How to Write a Letter of Recommendation With AI

Someone you respect asks you to write a letter of recommendation. You say yes immediately — they deserve it. Then you open a blank document and realize you have no idea how to start. You know they’re talented, you’ve seen them work, but translating that into a persuasive, well-structured letter feels like a different skill entirely.

You’re not alone. Most people write recommendation letters so rarely that every one feels like the first time. Managers write them once or twice a year. Professors get asked by dozens of students but still struggle with phrasing. The result is either a generic two-paragraph letter that doesn’t help the candidate or a rambling page that buries the strongest points.

This guide breaks down what makes recommendation letters effective, shows you the differences between professional and academic formats, walks through using an AI recommendation letter generator to produce a strong first draft, and provides two full examples you can study.

What Makes a Strong Letter of Recommendation

Hiring managers and admissions committees read hundreds of recommendation letters. Most are forgettable — they describe the candidate as “hardworking” and “dedicated” without offering any evidence. The letters that actually influence decisions share three structural qualities.

A Credible Opening

The first paragraph establishes who you are, how you know the candidate, and how long you’ve worked together. This is your credential. A letter from a direct supervisor who managed someone for two years carries more weight than one from a CEO who met them twice. State your relationship clearly: “I supervised Maria for 18 months as her marketing director at Acme Corp” tells the reader everything they need to know about your vantage point.

Specific Evidence in the Body

The body is where most letters fail. Saying someone is “a great team player” means nothing without a story. Saying “When our Q3 launch timeline was compressed by two weeks, she reorganized the content calendar, coordinated with three external vendors, and delivered every asset on the original deadline” paints a picture the reader can evaluate. Two or three detailed examples are more convincing than ten vague adjectives.

Quantify whenever possible. “Improved customer satisfaction” is weak. “Improved NPS from 32 to 51 over six months by redesigning the onboarding flow” gives the reader a concrete data point. Numbers ground your praise in reality.

A Forward-Looking Close

The closing paragraph should do two things: state your overall recommendation clearly (“I recommend her without reservation” or “She would be an excellent addition to your program”) and connect the candidate’s strengths to the specific opportunity. If you’re recommending someone for a product management role, mention how their cross-functional skills would translate. If it’s a graduate program, connect their research interests to what the program offers.

Professional vs Academic Letters: Key Differences

The two most common recommendation letter types — professional and academic — serve different audiences and follow different conventions. Writing one in the wrong format signals that the recommender doesn’t understand the context, which undermines the letter’s credibility.

Professional Recommendation Letters

• Audience: Hiring managers, HR teams, and sometimes executives making final decisions
• Focus: Workplace performance, specific accomplishments, leadership qualities, and how the candidate compares to peers in similar roles
• Tone: Direct and results-oriented. Readers want to know what the candidate did, how well they did it, and whether you’d hire them again
• Length: 350 to 500 words. Hiring managers scan quickly — get to the point
• Key elements: Measurable results, skills relevant to the target role, examples of problem-solving or leadership under pressure

Academic Recommendation Letters

• Audience: Admissions committees, scholarship panels, and fellowship review boards
• Focus: Intellectual curiosity, research ability, academic performance relative to peers, and potential for graduate-level work
• Tone: More narrative and personal. Committees want to understand how the student thinks, not just what grades they earned
• Length: 500 to 700 words. Academic committees expect more depth and context
• Key elements: Intellectual contributions (class discussions, research questions, thesis work), comparison to other students (“top 5% of students I’ve taught”), and growth trajectory

The AI recommendation letter generator lets you select between professional and academic formats, adjusting structure and language accordingly.

What to Include (and What to Leave Out)

Always Include

• Your relationship and context. Title, organization, how long you worked together, and in what capacity. This establishes credibility from the first sentence.
• Two to three specific achievements. Projects, results, or contributions with enough detail that the reader can picture the situation. One strong anecdote beats five generic claims.
• Character and work style. How the candidate approaches challenges, collaborates with others, and handles pressure. Use one brief example to illustrate.
• A clear endorsement. Don’t make the reader guess. State directly whether you recommend the candidate and how strongly.
• Relevance to the opportunity. Connect the candidate’s strengths to what the job or program requires. This shows you understand the context and aren’t sending a generic letter.

