Best AI Tools for Writing a Recommendation Letter in 2026

Jotform's free recommendation-letter generator asks 4 questions, the candidate's name, your relationship to them, the job they want, and one line about why they're qualified, then hands back a finished letter in under 60 seconds. ChatGPT costs the same amount of money, zero dollars, and takes 5 times as long to get right, because the input is a blank paragraph instead of 4 boxes, and the output is only as specific as what you type into it.

There are now at least 8 viable ways to get an AI's help with a recommendation letter. General chat assistants like ChatGPT draft from a loose paragraph of notes. Claude does the same, with a cleaner default privacy setting. Dedicated form-based generators like Jotform ask for specific fields instead of a prompt. Template.net leans on formatting, turning a plain paragraph into a document that looks like it came off letterhead. Browser extensions like JetWriter AI polish a draft where you're already typing it, inside Gmail or Google Docs.

The real dividing line here isn't price. Most of these tools have a usable free tier. It's how much specific detail about the candidate you're willing to type in yourself, and how many letters you're actually writing this season. Pick wrong on that axis and you either spend 20 minutes writing a prompt a form would have handled in 60 seconds, or you fill out a rigid template that produces the same generic letter every reference-letter mill on the internet already writes.

How to Choose an AI Tool for a Recommendation Letter

Before picking a tool, answer 5 questions honestly. Each one points at a different pick.

1. How much detail do you already have written down? A resume or 3 real specifics beats a blank page.
2. One letter, or many this season? 20 letters is a different problem than one.
3. Does the subject's privacy matter here? A letter names a real person, not a marketing brief.
4. Draft in a separate app, or where you already write? Some people finish faster inside Gmail.
5. What's the actual budget? A one-time letter rarely justifies a paid plan.

The criteria matter more than any specific pick below. A reader who answers these 5 questions honestly will end up in roughly the right place, even ignoring the shortlist entirely.

What Makes a Letter Actually Good, No Matter Which Tool Writes It

The tool matters less than what you feed it. A recommendation letter succeeds or fails on specificity, and every tool on this list, prompt-based or form-based, inherits whatever you put in.

Before opening any of them, write down three things: one project or moment where the candidate stood out, one number if you have one, a percentage, a deadline hit, a team size they managed, and one sentence about how they're different from other people you've worked with. That's the raw material a general assistant like ChatGPT or Claude turns into prose. Neither tool can invent it for you.

Weak input produces the same letter every time, regardless of tool: "X is a hardworking and dedicated professional who consistently exceeds expectations." That sentence could describe anyone, which is exactly the problem admissions committees and hiring managers flag first. A tool given real specifics writes something closer to "X rebuilt our onboarding flow in six weeks, cutting new-hire ramp time from three weeks to eight days," a sentence no other candidate's letter could accidentally duplicate.

This is also where a resume helps more than most people expect. If you're pasting material into ChatGPT or Claude, a candidate's resume is a faster shortcut to specifics than trying to remember them from a conversation months ago. HokAI's Kickresume vs. Teal guide covers the tools on the other side of that document, useful if you're helping a student or a junior colleague put one together before you write for them.

None of the tools compared here will catch a generic letter and flag it as generic on its own. That check is still on the person hitting send. Read the finished draft once and ask whether a sentence could be swapped into someone else's letter without anyone noticing. If the answer is yes, go back and add one more specific before you sign it.

What These Tools Actually Cost

These 5 criteria are also how HokAI's own guide to choosing a general AI writing assistant frames the broader decision. If a recommendation letter is one of several writing tasks you use AI for, that guide, and the wider AI writing assistant category, are worth a look beyond this one task.

Every tool below has a free tier that can produce a usable letter as of September 2026. The paid tiers buy speed, higher usage caps, or features unrelated to letter writing at all.

