Skip to content

Example Everything on this site is sample data — one of the configurations this template ships with, built from the same repository.

Example: AI use case catalog
Civic AI Community of Practice AI Use Case Catalog

Paused or retired · Mid-sized city — Human Resources

Internal tool

New-hire onboarding checklist assistant

Result in one line: Cut help desk tickets per new hire from 11 to 3 during the eighteen months it ran

A chat assistant that built a personalised onboarding checklist for each new hire and opened the routine access requests for them. Retired in 2026 when the HR system shipped the same feature.

Published January 27, 2026 Updated July 2, 2026

Documentation or write-up
Skills needed to set it up
Power user
Readiness
Reference only Human review built in
Data it touches
Personal information (PII) Internal, non-public data
Who sees the output
Internal staff
License
Not open source — description only
Cost to stand up
No new spend
Portable to other platforms
Partially — with rework
Use case category
Administrative & task automation
Reviews it went through
Privacy review Labor or workforce consultation
What do these mean?
Power user
Someone comfortable with spreadsheets, forms and low-code tools.
Reference only
Documentation and lessons — not something to deploy.
Human review built in
A person checks the AI's output before it is used.
Personal information (PII)
Names, addresses, IDs or other personal identifiers.
Internal, non-public data
Non-public operational data without personal identifiers.
Internal staff
Used only inside the organization.
Not open source — description only
The write-up is what is shared, not the artifact itself.
No new spend
Built with licences, staff and infrastructure the organization already had.
Partially — with rework
Some pieces are vendor-specific and would need swapping.
Administrative & task automation
Forms, intake, scheduling, records — the routine work that eats staff time.
Labor or workforce consultation
Discussed with the union or the affected staff before rollout.

Problem

Onboarding a new employee involved somewhere between 15 and 30 discrete tasks depending on the role, spread across HR, IT, payroll, facilities and the hiring department, and described in a 40-page handbook that nobody read on their first week.

The visible symptom was help desk volume: an average of 11 tickets per new hire in the first month, most of them variations on “I don’t have access to the thing I was told to use.”

What we built

A chat assistant in our low-code platform. On a hire’s start date it generated a checklist from their role, start date and work location, grouped into before day one, week one and week two. The hire and their supervisor saw the same list.

For the routine access requests — the ones that are determined entirely by the role — it opened the request automatically. Everything else it explained and pointed at.

A person approved every access request. The assistant drafted and tracked; it never granted anything.

Results

Help desk tickets per new hire fell from 11 to 3 within four months and stayed there for the eighteen months the tool ran. Checklist completion at 30 days went from something we could not measure at all to 94%.

Supervisors used it more than we expected and hires used it less. The most common feedback from hires was that they wanted the list in email, not in a chat window, which we should have taken more seriously than we did.

Why it was retired

Our HR system’s vendor shipped a native onboarding checklist feature in a 2026 release, covering most of what this did, integrated with the record we were querying through a connector anyway. Continuing to run a parallel tool would have meant two task libraries drifting apart, and we had already watched that happen once with the handbook.

We decommissioned it in July 2026 and moved the task library into the vendor’s feature. The entry stays here because the retrospective is the useful part: what the tool did, what the replacement does not do as well, and what we would keep.

Lessons learned

The task library was the asset and the assistant was the interface. Eighteen months later the library moved to a completely different system in an afternoon, which is the clearest possible statement of where the value was.

Build the thing that gets replaced. We would build it again knowing it had a two-year life, because the alternative was two more years of eleven tickets per hire while waiting for a vendor roadmap.

Ask people how they want to be reached before choosing the interface. A chat assistant was the platform’s default shape, not a decision, and email would probably have been used more.

What the replacement lost: our version explained why a task existed, in a sentence written by whoever owned it. The vendor’s checklist is a list of task names. We are trying to get the explanations back in as task descriptions.

How to reuse

There is nothing to install. Take the task library structure — a role, a task, an owner, a due offset from start date, and one sentence of why — and build it before you build anything else. If your HR system now offers a checklist feature, use it; that is the honest recommendation from the people who built the alternative.

About

Area of work
  • HR & workforce
  • IT & operations
  • Staff & partner coordination
Review status
Deprecated

How it's built

How AI is involved
AI is part of the solution
Types of AI
  • Chat assistant
  • Agents & automation
AI tools & models
  • Power Virtual Agents
  • Power Automate
  • SharePoint
Where it runs
  • Low-code platform

Sharing & licensing

Access terms
The tool has been decommissioned and there is nothing to install. The task library and the retrospective are linked above and are the reason this entry is still here.
Portability notes
The task library is a spreadsheet and moves anywhere. The assistant itself was built in the Power Platform against our own HR and identity systems, and by the time it was retired the platform-specific parts were the majority of it.

What it took

Cost to keep running
No ongoing cost
How it was bought
  • Existing enterprise licence
Who it affects
It reached every new hire, which was the argument for building it: the informal version of onboarding — a manager who remembers to explain things — was reliably better for people who already knew someone in the organization. We checked completion rates by department and by whether the hire was internal or external, and the gap between them narrowed. Nothing about a person's performance was inferred or recorded.

Data & access

No PII/PHI in the shared material
Yes
Data sources
  • HR position and start date records
  • Role-based task library
  • IT access request catalog
Data-governance caveats
The assistant read a new hire's role, start date and work location, and nothing else from the HR record. Conversation logs were kept 90 days and were never available to a hire's supervisor. All of it was deleted at decommissioning under a documented plan, which is in the retrospective.

Printed from AI Use Case Catalog
https://crypticpy.github.io/phct/examples/ai-use-cases/catalog/hr-onboarding-assistant/