Home Use Cases Work Insights About Contact
Back to Our Work

How an Enterprise HR Services Company Built a Support Agent That Answers Employees from the Employer's Handbook

Operations1
Open the walkthrough

Six-Step AI Workflow with Human Oversight

Each step shows the workspace, what the AI completed, and what still needs human review.

Steps 4 and 6 need a person. Step 4 is waiting for approval right now. Click it to open the full walkthrough.

Estimates are directional and based on stated assumptions. All names, organizations, and identifying details have been anonymized in accordance with our confidentiality agreements.

The Transformation

How an Enterprise HR Services Company Built a Support Agent That Answers Employees from the Employer's Handbook

Before
A person used to read every inbox each morning and log each question by hand.
A person used to read each question and pass it to the right team.
A person used to look up and type every answer, during office hours only.
A person used to log each call by hand and close the ticket at the end of the day.
After
every question that arrived, tied to a person before anything answers
every question sorted and understood before anything searches
the answer sent with the page it came from
the short queue only a specialist can settle
every conversation on the record, closed only when it is settled
Executive Summary

Answers every employee question the moment it arrives, on whichever channel it arrived, and never sends an answer it cannot point to a page for. Written for an HR services company where the question "how do I add my newborn?" comes in at 11pm on a Sunday and the person asking cannot wait for office hours. Every answer cites the employer's own handbook, every question the bot cannot settle reaches a specialist with the whole story already written down, and the questions the bot could not answer this week become the answers it gives next week. Six stations, run as a loop: what station 6 learns is what station 3 searches.

What Was Broken

A person used to read every inbox each morning and log each question by hand
A person used to read each question and pass it to the right team
A person used to look up and type every answer, during office hours only
The real cost
A person used to log each call by hand and close the ticket at the end of the day.

What We Built

Six stations and 13 subagents. Each subagent carries its own tasks and its own refusal.

A4
Channel Watch
Poll the chat widget, the mailbox and the SMS line, and stamp each message with where it came from
A3
Identity Match
Match the sender to an employee record by email, phone number or portal login
A3
Question Sorting
Classify each question as benefits, payroll, leave, onboarding or policy
A2
Context Building
Pull the sender's open tickets, past questions and current plan details alongside the question
A3
Knowledge Search
Search this employer's handbook, policy library and past settled tickets for the question
A2
Answer Drafting
Draft the reply from the retrieved passages and nothing else
A3
Resolution Watch
Ask whether the answer settled it, once, in the reader's own channel
A3
Handoff Routing
Route each stuck question to the team that owns that kind of question, on that account
A2
Case Brief
Summarize the whole exchange in three lines a specialist can decide from
A4
Ticket Writing
Record every conversation as a ticket: channel, category, the answer, and the pages it cited
A4
Satisfaction Watch
Send the one-question survey on the channel the person used, after the close
A1
Gap Learning
Identify the questions the knowledge base could not answer, out of the week's handoffs
A2
Service Report
Assemble the week's numbers: settled by the agents, handed to specialists, wait times, survey scores by category

How It Runs

1

Inquiry Arrival

Nobody is asked anything here

Agents watch every channel questions arrive on, merge the same person asking twice, and tie each question to an employee record before anything tries to answer it. Nothing waits on a person here. A question only reaches somebody when a later station cannot settle it.

2

Question Triage

Nobody is asked anything here

Agents sort each question into the kind of work it is, pull the record it will be answered against, and ask one short clarifying question when the message does not say enough. A person never sorts here. A question the agents cannot make sense of goes to station 4 with everything they tried.

3

Answer Delivery

Nobody is asked anything here

Agents search the employer's handbooks, policies and past settled tickets, write the answer with the page it came from, and stay with the question until it is settled or stuck. A person types no answer here. The questions the agents cannot settle move on with everything already tried attached.

4

Specialist Handoff

A person answers here

Agents turn each stuck question into a short brief that carries the transcript, what was already answered, and the two or three ways this kind of question usually settles. This is the station where a person works. Everything before it exists to keep this queue short.

5

Ticket Closing

Nobody is asked anything here

Agents write every conversation into the ticketing system with who answered what and from where, ask the one-question survey, and reopen any close the survey proves wrong. A person signs nothing here. The answer already happened at station 3 or station 4, and the record writes itself as it happens.

6

Service Review

A person answers here

Agents group the week's handoffs by what caused them, draft a new answer for every gap the specialists had to fill twice, and put the week's numbers in front of the person who signs it. A person signs the week. New answers do not go live without that signature, because next week the bot gives them to every employee who asks.

Where a Person Decides

Step 4, Specialist Handoff. Read three lines and pick an answer. Open the transcript only when the three lines are not enough. your answer is the only way through.
Step 6, Service Review. Settle the leave-of-absence trend, or name who will. Accept or reject the two proposed answers, then sign the week. new answers do not go live unsigned.

Operating Model

This changes how work flows through the team.