Two of these numbers deserve a caveat. Template.net's paid Plus and Max tiers are credit-based, and the pricing page does not display a plain dollar figure for either as of this writing, only the credit allowances. And Anthropic's $17 Claude Pro price applies only to an annual plan billed up front; paying monthly runs $20, the same as ChatGPT Plus.

The Shortlist: What Each Tool Is Actually For

ChatGPT and Claude are the default pick for most people writing 1 or 2 letters, because a recommendation letter is fundamentally a prompting problem, not a form-filling one. OpenAI builds ChatGPT to be flexible enough to draft from a resume and a few specific accomplishments. Claude, from Anthropic, does the same.

The caveat is identical for both: a thin prompt produces a thin, generic letter. Neither tool can invent specifics you didn't provide, and both will happily hand back a competent-sounding letter built on nothing at all if that's what you feed them.

Jotform skips the prompting step entirely. Its recommendation-letter generator, confirmed live on Jotform's own site, asks for the candidate's name, your relationship, the job title, a language, and a short note on why they're qualified, then produces a full letter in seconds, editable afterward with adjustable fonts, colors, and a PDF export. It's free with no signup. The caveat: 4 short fields produce a shorter, more generic letter than a paragraph of real detail would in ChatGPT. Jotform is the fast option, not the specific one.

Template.net takes a third approach: a prompt-based AI Writer, also free with no signup, aimed less at wording and more at output that looks like a finished document. It bundles fonts, layout, and export options a plain chat window doesn't offer. It has no template built specifically for recommendation letters, but its general "professional letters" prompt path can produce one.

The gap here is pricing transparency. Past the free tier, Template.net's Plus and Max plans are sold in credits rather than dollars, so you won't know the real monthly cost until you're inside the checkout flow.

More Picks: JetWriter AI, Grammarly and Gemini

JetWriter AI is the pick if you'd rather draft the letter where you already write, most often Gmail. It's a browser extension, not a separate app. Its free tier includes 15 advanced AI responses a month plus a smaller allowance of "regular" responses on models including GPT-5 mini and Gemini 3 Flash. The Elite plan, $18 a month, adds roughly 1,000 advanced responses and 50 automated Gmail draft replies a day, built for a higher-volume use case than a single letter.

Grammarly doesn't draft a letter from nothing, and shouldn't be asked to. Its job is the last pass: tone, sentence structure, and catching the stiff, over-formal phrasing that recommendation letters are unusually prone to. The free tier covers spelling and grammar plus 100 AI prompts a month; Pro, $12 a month, adds full sentence rewrites and tone adjustment, the feature that matters most here.

Gemini is a fourth general-assistant option from Google, functionally similar to ChatGPT and Claude for this task. Google's own product blog lists a 4-tier pricing ladder, cheapest at $4.99 a month.

A full breakdown of Gemini against the other 2 general assistants is in HokAI's assistant-choosing guide. The short version for this task: all 3 are close enough that the deciding factor should be data handling and what you already pay for, covered next.

ChatGPT or Claude: The General-Assistant Question

Between the 2 general assistants, the honest answer is that either produces a comparable letter from a comparable prompt. The difference that matters here is data handling, not writing quality, and it's covered in full below.

Price is close enough not to be the deciding factor: ChatGPT Plus and Claude Pro both land at $20 a month billed monthly, and both have usable free tiers for a single letter. If you're already paying for one of the two for other work, use that one. Starting from zero, Claude's opt-out-by-default privacy posture, detailed next, makes it the slightly better fit for a document about a real person who never agreed to be part of a training run.

ChatGPT vs. Jotform: Prompting vs. Filling In a Form

This is the actual decision most readers of this guide are making, and it comes down to how much you're willing to type before you get a draft.

Jotform wins on speed. 4 fields, 60 seconds, done, and the output is a complete, properly formatted letter with a signature block. If you need something serviceable in the next 5 minutes and don't have strong opinions about the wording, Jotform is the better tool, full stop.

