Why this exists
Four broken loops, one platform
Most talent tools are transaction processors — they record decisions but don't reason across them.
Sutra connects four people who currently have no shared view: the recruiter filling a role, the manager losing someone, the employee ready to move, and the hiring manager who needs them.
🏢
Internal talent goes unnoticed
547 Airbnb employees match at least one open role right now. Hiring managers default to external search — incurring cost, ramp time, and the risk of losing the internal employee when they feel overlooked.
📂
Strong candidates expire prematurely
External candidates who didn't get an offer — due to timing or headcount — are abandoned in ATS limbo. A candidate strong enough to enter the pool in Year 1 is often still relevant in Year 3.
📉
Managers can't see what's coming
People managers have no early-warning system for flight risk. By the time someone resigns, the signals were visible for months — in tenure data, promotion gaps, and 1:1 cadence — but no tool connected them.
🔒
Employees have no agency
Internal candidates have no way to signal openness without tipping off their manager. They leave because they never knew an opportunity existed, or because the process felt opaque and political.
The through-line: Greenhouse records what happened. Sutra reasons about what should happen next — across recruiters, managers, and employees simultaneously. "Greenhouse is your transaction processor. Sutra is your talent memory."
Who uses Sutra
Four roles, one platform
Each persona has a different scope of access. The asymmetry is intentional — churn risk is sensitive personnel data, and employee mobility signals are private by default.
👩💼
Jordan Rivera · Recruiter
Full access · Senior Recruiter, Tech · 12 open roles
GoalFill roles faster, reduce regrettable attrition
Pain todayManually building context across ATS, spreadsheets, and Slack
Signal neededWho's ready to move, who's going cold, who might leave
Access
✓ Open Roles
✓ Role Breakout
✓ Candidate Packet
✓ At Risk
✓ Settings
👩💻
Priya Sharma · Hiring Manager
Scoped access · Sr. Director, Trust & Safety PM · 1 open role
GoalHire the right person for their team, with minimal friction
Pain todaySlow feedback loops with recruiting, unclear pipeline status
Signal neededWho to interview next, what's waiting on them
Access
✓ Open Roles (scoped)
✓ Role Breakout
✓ Candidate Packet
At Risk
Churn scores
🧑💼
David Park · People Manager
Team-scoped access · Head of Growth Operations · 4 direct reports
GoalRetain and grow his team, spot flight risk early
Pain todayNo structured way to track promotion readiness or retention signals
Signal neededWho needs attention, who's ready for a new challenge
Access
✓ My Team
✓ Team Insights
✓ Member Profiles
✓ HM Objectives
At Risk (org-wide)
👩🔬
Camille Dubois · Employee
Self-only access · Operations Analyst L3 · 41mo in role
GoalExplore internal moves without risking her current role
Pain todayNo transparent way to signal interest or understand what roles she'd match
Signal neededWhat roles she matches, what gaps exist, who has endorsed her
Access
✓ My Profile
✓ Role Matches
✓ Skills Gap
✓ Endorsements
Other employees' data
Journey · Recruiter
From portfolio triage to candidate decision
A recruiter managing 12 open roles across multiple hiring managers. The journey traces a morning triage session that surfaces a stalled role, an internal match, and a retention risk — all without leaving Sutra.
👩💼
Jordan Rivera · Senior Recruiter, Tech
Managing 12 open roles · Airbnb HQ · 4 years recruiting
sutra · PM — Trust & Safety
sutra · PM — Trust & Safety · Internal (3)
sutra · Mia Tanaka · Candidate Packet
sutra · At Risk · 312 flagged
Journey · Hiring Manager
Scoped view, same intelligence
The HM sees only their own role instances. No At Risk tab. No churn scores. Otherwise the same
action band and candidate experience — designed so HMs get the context they need without accessing
sensitive personnel data that should stay with recruiting.
👨💻
Priya Sharma · Senior Director, Trust & Safety PM
Hiring for 1 open role · PM — Trust & Safety (L4)
sutra · PM — Trust & Safety · HM view
Journey · People Manager
From flight risk signal to resolved action
A people manager who has no structured system for tracking team health. The journey traces a morning check-in where a flight risk surfaces, David reads the full context on a team member's profile, and takes action — without a single Slack message to HR.