Role
Responsibility
Specialist Handoff owner
Read three lines and pick an answer. Open the transcript only when the three lines are not enough. your answer is the only way through.
Service Review owner
Settle the leave-of-absence trend, or name who will. Accept or reject the two proposed answers, then sign the week. new answers do not go live unsigned.

What Transfers, What Must Be True

What transfers
A person is in the loop wherever a stuck question needs an answer or a new answer is about to reach every employee who asks. Two of the six stations refuse to proceed on their own, Specialist Handoff and Service Review, and those are the two a person carries. Everywhere else the agents run at volume, around the clock, and reach you only when they cannot settle something.
Specialist Handoff stops for a person, and your answer is the only way through.
Service Review stops for a person, and new answers do not go live unsigned.
Every subagent says what it will not do. 13 of them do.
What must be true in your environment
The agent can read the systems your records already live in. This one reads 20.
Somebody owns Specialist Handoff and has time for it.
Somebody owns Service Review and has time for it.

Failure Modes

What breaks this pattern:

✗ Answering a name-only match

The agent matches an employee on name alone and pulls up the wrong record. It then answers a stranger's benefits question with someone else's plan details.

✗ Answering from general knowledge

The agent finds no passage from this employer and answers from what it knows about benefits in general. The answer sounds right and describes a plan the employee does not have.

✗ Routing to a general queue

The agent hands a stuck question to a shared queue with no owner. Nobody's name is on it, so nobody picks it up, and the employee asks again in a week, angrier.

✗ Closing over an open question

The agent closes a ticket whose last message is still a question. The employee thinks they were ignored, and the metrics count it as resolved.

Directional Outcomes

What the agent counts, and the station that counts it.

These counts are the tallies from one monitored run of the agents. They are not monthly or annual totals.

Questions logged and tied to a person
Counted at Inquiry Arrival
2,480
Tied to the wrong person
Counted at Inquiry Arrival
0
Could not tie, held
Counted at Inquiry Arrival
11
Questions sorted
Counted at Question Triage
2,412
Sent down the wrong path
Counted at Question Triage
0
Held as unclear, passed to a person
Counted at Question Triage
38
Answers sent with a source
Counted at Answer Delivery
2,374
Answers sent without a source
Counted at Answer Delivery
0
Questions passed to a person
Counted at Answer Delivery
388
Tickets written and closed
Counted at Ticket Closing
2,375
Closed over an open question
Counted at Ticket Closing
0
Reopened after a bad survey
Counted at Ticket Closing
37
Our measurement policy: We do not publish precise ROI without baseline methodology. Every figure above carries its basis.

What Runs Where

Every step names the subagent that does the work, the record it writes, the thing that raises a question for a person, and what it is allowed to touch. This is drawn from the source, not from a diagram somebody kept in sync by hand.

1Inquiry Arrival
subagentchannel-intake
writesqueue/open/<conv-id>.json
raiseshold-unmatched-sender
may touchqueue/**, rosters/** read-only
2Question Triage
subagenttriage-question
writesqueue/triaged/<conv-id>.json
raiseshold-unclear-question
may touchqueue/**, tickets/** read-only
3Answer Delivery
subagentanswer-question
writesqueue/answered/<conv-id>.json
raisesno-passage-found
may touchqueue/**, knowledge/** read-only
4Specialist Handoff
GATE
subagenthandoff-brief
writesqueue/handoff/<conv-id>.json
raisesawait-specialist-answer
may touchqueue/handoff/**, everything else read-only
5Ticket Closing
subagentclose-ticket
writestickets/closed/<conv-id>.json
raisesreopen-on-bad-survey
may touchtickets/**, queue/answered/** read-only
6Service Review
GATE
subagentreview-and-learn
writesknowledge/proposed/<answer-id>.json, reports/week-<n>.json
raisessign-weekly-review
may touchknowledge/**, reports/**, tickets/** read-only

Stack

Every system this agent reads or writes.

System
Read at
Stations
past settled tickets
Answer Delivery
1 of 6
the SMS line
Inquiry Arrival
1 of 6
the chat widget
Inquiry Arrival
1 of 6
the conversation archive
Ticket Closing
1 of 6
the employee portal
Inquiry Arrival
1 of 6
the employer account list
Question Triage
1 of 6
the employer handbooks
Answer Delivery
1 of 6
the employer rosters
Inquiry Arrival
1 of 6
the enrollment records
Question Triage
1 of 6
the knowledge base
Service Review
1 of 6
the plan documents
Answer Delivery
1 of 6
the policy library
Answer Delivery
1 of 6
the reporting dashboard
Service Review
1 of 6
the specialist queue in Teams
Specialist Handoff
1 of 6
the support mailbox
Inquiry Arrival
1 of 6
the survey tool
Ticket Closing
1 of 6
the team directory
Specialist Handoff
1 of 6
the ticket history
Question Triage
1 of 6
the ticketing system
Specialist Handoff
2 of 6
the weekly review report
Service Review
1 of 6
Next Step

Want to see if this pattern fits your customer support bots?

No build commitment·Real samples, not a demo·Estimate in writing