Leave Out

• Personal details unrelated to the role. Family situation, age, appearance, health, or religious beliefs. Besides being irrelevant, including these can create legal issues for employers.
• Generic superlatives without evidence. “Outstanding individual,” “exceptional talent,” “truly remarkable person” — these phrases appear in nearly every letter and carry zero information. Replace them with specifics.
• Criticism disguised as praise. “Despite their limited experience, they showed surprising ability” sounds backhanded. If you can’t write a genuinely positive letter, it’s better to decline the request.
• Information the candidate hasn’t approved. Performance issues, personal struggles, or reasons for leaving a previous role should not appear unless the candidate explicitly asked you to address them.

How to Write One With AI

The NavioHQ Letter of Recommendation Generator produces a complete first draft based on a few inputs. Here’s how to get the best results.

Start with the basics. Select the letter type (professional or academic), enter your relationship to the candidate (direct manager, professor, colleague), and note how long you’ve known them. The generator uses these details to frame the opening paragraph with the right level of authority.

Feed it specifics. The most important input is the candidate’s achievements and skills. Don’t write “good at communication” — write “led weekly client presentations for a $2M account and consistently received positive feedback.” The more concrete your input, the more convincing the output. List three to four bullet points of real accomplishments.

Set the tone and length. Choose formal for corporate environments or academic institutions, warm for nonprofits or creative roles, and concise for cover-letter-style submissions. The generator adjusts vocabulary and sentence structure to match.

Generate and review. The output is a structured letter with opening, body, and closing. Read it through once for accuracy, then move to the editing phase (covered in the next section). The generator handles the hardest part — getting words on the page in the right structure — so you can focus on adding the personal details that make the letter authentic.

Full Letter Examples

These two examples show what a strong finished letter looks like — one professional, one academic. Both were generated with the recommendation letter generator and then edited with personal details and anecdotes.

Professional: Senior Marketing Analyst

Dear Hiring Committee,

I am writing to recommend Elena Torres for the Senior Marketing Analyst position at your organization. I served as Elena’s direct manager at Meridian Digital for three years, overseeing her growth from a junior analyst to the person I relied on most for strategic campaign decisions.

Elena’s strongest quality is her ability to translate complex data into recommendations that non-technical stakeholders actually act on. When our CEO questioned the ROI of our paid social program, Elena built a multi-touch attribution model that traced $1.4M in closed revenue back to specific ad campaigns. That analysis saved the program from being cut and informed our budget allocation for the following two quarters.

She also redesigned our monthly reporting dashboard from scratch. The previous version took 12 hours to compile manually each month. Elena automated the entire pipeline, reducing reporting time to 45 minutes and surfacing metrics the leadership team didn’t know they could track. Three other departments adopted her dashboard template within six months.

Beyond technical skills, Elena mentors junior team members with genuine patience. She created a “data office hours” block every Thursday where anyone on the team could bring analytics questions. Attendance was consistently high, and two analysts she mentored have since been promoted.

I recommend Elena without reservation. She combines deep analytical thinking with the communication skills to make data accessible, and she does it while elevating the people around her. She would be a significant asset to any data-driven marketing team.

Sincerely,
James Whitfield
Director of Marketing, Meridian Digital

Academic: Graduate School Applicant

Dear Admissions Committee,

I am pleased to recommend Anika Patel for admission to your doctoral program in Cognitive Neuroscience. I have known Anika for two and a half years — first as her instructor in Advanced Research Methods and later as her thesis advisor for her senior honors project on attention and working memory.

Anika stands out in a way I rarely see at the undergraduate level: she questions assumptions. In my seminar, most students accept published findings at face value. Anika read Cowan’s seminal work on working memory capacity and raised a methodological concern about the visual change detection paradigm that I had not considered in 15 years of teaching the paper. That kind of critical engagement is what separates students who consume research from those who advance it.

Her honors thesis examined whether attentional load modulates false memory formation using a modified DRM paradigm. She designed the experiment, programmed the task in PsychoPy, recruited and tested 87 participants, and analyzed the data using mixed-effects models — all with minimal guidance. Her results showed a significant interaction between attentional load and false recognition rates, and I encouraged her to submit the work for publication. The manuscript is currently under review at Memory & Cognition.

