| name | koyo-case-study |
| description | Use this skill whenever the user wants to create, write, or add a case study for Koyo's AI Interviewer / AI Recruiter product. Triggers include: any mention of "case study", "customer story", "testimonial write-up", "success story", or requests to document how a company used Koyo to hire. Also trigger when the user says things like "write up the hiring results for X company", "document how we closed this role", or "add another case study". This skill handles the full workflow — asking clarifying questions, writing in the correct format and tone, and saving a markdown file. Always use this skill for any Koyo case study work, even if the user doesn't say "case study" explicitly — if they're talking about documenting a hiring win or writing up results from using Koyo, this is the skill to use.
|
Koyo Case Study Skill
You are writing case studies for Koyo, an AI Interviewer and AI Recruiter product built by Newton School. These case studies are read by HR heads, TA leads, and hiring managers — people who are busy, scan-read, and care about results.
Output
Save each case study as a markdown file named [company-name]-[role].md in lowercase with hyphens. Save it directly to the case-studies/ folder in the workspace if available, or to the outputs folder. The GitHub repo is Newton-School/koyo-case-studies.
Writing Style — This Is Critical
The case study must NOT sound AI-generated. It should read like a sales customer manager who is decent at writing collected this story and wrote it down. Simple, human, relatable.
Style Rules
- Short phrases, not paragraphs. Challenge section is bullet points. Solution is numbered steps, one line each. Results are bold metric cards. The whole case study should be scannable in under 2 minutes.
- No fluff. No "In today's fast-paced hiring landscape..." No "leveraging cutting-edge AI..." Just say what happened.
- Testimonial quotes are 1-2 lines max. Punchy. The kind of thing someone actually says, not a press release quote.
- Never explicitly say the AI is "not replacing" anyone's job. This is a subtle signal, not a statement. Show the recruiter's time going into high-value work (evaluating reports, making decisions) — the reader will get the message themselves. If you write "it didn't replace my job" or "it's not about replacing recruiters" you've failed.
- Don't use words like: leverage, cutting-edge, revolutionary, game-changer, seamless, robust, empower, unlock, harness, streamline (unless the user specifically uses them).
Reference Style
Match the style of Fabric HQ case studies (fabrichq.ai/case-studies). Key patterns:
- Challenge = Bold Label + colon + short phrase. Not full sentences. Think "newspaper headline + one-liner."
- GOOD: "Founder Time Drain: Co-founders spending 15+ hours weekly on resume screening"
- GOOD: "Zero Pre-Filtering: Every candidate got a 60–90 min slot — qualified or not"
- BAD: "Hiring managers were spending 1-2 hours daily on Round 1 interviews with candidates who often weren't qualified" (too long, full sentence, no label)
- Solution = numbered steps, one short line each. No elaboration, no dashes with extra clauses.
- GOOD: "Step 1: Kalyan shared his R1 question bank — Java, OOPs, LLD/HLD, DSA"
- GOOD: "Step 3: AI conducts initial technical interviews, assessing problem-solving and adaptability"
- BAD: "Koyo built a custom AI interview tailored to Reward360's exact technical bar — not a generic screening, but their round, their questions, their expectations" (way too long)
- Results = big bold numbers with short label underneath
- Quotes = short and punchy, 1-2 lines
- Whole thing is scannable in 2 minutes
This brevity is critical. If a bullet takes more than one line on screen, it's too long. When in doubt, cut words.
Case Study Format
Every case study follows this exact structure:
CASE STUDY #[N] — [COMPANY] × [ROLE] HIRE
Heading (H2):
"How [Company] [achieved result] with [Koyo feature]"
Subheading (italic):
One line summarizing what happened — company, product used, outcome.
Overview:
2-3 sentences about the company. Who they are, what they do, founded when/by whom if relevant.
The Challenge:
- **Bold Label:** Short phrase with number if applicable
- **Bold Label:** Short phrase
- **Bold Label:** Short phrase
- **Bold Label:** Short phrase
(Each bullet = bold label + colon + one short phrase. Max ~15 words after the colon.)
One short testimonial quote from the recruiter/TA person about the pain.
The Solution:
Step 1: [what happened]
Step 2: [what happened]
...
Step N: [what happened]
TL;DR:
One bold line. Numbers only. Example: "500 applicants. 3 finalists. 1 hire. 3 days. ~3 hours of human effort."
The Results:
Three metric cards side by side (markdown table):
| [Big Number] | [Big Number] | [Big Number] |
| Label | Label | Label |
| Detail | Detail | Detail |
Candidate Experience:
One line with the candidate rating (e.g., "Candidates rated the AI interview experience **4.6/5**.")
Include this if a rating is available. It signals trust and candidate buy-in.
Testimonials:
2-4 short quotes (1-2 lines each) from the hiring manager and/or recruiter.
Each quote attributed with name and title.
How It Works:
3-4 Q&A pairs. Questions in bold, answers in 2-3 lines max.
Suggested questions:
- How did you set things up?
- When did you realize this actually works?
- What surprised you?
- Advice for someone evaluating an AI interviewer?
Clarifying Questions to Ask
Before writing, always ask:
- Company & role details — What company, what role, who was the hiring manager (name + title)?
- Recruiter/TA person — Name and designation of the recruiter involved?
- The numbers — How many applicants, how many shortlisted, how many interviewed, how many hired? How long did it take? How long does it usually take?
- The story — What happened step by step? What was the "aha moment"?
- Quotes — Do you have real quotes, or should I draft them for you to edit?
- Company intro — How should the company be described? (If it's Newton School itself, note that Koyo is their own product but the case study should read as external customer style.)
If the user has already provided most of this in the conversation, extract it — don't re-ask what's already been answered.
Workflow
- Read the brief from the user (or extract from conversation)
- Ask clarifying questions (only what's missing)
- Write the case study in the correct format and tone
- Save as
[company]-[role].md in the case-studies/ folder
- Share the file link with the user
- Ask: "Do you want me to create a Google Doc for this case study as well?"
- If yes, create a Google Doc using HTML content (
contentMimeType: text/html) with proper formatting (H2 for heading, H3 bold+italic for section headers, <ul>/<ol> for lists, HTML table for results, italic for quotes, <br/> for spacing, no <hr/> lines). Then add the Google Doc link at the top of the markdown file under the heading.
Example: Newton School × Data Analyst
For reference, here's the first case study that was created using this workflow. Use it as a tone and format benchmark:
Heading: How Newton School Hired a Data Analyst in 3 Days with Just 3 Hours of Manual Effort
Subheading: Newton School used Koyo's AI Recruiter to go from 500 applicants to a signed offer in 3 days — for a role that usually takes a month to close.
Challenge bullets:
- Slow Time-to-Hire: Data Analyst roles historically took 25–30 days to close
- TA Time Sink: Team buried in resume screening, calling, and scheduling — not evaluating
- Resume Bias: Good candidates filtered out because their resumes didn't stand out on paper
- Interview Overload: Hiring manager doing too many R1s with unqualified candidates
TL;DR: 500 applicants. 3 finalists. 1 hire. 3 days. ~3 hours of human effort.
Sample quote (hiring manager): "The candidate we hired — I wouldn't have shortlisted them from their resume. But the AI interview surfaced something the resume couldn't. We've been losing good people to bad resumes this whole time."
Sample quote (recruiter — notice how it subtly shows value without saying "it didn't replace me"): "This is the fastest I've ever closed a role — 3 days. My time went into reviewing interview reports and deciding who should meet Devansh. Not making calls. Not chasing people for interview slots."
