Yes, you can write a grant application with a free AI plan, and many small businesses do. The limits bite on long documents: usage caps that reset every few hours, file-size limits, weaker memory across sessions and no guarantee of private handling, so plan to work section by section and check every fact.
A business bid is a different job from a charity application. An equipment grant, an innovation or growth fund, a local business support scheme or a skills grant for an owner-run company comes with a scoring grid, a budget with match funding and an eligibility test that can sink the application in the first paragraph. Free AI helps with the writing and the structure. It does not help with the decision of whether to apply, or with the facts.
What a business bid needs that AI can help with
Break the application into five jobs, because free plans are strong at some and weak at others.
- Reading the funder's guidance. Eligibility, scoring criteria, eligible costs, deadlines, word limits. AI summarises well if you give it the text.
- Checking fit. A table of "criterion, evidence we have, evidence we lack". AI is good at the shape and bad at knowing your business, so you fill the evidence.
- Drafting each answer. Section by section, to the word limit, in the funder's own vocabulary.
- The budget narrative. Words that explain each cost line. AI writes the prose. The numbers are yours.
- Final checks. Cross-checking claims, figures and word counts against the form.
Free plans cope with all five if you work in small pieces. The sections below cover what each plan does, where the caps interfere and how to stay inside them.
What the three free plans give you, as of November 2026
Limits change often and vendors describe them in different ways. This is what their own pages said when I checked.
| Plan | What the vendor says | What that means for a bid |
|---|---|---|
| Claude Free | Usage is limited every 5 hours and every week; includes file creation, web search, memory and up to five projects; Opus is not included | Good for long reading and careful drafting; a heavy session can stop mid-section; five projects is enough for several bids |
| Gemini Free | Deep Research and Gemini Notebook are listed on the free plan; limits refresh every 5 hours up to a weekly cap, and paid tiers raise them (AI Plus 2x, AI Pro 4x); the context window without a plan is 32K tokens | Useful for background research on a sector or a funder, though Google says Deep Research may be unavailable at busy times without a plan; check the usage-limits setting in the app to see where you stand |
| ChatGPT Free | Unlimited text chats (subject to abuse guardrails); messages with uploads, deep research, memory and image creation are "limited"; 3 file uploads a day; projects with up to 5 files each; a 27K-token context window on the Instant model, about 12 pages of input | Fine for drafting short answers one at a time; the daily upload cap and the short context mean you paste the relevant section of the guidance, not the whole pack |
For reference, the cheapest paid steps on the vendors' own pages are ChatGPT Go at $8 a month (Plus is $20), Google AI Plus at $4.99 a month (AI Pro $19.99) and Claude Pro at $17 a month on an annual subscription (billed $200 up front) or $20 monthly. If the free allowance cuts you off twice in a bid week, one month of one paid plan is probably the cheapest fix; a bid that wins is worth far more than the subscription. The cost side for professional grant writers is in AI grant writing software: what each costs a freelance grant writer.

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 nowStep 1: turn the funder's guidance into a scoring sheet
Before drafting, paste the eligibility section, the scoring criteria and the list of eligible costs into the chat as text. Do not rely on a PDF upload alone. ChatGPT Free allows three file uploads a day, Claude analyses charts and layout only in PDFs of 100 pages or fewer (longer ones are read as text), and an upload that fails halfway will waste an hour. A prompt:
Below is the guidance for a small-business equipment grant. I run a craft brewery supplier with 9 staff.
Task: produce a table with columns: criterion, maximum score or weight, what evidence the funder asks for, word limit.
Then list: eligibility rules that could rule us out, eligible and ineligible costs, deadlines, and anything that must be attached.
Quote the funder's own words for each rule. Do not add rules that are not in the text.
An illustrative output includes a row such as "Value for money, 20 points, quotes from at least two suppliers, 250 words" and a flag: "Costs incurred before the award date are ineligible". That last line saves a business from buying the equipment too early. Check every quoted rule against the original and note the page. If the tool says something the guidance does not, delete it.
