Founding partners: 10 teams get hands-on setup and a locked price for 12 months. Apply →

Product / MCP

Run your hiring from ChatGPT or Claude.

Connect Ryuc to the AI assistant you already use. Ask what happened, and let it do the work: post a job, screen CVs, invite the shortlist, book round two, draft the offer. It asks before it sends anything.

Answers in one question.

A role over a week or a month, or one candidate from application to offer.

Work, not just answers.

The assistant chains the steps: from a pasted JD to a posted job to a screened shortlist.

Your rules apply.

The same checks, templates and activity log as the app. Read-only keys if you want.

Examples

What to type, tool by tool.

Two or three things you could say for each, in plain words.

Ask anything

Numbers, reports and stories, for a day, a week, a month or any dates.

15 tools
get_dashboard

Hiring numbers for today, 7 or 30 days: applied, screened, invited, passed and failed round 1, take-homes, rounds, offers and joined.

  • How many people passed round 1 today?
  • How many applied this week?
  • Give me this month in numbers.
role_report

One role over a week, a month or custom dates: sources, screening, interviews, top candidates, drop reasons and what is stuck.

  • How did the data analyst role go this month?
  • Weekly report for every open Acme role.
  • What is stuck on the sales role?
candidate_story

One candidate from application to offer, in order: every email, reply, score, decision and who made it.

  • What happened with Priya so far?
  • Did Arjun reply to the round two email?
  • Why was Sam dropped, and by whom?
search_candidates

Find people by name, email, role, stage, CV suggestion or interview decision.

  • Show me everyone shortlisted for the sales role.
  • Who failed round 1 this week?
  • Find Rahul Mehta.
get_candidate

One person: details, CV screening summary, latest interview, rounds and take-homes.

  • Pull up Arjun's profile.
  • What did the CV screening say about Neha?
list_interviews

Finished AI interviews in a period, newest first, with score and decision.

  • List today's interviews with scores.
  • Which interviews this week need a decision?
get_interview_report

The full scorecard of one interview: summary, strengths, concerns, rubric scores, questions asked.

  • Show me Priya's scorecard.
  • What did the interviewer not get to ask her?
  • Compare the top two interviews.
list_roles

Open, on-hold or closed roles, with how many people are at each stage.

  • Which roles have nobody at offer yet?
  • List all open roles for BlueRiver.
list_clients

The companies you hire for, with their open roles.

  • Which clients have open roles?
  • How many roles is Acme running?
list_upcoming_rounds

Rounds booked in the next days, and requests still waiting for a time.

  • What interviews are on this week?
  • Who has not picked a time yet?
list_offers

Offers, offered pay, joining dates and placement fees.

  • Who is joining in November?
  • Which offers are still unanswered?
list_take_homes

Take-home assignments per role, who got them and who submitted.

  • Who has submitted the take-home?
  • Which take-homes are overdue?
get_take_home_submission

One submission with the AI review against your rubric.

  • Read Maya's take-home and its review.
  • Did anyone score above 80 on the take-home?
get_screening_batch

Progress and scores of a batch of CVs, as they finish.

  • Is the screening done?
  • Who came out on top in that batch?
get_job_board

Your careers page, live jobs, applications per job and per source.

  • Where are applicants coming from this month?
  • How many applied to the analyst job?
  • Send me my careers page link.

Roles and job posts

From a pasted job description to a live job, without opening the app.

4 tools
create_role

Creates the role from a JD: skills, level, the interview style (technical, product, leadership, executive, general) and its rubric, coding or case study on or off.

  • Here is the JD for a product manager. Set it up with a case study.
  • Create a senior backend role with a 15-minute coding task in Python.
  • Make a director of sales role for Acme.
post_job

Publishes the role on your careers page, with Google for Jobs data and the job feed: location, type, pay shown or hidden, client named or not, questions for applicants.

  • Post it in Bengaluru, hybrid, 30 to 40 lakh, pay shown.
  • Post the ops role as remote, and ask for notice period.
  • Keep the client confidential on this one.
unpublish_job

Takes a job off the careers page and the feed; applicants stay.

  • The analyst role is filled, take it down.
  • Pause the sales job for now.
sync_job_boards

Tells search engines about every live job now.

  • Sync the job boards.
  • I changed the pay on two jobs, push the update.

Candidates and CVs

Drop CVs into the chat; they are screened like any upload.

4 tools
upload_cvs

Your assistant reads the CVs you share (PDF, Word, images, text) and sends their text; each is scored against the role. Hidden instructions are flagged, never followed.

  • Screen these 12 CVs for the backend role.
  • Here are three referrals for the PM job, add and screen them.
add_candidate

Adds one person by hand, without screening.

  • Add Rahul, referred by Meera, to the analyst role.
  • Add a walk-in: Kavya, kavya@example.com, sales.
screen_cvs

Screens people already in your list, in the background.

  • Screen everyone imported yesterday.
  • Re-screen the sales candidates against the new JD.
move_stage

Moves people through the pipeline, with a reason when they drop.

  • Move the top three to submitted.
  • Drop everyone below 40: rejected at CV screening.

Interviews and assessments

Invites, links and take-homes, sent in your name.

