01
Onboarding coordination
Accounts, buddies, first-week schedule — chased across teams until each step is verifiably done.
Solutions · People
The AI people-ops employee your team actually hears back from.
Onboarding coordinated to done, policy questions answered with the source attached, records kept current, survey themes summarized — while offers and compensation stay entirely human.
Offers & comp never automated Policy answers always cite the source Confidential by scope, not promise
people · ops
This week People · Harbor
TaskState
Day-1 pack sent · equipment confirmedDONE
Parental leave question — policy citedANSWERED
PTO conflict in support coverageFLAGGED
Offer letter for reviewHUMAN
23 REQUESTS HANDLEDEVERY ANSWER SOURCED
<10m
Median answer time on policy questions
100%
Onboarding steps verified done
0
Automated compensation decisions
Representative results · your workload will vary
The job
What your people employee owns.
02
Policy Q&A
Answers from your handbook with the section cited, and escalates when the handbook doesn't say.
03
Records hygiene
Keeps titles, managers, and leave balances consistent across systems — as held proposals.
04
Survey summaries
Turns engagement feedback into themes and verbatims, anonymized by rule before anyone reads it.
Guardrails
Helpful with people. Careful with people data.
Scope is the protection: the agent sees only the fields its tasks require, answers only from approved sources, and the sensitive core — offers, compensation, performance, disputes — is excluded by structure.
Access itself is logged, not just actions. Every decision stays on the record.
people · guardrails
AreaState
Handbook answers (sourced)AUTO
Record updatesHOLD
Survey dataANONYMIZED
Offers · comp · performanceHUMAN ONLY
Integrations
Plugged into your stack.
Scoped to people workflows — coordinating onboarding doesn't grant reading compensation, ever.
GM
GmailEmployee comms
SL
SlackQuestions & nudges
NO
NotionHandbook
GO
Google CalendarScheduling
GO
Google DriveDocuments
ZO
ZoomOnboarding sessions
+ thousands morevia Syntrum connectionsPolicy questions get answered in minutes with the handbook section attached. Our team finally trusts the answer — and the process.
Straight answers.
Offers, compensation, performance reviews, and disputes. These are excluded by scope, not by instruction — the agent cannot see the data, so it cannot act on it.
From your handbook and approved policies only, with the section cited in every answer. If the source doesn't cover it, the question routes to your team instead of being improvised.
The agent operates on least-access scope per task, survey feedback is anonymized by rule before summarization, and every read and write is on the audit trail.
It's one standard role on any plan — from $12/month with your own API key on Solo, or on Team with model usage included. There are no per-seat charges for the people reviewing its work.
Start small
Put one agent to work this week.
Start with one repeatable job. Keep approval over every action, see the results, and expand when you are ready.