Step 2: a fit table before any writing
Make a second table: each criterion against the evidence you actually have. Fill it yourself. A sample, for a supplier applying for a cold-store grant:
| Criterion | Evidence we have | Evidence we lack |
|---|---|---|
| Need | Spoilage log for the last 12 months; two customer complaints | Current cold-store capacity in cubic metres |
| Jobs | Two planned hires, with job descriptions | Salary budget and start dates |
| Match funding | Bank facility agreed in principle | Written confirmation from the bank |
| Value for money | One supplier quote | A second quote |
The "evidence we lack" column is the work plan for the week. AI cannot find a second quote, and a bid that claims jobs without a plan scores poorly. If most of the right-hand column is full, either wait for the next round or spend time on evidence, not prose.
Step 3: draft one answer at a time
Free plans do best with a single question, the scoring criterion for that question, your evidence and the word limit. Do not paste the whole form and ask for everything. Three reasons: the output wanders; a long chat can use up the cap; and long conversations can lose track of earlier instructions.
Question: "Explain how the project will improve the productivity of the business" (250 words max, scored 20 points on measurable improvement).
Our evidence: spoilage was about 6% of chilled stock last year by value; a second cold store would let us buy in larger lots; we plan to cut weekly delivery runs from 5 to 3.
Write a draft using only these facts. Use the funder's own terms: "productivity", "measurable". Put the measurable outcomes in the first two sentences.
Mark with [EVIDENCE NEEDED] any claim that needs a number I have not given you.
An illustrative response gives a tight 230-word draft, with "[EVIDENCE NEEDED: how much of the spoilage is avoidable]". You fix three things: the draft calls the business "innovative" without proof; it adds a sentence about "sustainable growth" that is not in your evidence; and its tone is more confident than your numbers justify. Cut the sentence, supply the missing number, and keep the funder's vocabulary. The step-by-step version of this for a general reader is in how to use AI to write grant applications.
Step 4: the budget narrative, with numbers entered by you
The budget table is arithmetic and the narrative is prose. Keep them apart. Type the table into a spreadsheet, total it with a formula and read every line yourself. Then give the tool the final table and ask it to write one sentence per line explaining why the cost is needed, what it buys and where the figure comes from.
An illustrative table: refrigerated unit installation, $18,400 (quote one); electrical upgrade, $3,100 (quote one); two part-time hires for six months, $14,000; training, $1,200. Total: $36,700. Match funding required: 40 per cent, so $14,680 from the business. The tool writes the narrative: "The unit installation is the largest cost and is supported by two quotes dated October. The electrical upgrade is required for the unit and is quoted separately by the installer." You check three things: the quotes exist, the dates are right, and the percentage matches the funder's rule. AI is known to make plausible slips on arithmetic, such as a percentage of the wrong total. Do the sum in the spreadsheet, not in the chat.
Step 5: citation and claim checks
Funders do read sources. If you quote a market statistic, an industry report or a regulation, you need to have opened it. Free plans can invent references, with a real-looking author, a plausible title and no existing page. Handle it like this:
- Ask the tool for the claim and where it would look, not for the source itself.
- Find the source yourself in a search, and open it.
- Copy the figure and the exact wording from the source, with its date.
- Delete every statistic you cannot trace to a page you have read.
A realistic mistake: a draft said "an estimated 30 per cent of chilled stock in the sector is wasted". No such figure could be found, and the reviewer would have asked for the source. The fix was to use the business's own spoilage log, which was stronger evidence anyway. The same principle applies to any claim about competitors, markets or regulation. For more on how funders react to AI in bids, see whether you can use AI for grant writing without funders rejecting it.
Which free plan for which job
After a few bids you will find that the plans differ in what they are good at for you. A sensible split to test, not a rule:
| Job in the bid | Try first | Why, and what to check |
|---|---|---|
| Summarising long guidance into a scoring sheet | Claude Free | Handles long text carefully; confirm quoted rules against the PDF |
| Background on the sector or the funder's past awards | Gemini Free (Deep Research) | Gives a sourced overview; open the sources, as some will be thin |
| Tightening a draft to the word limit | Any | Count the words yourself afterwards; word counts from chat tools are often wrong |
| Practice questions from a panel | ChatGPT Free | Quick back-and-forth; useful for interview-style stages |
| Final read for contradictions | A different tool from the drafter | A second tool will not share the first one's blind spots |
Before and after: an answer that would lose points
This is an illustration of what free tools produce when the prompt is thin, and what changes with specifics. A growth-fund question asks for the expected benefit to local jobs, 150 words, scored on "realistic, measurable job creation".