4 tools
send_interview_invites

Emails AI-interview links from Ryuc in your name, with your template. Asks you first.

  • Invite everyone shortlisted above 75.
  • Send Priya and Arjun their interview links.
  • Invite the 5 best from today's batch.
create_take_home

A take-home drafted from the JD, with a rubric only reviewers see, or your own task.

  • Draft a two-hour take-home for this role.
  • Use this task as the take-home, due in 3 days.
send_take_home

Sends each candidate a personal take-home link with a deadline.

  • Send the take-home to the three who passed.
  • Send it to Arjun only.

Scheduling

Round two, booked from the chat.

5 tools
ask_for_availability

Emails the candidate a time picker; the interviewer confirms; both get a calendar invite and reminders.

  • Get a second round with Neha for Arjun next week.
  • Set up a 60-minute final with the client for Priya, London time.
set_round_time

Books a time agreed elsewhere, or moves a booked round; invites update for both.

  • Move it to Thursday at 4 pm.
  • Book Sam for 3 November, 11:00 IST.
ask_for_new_times

Cancels the booked time and asks the candidate to pick again.

  • The interviewer is sick, ask Priya for new times.
  • Reschedule Arjun's round.
cancel_round

Cancels a round and tells both sides.

  • Cancel Sam's round two, the role is on hold.
resend_scheduling_email

Sends the current step again: the time picker, the confirm link or the invites.

  • Nudge Priya about picking a time.
  • Resend the calendar invite to the interviewer.

Emails, offers and joining

The end of the funnel, handled.

5 tools
send_rejection

The "not moving forward" email from your template, and a move to Dropped with the reason. Asks you first.

  • Let the ones below 50 know, kindly.
  • Send a rejection to everyone who failed round 1 this week.
send_email

An email you wrote together, sent in your name; replies come back to your team.

  • Email Priya that the client loved her and will be in touch.
  • Ask Arjun for his notice period.
update_offer

Records the offer status, offered pay and joining date.

  • Mark Priya's offer sent: 24 lakh, joining 1 December.
  • Arjun accepted.
create_offer_letter

Drafts the offer letter for India, the US, the UK or Australia, with that country's usual clauses.

  • Draft an India offer for Priya: 24 lakh CTC, joining 1 December.
  • Make a UK offer letter for Sam, £65,000, 3 months probation.
  • US offer for Maya in California, $140k, at-will.
confirm_joining

Marks the joining day: placement, fee from the client's terms and the replacement period.

  • Priya joined today.
  • Arjun started on Monday, what is our fee?

Agentic

Whole jobs, not single steps.

Ask for the outcome; the assistant picks the tools and checks with you before anything goes out.

From JD to shortlist, in one chat

  1. Paste the JD: create_role sets it up with the right rubric.
  2. "Post it": post_job puts it on your careers page.
  3. Next day: "Screen these CVs" (upload_cvs), then "who is on top?"
  4. "Invite the top five": send_interview_invites, after you confirm.

The Monday review

  1. "How did last week go on every open role?" role_report for each.
  2. "Who is stuck?" Shortlisted but not invited, interviews needing a decision.
  3. "Invite them, and remind the ones who haven't started."
  4. "Let everyone below 50 know." send_rejection, after you confirm.

From interview to offer

  1. "Tell me about Arjun's interview." candidate_story and the scorecard.
  2. "Book round two with Neha next week." ask_for_availability.
  3. "He passed. Draft an offer: 30 lakh, 1 December." create_offer_letter.
  4. "He joined today." confirm_joining works out the fee.

Set up and safety

Two minutes to connect. You stay in control.

Set up, once

  1. Make a key. In Ryuc: Settings, AI assistants. Choose read-only or read and act.
  2. Add the connector. Claude: Settings, Connectors, Add custom connector. ChatGPT: Settings, Connectors (developer mode). Paste the link Ryuc gives you.
  3. Ask. Claude Code and Cursor work too, with mcp.ryuc.app and the key as a header.

Safety, always

  • A key per person, shown once and revocable at any time
  • Read-only keys for people who only need answers
  • The assistant sees your workspace only and acts as you; every action is on the activity log
  • Tools that email candidates tell the assistant to confirm with you first
  • Text written by candidates (CVs, answers) is treated as data, never as instructions
  • Limits per call and per day on emails and CVs
What is MCP?

The Model Context Protocol: an open standard that lets AI assistants use other apps' tools. Ryuc runs an MCP server at mcp.ryuc.app.

Which assistants work?

Claude (web, desktop and Claude Code), ChatGPT with developer-mode connectors, Cursor, and any other assistant that supports remote MCP servers.

Can it send emails without asking?

Its tools tell it to confirm with you first, and a read-only key cannot act at all.

How do CVs get from the chat into Ryuc?

Assistants cannot hand over files, so yours reads each CV and passes its full text. It is screened exactly like an upload, with hidden instructions flagged.

Does my data train their models?

That depends on your assistant's own settings and plan. Ryuc only shares what the assistant asks for, for your workspace.

Let Ryuc take the first round.

A 15-minute demo with your own job description.

Early access

Ryuc is invite-only for now.

Leave your work email and our team will reach out to set you up.

Already have access? Open your workspace