ChatGPT wins on quality, but only if you feed it real material. Paste in the candidate's resume, name 2 specific projects or moments, and describe your working relationship in a sentence, and ChatGPT will produce a letter with the kind of concrete detail that makes a recommendation letter credible instead of interchangeable. Give it the same 4 scraps of information Jotform's form asks for, and the two tools produce roughly the same letter, except ChatGPT took longer.

The winner, specifically: anyone who already has a resume or a few concrete accomplishments in hand should use ChatGPT or Claude. Anyone starting from nothing but a name and a job title, who wants something usable in under 2 minutes, should use Jotform.

Data Privacy: What Happens to the Details You Type In

A recommendation letter is, by definition, a document full of specifics about someone else: their performance, their character, sometimes their weaknesses. That makes the question of what happens to your input after you hit submit more relevant here than in most AI writing tasks.

Anthropic changed its consumer terms in 2025 so that Claude conversations are used to train future models by default unless a user turns that setting off, at which point conversations are kept for 30 days for safety and technical purposes, then deleted. Claude for Work, Claude Enterprise, and Claude for Education are excluded from training regardless of the individual setting. Turning off training on a personal account takes 3 steps:

1. Open Settings from the profile menu.
2. Go to Privacy, or Data Controls.
3. Turn off the option that allows conversations to train future models.

By contrast, OpenAI's consumer tiers keep training turned on unless you dig into settings and switch it off yourself, and even then a full opt-out with guaranteed non-retention is really only available on the paid Team or Enterprise plans. Jotform's generator sends a narrower slice of information than a full chat session, just a name, a relationship, a job title, and one qualifying sentence, though the same underlying vendor-data questions still apply to a structured form as to an open conversation.

None of this makes any of these tools unsafe to use. It does mean that typing a candidate's full resume and a paragraph about their personal struggles into a free ChatGPT session is a different privacy decision than typing 4 short fields into Jotform, and the tool you pick should account for that, not just for writing quality.

Who Should Skip All of This

0 is a marketing team platform, not a personal writing tool, and it shows in the price: $69 a month for a single seat, built around brand voice consistency and team collaboration.

Jasper's own pricing page is explicit that the product targets "content marketers, product marketers, and performance marketers," not someone writing one letter about a former employee. If a coworker suggests Jasper for a recommendation letter, that's the wrong tool for the job, not a wrong opinion about AI writing in general. Jasper, Jotform, and Template.net all sit in HokAI's AI content creation category, a useful place to browse if none of the picks above fit, but Jasper belongs in that category's marketing-team half, not its personal-letter half.

0 is worth a specific mention because it looks, on the surface, like it should be on this list. It's a 40-plus template library at $7.50 a month for unlimited generations, cheaper than nearly everything else here.

A check of Rytr's own use-case and pricing pages turns up no dedicated recommendation-letter or reference-letter template, only adjacent formats like cover letters and job descriptions. You can still get a recommendation letter out of Rytr with a generic prompt, the same way you could with ChatGPT, just without the letter-specific structure a form like Jotform's provides and without the broader reasoning a full chat assistant offers.

The Case for a Template Tool Instead

The strongest argument against everything above is volume. This guide is written for someone writing one letter, maybe 2. A professor writing 20 recommendation letters in a single graduate-admissions season is solving a different problem, and for that person, the verdict flips.

Re-prompting ChatGPT or Claude from scratch for each of 20 students means re-explaining the relationship, the program, and the format every single time, or maintaining your own external prompt template and pasting it in 20 times, which is real, repeated manual work. A form-based tool like Jotform, or a saved prompt template inside whichever assistant you already use, removes that repetition: the structure stays fixed, and only the candidate-specific fields change each time.

If you're in that position, the right pick is whichever tool lets you save a reusable structure, not whichever tool writes the single best sentence. That is a real, common case, and the general-assistant recommendation above is not built for it.