Anika is also an exceptionally clear writer. Her thesis draft required fewer revisions than those of most graduate students I advise. She presents complex statistical analyses in accessible language without sacrificing precision — a rare combination.

Among the roughly 300 undergraduates I have taught and mentored over the past decade, Anika ranks in the top five. She has the intellectual curiosity, methodological rigor, and persistence that doctoral work demands. I recommend her with the highest enthusiasm.

Sincerely,
Dr. Robert Chen
Professor of Psychology, Westfield University

Editing AI Output Into Something Personal

An AI-generated draft gives you structure and phrasing. Your job is to make it sound like it came from you, about this specific person, for this specific opportunity. Here’s how to do that in 15 minutes.

Replace Generic Claims With Real Stories

The generator might produce: “She demonstrated strong leadership skills during complex projects.” That’s a starting point, not a final sentence. Replace it with what actually happened: “When our biggest client threatened to leave during a product migration, she stepped in as the primary point of contact, ran daily stand-ups for three weeks, and retained the account.” The reader doesn’t need to be told she’s a strong leader — the story shows it.

Add One Personal Observation

Include at least one detail that only someone who worked closely with the candidate would know. It could be how they handled a stressful week, a habit that made them effective, or a moment that changed your impression of them. “I knew Elena was ready for a senior role when she started anticipating questions from the VP before our weekly reviews — preparing answers for objections he hadn’t even raised yet.” These details are what AI can’t generate and what readers trust most.

Match Tone to Context

A letter for a startup should read differently than one for a law firm. If the AI draft feels too formal for the context, loosen the language. If it’s too casual for an academic committee, tighten it. Read the job posting or program description and mirror their tone — if they describe their culture as “collaborative and fast-paced,” use similar language in your letter.

Check for Placeholder Language

AI drafts sometimes include brackets like [specific project name] or [number of years]. Scan the entire letter for anything that looks templated. Every sentence should read as if it was written from scratch for this candidate and this opportunity.

Common Mistakes That Weaken Letters

Even well-intentioned recommendation letters can undermine a candidate. These are the patterns that hiring managers and admissions readers notice — and they’re all avoidable.

Being Too Generic

“John is a hardworking, dedicated employee who always meets deadlines.” This sentence could describe anyone. It tells the reader that either you don’t know the candidate well or you didn’t invest effort in the letter. Both interpretations hurt the candidate. Replace every generic adjective with a specific example.

Writing Too Much (or Too Little)

A two-sentence letter feels dismissive. A two-page letter suggests you can’t prioritize. Aim for one page — 400 to 600 words for professional letters, 500 to 700 for academic. If you find yourself stretching to fill space, you probably need more specific examples rather than more paragraphs.

Focusing on Yourself Instead of the Candidate

Some letters spend three paragraphs establishing the recommender’s credentials before mentioning the candidate. Your title and relationship belong in the first sentence. After that, every sentence should be about the candidate. The reader doesn’t need your career history.

Damning With Faint Praise

Phrases like “adequate performance,” “pleasant to work with,” or “met expectations” read as coded criticism. If you can’t write enthusiastically about the candidate, have an honest conversation with them about whether your letter will help their application. It’s better to decline than to send a lukewarm letter.

Sending the Same Letter Everywhere

A letter addressed “To Whom It May Concern” with no reference to the specific role or program signals that you didn’t tailor it. Change the addressee, connect the candidate’s strengths to the specific opportunity, and adjust the closing recommendation. The AI generator makes it fast to create customized versions for different applications.

Build Your HR Toolkit

Recommendation letters are one part of the hiring and talent management cycle. These tools cover the rest:

• Letter of Recommendation Generator — professional and academic letters with customizable tone and length
• Performance Review Generator — structured evaluations for self-reviews, manager reviews, and 360-degree feedback
• Job Description Generator — ATS-friendly postings for any role and seniority
• Interviewer Question Generator — behavioral, technical, and situational questions with scoring rubrics

If you’re writing performance-related content regularly, the performance review writing guide and job description writing guide cover those formats in detail.

Frequently Asked Questions

Open the Letter of Recommendation Generator, enter the candidate’s key achievements and your relationship, and you’ll have a structured draft ready to personalize in under two minutes. The candidate trusted you enough to ask — give them a letter that matches that trust.