Thin prompt output: "Our innovative project will create exciting new employment opportunities and drive sustainable growth across the community, positioning the business as a leader in its field." It sounds fine, and a scorer will give it almost nothing: no number, no role, no date.
After you supply the evidence: "We will recruit two warehouse operatives (full time, 37.5 hours) by month 3 of the project, and one part-time customer service coordinator (20 hours) by month 6. Salaries are in the budget at lines 6 and 7. The roles are needed because the new cold store lets us handle 40 per cent more stock without extra delivery runs. We will report headcount to the funder quarterly." Every sentence is a claim you can prove, and the match with the budget lines is the thing scorers reward. The fix was not better prose; it was better input. If a draft reads as polished but empty, the cause is nearly always missing evidence, not a weak tool.
When a free cap cuts in mid-draft
A common failure: you are on the fourth long answer, the tool stops with a message that you have reached your limit, and it resets in a few hours. If the deadline is tomorrow, that is a problem. Prevent it with three habits.
- Save as you go. After each accepted draft, paste it into your own document straight away, not at the end of the session.
- Keep a "continuation note". One short block listing which questions are done, which tool and chat you used, and the facts note. If you have to switch to a second tool, paste the block in and continue.
- Do the heaviest request first. A long document read uses far more allowance than a paragraph edit. Spend allowance on the hard tasks when it is full.
An illustrative case: a four-person firm hit the weekly cap on Wednesday evening. Because each answer was already saved and the facts note was ready, they pasted the remaining two questions into a second tool and finished on schedule. The bid read slightly differently in those two answers, so they spent twenty minutes smoothing the voice in the last read-through. That cost is the price of switching mid-bid, and it is a good argument for starting a bid a week before the deadline, not two days.
A reviewer prompt to run before you submit
When the draft is complete, ask a tool, ideally one that did not write it, to mark the bid as a scorer would. Paste the criteria and the full text.
Act as a grant assessor. Here are the funder's scoring criteria and my complete application.
For each criterion: give a score out of the maximum, quote the sentence that earns the most points, and quote the weakest sentence.
List every claim that has no number or evidence. List any answer that does not use the funder's terms.
Do not rewrite anything. Do not invent facts.
Treat the scores as a rough guide, not a prediction. The useful output is the list of weak sentences and unsupported claims. Fix those by adding evidence, not adjectives. Then run a last human check against the form: right questions answered in the right boxes, attachments labelled as the funder asks, names and dates consistent across the document.
Red flags that mean "do not apply"
AI will happily write a bid for a grant you are not eligible for. Before you start, test these, and treat any "no" as a stop sign unless the funder says otherwise in writing.
- Is your business type, size and sector on the eligible list?
- Does the project start after the award date, or does the funder allow earlier spending?
- Can you fund the match, in cash, on the funder's timetable?
- Do you have every required attachment, such as accounts, quotes, insurance or a declaration?
- Does the grant interact with other support you receive, in ways the rules limit?
If the answers are unclear, email the programme contact with specific questions, and keep the reply with the bid papers.
Working around the free-plan limits
- Caps that reset every few hours. Plan the day in blocks. Do the heaviest tasks (reading guidance, drafting long answers) first, and the light tasks (word counts, tidying) when the cap is near.
- Long chats degrade. Start a new chat for each question and paste in a short "bid facts" note, a half-page of verified facts about the business. In Claude, five free projects let you keep one note per bid; ChatGPT Free also has projects, with up to five files in each.
- File limits. Paste text instead of attaching large PDFs. You control what the tool sees and avoid wasted uploads.
- Memory you do not control. Do not assume the tool remembers your business details. Paste them each time, from a note you trust.
- Different tools, different strengths. Try one answer on two plans. Keep the draft that sticks closest to your evidence, not the one that reads best.