What Would Change This Verdict

Two things would move this guide's pick. If OpenAI ships a genuine, working full opt-out for ChatGPT's free and Plus tiers with no Team or Enterprise plan required, the privacy gap with Claude closes, and the decision goes back to being a coin flip based on whichever tool you already use.

And if any of the form generators, Jotform included, add an optional free-text field for real specifics instead of 4 short fixed fields, the speed-versus-quality tradeoff in the ChatGPT-versus-Jotform comparison above stops being a tradeoff at all.

Until one of those happens, the answer stays the same: bring your own specifics to a general assistant if you have them, use Jotform's form if you don't, and skip anything built for a marketing team.

If you're not sure which broader AI assistant is worth committing to beyond this one task, HokAI's Smart Match walks through that decision based on what you actually use AI for, not just this one letter.

Frequently asked questions

Can AI actually write a good recommendation letter?

Yes, if you give it real material. A general assistant like ChatGPT or Claude can turn a resume and two or three specific accomplishments into a letter with the structure and tone a hiring committee expects. Without that input, both a chat assistant and a form-based generator like Jotform produce the same generic result.

Is it dishonest to use AI to write a recommendation letter?

Not inherently. The letter still needs to reflect your own honest opinion of the candidate, since you're the one signing it. Using AI to structure and phrase that opinion is closer to using a template than to fabricating an endorsement.

Which AI tool is free for writing a recommendation letter?

ChatGPT, Claude, Gemini, Jotform, Template.net, and JetWriter AI all have a usable free tier for a single letter. Jotform's free tier includes its full recommendation-letter generator with no signup required.

Will ChatGPT keep or train on my recommendation letter?

ChatGPT's free and Plus tiers train on conversations by default, with no full opt-out short of a paid Team or Enterprise plan. Claude lets a user turn training off directly in account settings, after which conversations are deleted after 30 days.

What's the fastest way to get a recommendation letter written with AI?

Jotform's dedicated generator is the fastest option: four fields and under a minute to a finished, formatted letter. A general assistant like ChatGPT takes longer but produces a more specific result if you paste in a resume or a few real accomplishments first.

Covered in this guide

• ChatGPT: ChatGPT is OpenAI's AI assistant with 900 million weekly users and GPT-5.6 Sol, covering writing, coding, image generation, and web search with a free plan and Plus at $20/month.
• Claude: Claude API with Sonnet 5 (agentic, near-Opus performance at half the cost) and Claude Science (auditable research workbench). Free to Pro.
• Jotform: Jotform builds 10,000+ form templates with 150+ integrations. Free tier; plans from $34/month with HIPAA support.
• Anthropic: Anthropic, founded 2021 by 7 ex-OpenAI researchers, builds Claude and was valued near $965B after its May 2026 Series H round.
• Gemini: Google's multimodal AI model family for reasoning, coding, and creative tasks
• Google: Google (Alphabet, NASDAQ: GOOGL), founded 1998, serves 8B+ monthly Search users with Gemini 3.5, 190,820 employees, and $402.84B FY2025 revenue.
• Grammarly: AI-powered writing assistant for better communication
• Jasper: The agent workspace built for modern marketing teams to orchestrate intelligent agents and deliver end-to-end marketing workflows.
• JetWriter AI: AI writing assistant browser extension used by 600,000+ people for email replies, grammar fixes, summarization, and tone changes across Gmail, LinkedIn, and any website. Free and paid plans available.
• OpenAI: OpenAI builds the GPT-5.6 model family (Sol, Terra, Luna), o3, ChatGPT (900M+ weekly users), and the OpenAI API. Closed a $122B round at an $852B valuation in March 2026, the largest private funding round in history.
• Rytr: AI writing assistant with 40+ templates for short-form content. Free tier available; Unlimited from $7.50/month (annual). Used by 8 million+ creators.
• Template.net: Template.net is an AI Production Engine that converts text prompts into fully editable, production-ready documents, presentations, and designs. Free to generate, with credits-based premium plans.

Sources