🧑💼
David Park · Head of Growth Operations
4 direct reports · Growth Operations · 3 years managing at Airbnb
sutra · Manager — David Park · My Team
sutra · Manager · Camille Dubois
sutra · Manager · Team Insights
Journey · Employee
Private exploration, visible when it counts
An employee who wants to explore internal opportunities without risking her current role or manager relationship. The journey shows how Camille uses Sutra to understand her matches, track her skill gaps, and signal interest on her own terms.
👩🔬
Camille Dubois · Operations Analyst L3
41 months in role · Growth Operations · Exploring a PM transition
sutra · My Profile — Camille Dubois
sutra · My Profile — Camille Dubois
Non-negotiables
Design principles that must hold
These are product commitments, not preferences. Every design decision should be evaluated against them.
🔍
No hallucinations
Every AI output must be source-traceable. No inference is displayed without an auditable signal. Self-reported skills are always labeled as such.
🤝
Human in the loop
Sutra never takes autonomous action on a candidate or employee. No auto-advance, no auto-reject, no auto-outreach. Every consequential action requires a deliberate human click.
🔒
Privacy by default
Employee mobility signals are private unless the employee opts in. Churn risk data is recruiter-only — it never surfaces to hiring managers or people managers directly.
✅
Internal before external
Every open role checks internal matches first. The internal candidate count is the first thing visible on every role card — not an afterthought buried in filters.
👁
Action over awareness
Every screen drives toward a specific next action. Awareness without a clear path to act is noise. If Sutra surfaces a signal, there must be a button to respond to it.
♾️
Talent memory compounds
Candidates, endorsements, and 1:1 notes don't expire. The value of the talent graph grows over time — a strong candidate from Year 1 is often still relevant in Year 3.
Discussion
Open questions for this prototype
Tensions this journey surfaces that need resolution before v1 scoping.
Q1
What does "Create promotion plan" actually create — and who owns it after the click?
Right now the resolve action is a state change in Sutra. A plan that lives only in Sutra won't drive HR action. Does this need to generate a structured artifact, trigger an HRBP workflow, or push into a performance system? The follow-through mechanism is undefined for v1.
Q2
What motivates an employee to give endorsements proactively — and what stops them from gaming the system?
Camille can endorse Sasha, but the current flow is entirely opt-in with no nudge. Without a participation driver, the endorsement graph will be sparse. Conversely, if endorsements visibly boost match scores, reciprocal endorsement rings could corrupt the signal. What's the right feedback loop design?
Q3
How do manager resolve actions sync back to the ATS and HRIS — or do they need to?
When David snoozes a flight risk or marks a promotion plan in progress, that context lives in Sutra. If the HRBP or recruiter is simultaneously managing this person in Workday or Greenhouse, there's no shared state. Is Sutra the system of record for these actions, or does it need a write-back integration?
Q4
Should Manager Objectives be editable by the manager — or are they set top-down by their manager or HR?
The current prototype loads HM objectives at initialization. In practice, some objectives will be manager-authored, some assigned by their own manager, and some HR-mandated. Getting the edit permissions wrong here creates a political problem: a manager who can erase a "missed" objective, or one who can't add their own goals, won't trust the view.
Q5
Is there a succession planning workflow, or does Sutra only surface risk after it's already acute?
The current manager journey responds to existing flight risk signals. There's no forward-looking view — no "if Camille leaves in 6 months, who is closest to ready?" scenario. Succession planning is a distinct need that the tenure and skill data could support. Is that in scope for v1, or explicitly deferred?
Q6
Does an employee know what their manager can see — and is the current visibility model consistent across all four personas?
Camille's "Open to moves" toggle is private, but her match count (3) is visible to David on Team Insights. A recruiter sees her churn risk; David does not. These distinctions are principled but not communicated to Camille anywhere in the prototype. The transparency layer — what each persona sees about others — needs explicit documentation before any of the four views go to real users.