Privacy on free plans
Consumer plans from the main vendors let you switch off use of your chats for training in privacy settings, and business plans do not train on business content by default. A free plan is a consumer plan, so switch the setting off before the first paste and check the vendor's page for current terms. Keep out of the chat: customer names, staff details, bank information, any figures covered by a confidentiality agreement and unpublished intellectual property. For drafting, round the numbers or describe them ("a six-figure turnover") and enter the real ones in the final form. A grant body will also ask who has seen your commercial information; "an AI tool, with training switched off, and no customer data" is an honest answer.
Follow the funder's rules on AI use
Some funders say nothing about AI, some ask you to declare it, and some restrict it. Look in the guidance notes, the terms and the declaration section of the form. If the rule is clear, follow it. If it is unclear, ask the programme contact: a one-line email costs nothing. Whatever the rule, you remain responsible for every statement in the application, so the review is human. Keep a record for each bid: which tool, which plan, which sections it helped with, and what you checked. That record is useful if a funder asks, and it helps you improve the next bid.
A week-long timeline for a small business bid
An illustration for a ten-person kitchen fitter applying for a workshop equipment grant with a Friday deadline.
- Monday: read the guidance, build the scoring sheet and the fit table; decide go or no-go.
- Tuesday: gather the missing evidence: a second quote, the match funding letter, job descriptions.
- Wednesday: draft answers one question at a time; the heaviest sections first while the cap is fresh.
- Thursday: budget table, narrative and claim checks; a second person reads the whole bid as a scorer would, with the criteria next to it.
- Friday morning: final word counts, attachments, submit with time to spare.
The timeline works on free plans because each day uses a bounded amount of AI. If you would rather build a standing routine for finding and preparing bids across the year, that is the sort of thing my AI implementation consultation can map with you. And if your funder is a charity or a sports body rather than a business scheme, this tutorial on a sports club grant application covers what those funders look for.
Free AI and grant bids: common follow-ups
Which free plan is best for a grant application?
None wins on every task. Claude's free plan handles long documents and careful editing well but has 5-hour and weekly caps. Gemini's free plan includes Deep Research for background. ChatGPT's free plan is the most familiar. Run the same section through two and keep the one that follows the funder's wording most closely.
Do funders ban AI in applications?
Some do not mind, some require disclosure and some restrict it. The rule is in the guidance notes or the application form, so read it before drafting and follow it exactly. If the guidance is silent, ask the funder or keep a record of how AI was used so you can describe it honestly.
Can I upload the whole funder guidance PDF to a free plan?
Often yes, but limits apply and differ by plan. ChatGPT Free allows three file uploads a day. Claude reads PDFs up to 1,000 pages, but analyses charts and layout only up to 100 pages. If an upload fails, paste the eligibility and scoring sections as text instead, which is also better practice because you control exactly what the tool sees.
Should I paste my financials into a free plan?
Keep it minimal. Use rounded or masked figures for drafting, and enter the exact numbers yourself in the final budget. Free consumer plans have training settings you can switch off, but business-grade privacy comes with business plans. Never paste bank details, customer data or anything covered by a confidentiality agreement.
Further reads
- Grant Writer Rates: What to Charge When AI Drafts Faster — What a grant writer charges when AI drafts faster.
- Heat Pump Quote Template: Explain Options With AI, No Savings Promises — Present supplier quotes clearly, a skill bids need.
- How to Manage a Community Centre: Where AI Helps and Where It Cannot — Where AI helps a community organisation's admin.
- How Freelancers Use AI for Invoices, Proposals and Chasers — Proposal writing habits that carry over to bids.
- Can AI Help a Small Architecture Practice? Where It Saves Time — Another small-firm example of AI saving admin time.
- Grant Writing AI Agent: What It Can Draft in a Business Grant Bid — What an AI assistant can draft in a small firm's grant bid, the eligibility and figure checks a person must do, and a reusable Project set-up to copy.
- ChatGPT Prompts for Grant Writing: Needs to Outcomes, Fact-Checked — Eight ChatGPT prompts for needs statements, project descriptions and outcomes, each paired with a verification checklist for invented evidence.
- Logic Model Template: Build One With AI From a Client's Programme — Build a logic model from a client's programme notes with AI, then gap-check the causal links and match the funder's wording, with a filled example.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: ChatGPT pricing page and OpenAI's file uploads help article, Claude's pricing page and file upload help article, Google's Gemini plans page and Gemini Apps usage-limits help page, all read on 4 November 2026.