimers, and that is a decision for you and your adviser. The AI-written cover letter and FAQs should not restate those numbers as outcomes. They should point to the proposal pages where the tool prints them, with the assumptions beside them.
Which words should I ban in AI-written solar copy?
Search every draft for: save, savings, free, pays for itself, payback, return, guarantee, always, will, green, eco, clean, climate, carbon, planet, sustainable and energy efficient. Not every hit is a problem, but each one needs a source in your fact sheet or a rewrite into a specific, modelled statement.
Does the EU environmental-claims rule apply to a small installer?
It applies to traders marketing to consumers in EU markets, through national laws that implement Directive 2024/825 from 27 September 2026. Whether it covers you depends on where your customers are, so ask a qualified adviser. Writing specific, checkable statements instead of generic green wording is sensible everywhere.
Should the cover letter repeat the generation figure?
It can, if it says modelled, gives the figure the tool printed, and names where the assumptions sit. Keep the figure identical to the proposal. A letter that rounds 4,350 up to 'about 4,500' creates a second number that the customer can hold you to.
Can AI write the 'how your system works' section?
Yes, as an explanation of the parts and what each does, built from your equipment list. Keep numbers out unless they are in the fact sheet, avoid the words green and clean, and have someone who knows the design read it for technical errors before it is sent.
Further reads
- Solar Installer Software: CRM, Design and Tracking, and Where AI Helps — Compare the CRM, design and tracking tools around proposals.
- Roofing Proposal Software: Adding AI-Written Scopes and Warranty Wording — Another trade's proposal text, with warranty wording to compare.
- AV Proposal Software: Adding AI-Written Scope and Exclusions — How scope and exclusions get added to proposal software.
- Can AI Write a Solar Sales Script? Discovery Questions and Objections — Prepare spoken answers that match the written proposal.
- Solar Permit Application: Can AI Draft It? What Installers Must Verify — The permit paperwork that follows an accepted proposal.
- Solar Installer Marketing: Local Pages and Case Studies Written With AI — Apply the same fact-sheet rule to town pages and case studies.
- AI for Solar Installation: Five Jobs a Small Installer Can Automate — Five installer jobs with illustrative hours saved each: lead qualification, survey notes, proposal wording, permit paperwork drafts and customer updates.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Directive (EU) 2024/825 on consumer protection and the green transition (recitals 4, 7, 9 and 12; Article 4 on application dates); OpenSolar help articles on Scroll proposal templates and on how financial savings are calculated; read November 2026.