Hiring is a writing job wearing a deadline. Every open role needs a job description, a screening list, an outreach note, a follow-up, a scorecard, and a stack of resumes read closely enough to be fair. Multiply that by five open roles and the writing alone can eat your week, which is how good candidates end up waiting three days for a reply they should have had in an hour.
One agent, running on Keimodel credits, takes the drafting off your plate without taking the judgment. It writes the job description, builds a structured question set that asks every candidate the same things, summarizes a resume against the role, and turns your notes into a clean scorecard. It can read an uploaded resume or CSV and pull out what matters. The rule that makes this safe is simple: the agent drafts, a human decides. Anything sensitive, an offer, a policy, a termination note, goes to HR or legal before it goes anywhere near a candidate or an employee.
Two guardrails are worth stating up front. Score candidates on skills and experience only, never on protected characteristics like age, race, gender, religion, disability, or family status, and keep a person in the loop on every decision. Set your roles and voice in Memory, and the prompts below give you consistent, reviewable drafts you can move on the same day.
Capabilities this leans on
Do this first. The agent will write to your roles, your voice, and your guardrails, so every draft starts consistent and fair.
Remember these facts about our hiring: we're a 60-person SaaS company, we hire for engineering, sales, and support, and our employer voice is straightforward and human, not corporate. We always state the salary range in job posts and we're an equal-opportunity employer. House rule: evaluate candidates only on skills, experience, and role-relevant criteria, never on age, race, gender, religion, disability, family status, or any protected characteristic. Anything sensitive (offers, policies, terminations, anything legal) is a draft only, for HR or legal to review before it's used.
Give the basics; get a post in your voice with the range and an inclusive close.
Write a job description for a mid-level backend engineer, remote within the US, salary range $130k to $160k. Cover what they'll work on, what we need (skills and experience, not buzzwords), and why someone would want the job. Plain, human voice, include our equal-opportunity line and the salary range.
Now scan that post for any wording that could discourage qualified people or read as biased, and rewrite those lines to be neutral and inclusive.
Give me a 280-character version for LinkedIn and a two-line version for a Slack referral channel.
What you get: A full job post in your voice, checked for inclusive language, plus short versions for every channel.
Ask every candidate the same role-relevant questions so the comparison is fair.
Build a structured phone-screen script for that backend engineer role: 6 questions on skills and experience only, each with a one-line note on what a strong answer looks like. Nothing about age, family, health, or background that isn't job-related.
Add two short scenario questions that test how they'd debug a slow query and how they handle a disagreement with a teammate.
Save this as a Skill called 'Backend screen' so I can reuse it, and remind me that scoring should be on the answers only.
What you get: A consistent, role-relevant screening script saved as a Skill you can run on every candidate.
Draft the whole candidate sequence so nobody waits days for a reply.
Write a short, warm outreach note to a passive candidate for the backend role: reference their relevant experience, say why I think it's a fit, and ask for a 20-minute call. Under 90 words, no fake flattery.
Write a 3-touch follow-up sequence for a candidate who went quiet after a first screen: a day-3 nudge, a day-7 'still interested?', and a day-14 polite close that leaves the door open.
Write a kind, specific rejection note for a candidate we screened but won't move forward, focused on fit for this role, never on the person. Keep it human and short.
What you get: Outreach, follow-up, and a respectful rejection, all drafted so every candidate hears back.
Standardize the write-up so the hiring panel compares like for like.
Turn my rough interview notes (pasted below) into a structured scorecard: rate technical skill, problem-solving, and communication on a 1 to 5 scale with a one-line justification each, list strengths and concerns, and give a clear recommend or not. Base it only on what's in my notes, and skills only.
Flag anything in my notes that's about a protected characteristic or otherwise not job-related, and leave it out of the score.
Draft the two-line summary I'll post for the hiring panel with my recommendation and the single biggest open question.
What you get: A consistent, defensible scorecard built only from what you observed, with non-job-related notes flagged out.
Upload the resumes; get fair, skills-based summaries instead of skimming at midnight.
I'm uploading five resumes for the backend role. For each, give a 4-line summary: years of relevant experience, the strongest matching skills, any gap against the must-haves, and one question I'd ask in a screen. Skills and experience only, ignore anything about age, photos, or personal details.
Now rank them by fit to the role's must-have skills and explain the ranking in one line each. This is a first pass for me to review, not a decision.
Draft the screen invites for the top three in our voice, with two time slots each.
What you get: Fair, skills-based summaries and a first-pass shortlist you review before anyone gets contacted.
Get a solid first draft fast, then send it to HR or legal before it's used.
Draft an offer letter for the backend engineer: $145k base, standard benefits, remote, start date to be confirmed. Warm and clear. Mark it clearly as a draft for HR and legal to review before sending, and leave blanks for anything I should confirm.
Draft a short, plain-language remote-work policy section on core hours and response expectations. Same rule: this is a draft for HR or legal to approve, not final.
(From Slack) Quick check: candidate asked for $160k against our $130k to $160k range. Draft a calm reply that we can meet that and confirm next steps, marked as a draft for me to approve.
What you get: Ready-to-review drafts of sensitive documents, each labeled so a human signs off before it goes out.
SaaS founders
Landing copy, changelogs, lifecycle emails, support macros, and churn analysis.
Sales & outreach
Research prospects, personalize cold email at volume, and turn calls into CRM notes.
Freelancers & consultants
Proposals, scoping, positioning, and client updates without the admin tax.
Open the Agent, paste any prompt above, and change the details to fit your business.