Three production workflows, mapped across every toolset.

03 / Ticket Triage and Response

ServiceNow opens a ticket. An employee gets the email notification. The agent reads the ticket email, understands the question, and drafts or sends a reply to the student or internal staff. Human stays in the loop for approval until you trust it.

# ServiceNow webhook -> your endpoint -> agent
on ticket_created(payload):
    ticket = parse_email(payload.short_description, payload.description)

    # classify so the right specialist handles it
    route = classifier.run(ticket)        # student_q | internal | billing | escalate

    if route == "escalate":
        notify_staff(ticket); return

    draft = reply_agent.run(
        question = ticket.question,
        context  = kb_lookup(ticket.topic)   # grounded, no guessing
    )

    # human-in-the-loop until trust is earned
    if confidence(draft) >= 0.85:
        send_reply(ticket.requester, draft)
    else:
        queue_for_approval(staff, draft)

01 / Internship Research and Behavioral Nudges

A scheduled agent goes out, finds open internships, captures deadline windows and submission requirements, and stores the steps. Then it nudges students to act using behavioral science, not soft recommendations. The nudge creates movement, it does not just suggest.

The nudge ladder // behavioral science, not recommendations

# nightly cron -> batch research -> per-student nudge schedule
on schedule("0 2 * * *"):
    listings = batch_research(
        sources = internship_sources,
        extract = ["deadline", "requirements", "submission_steps"]
    )                                     # parallel subagents, async

    for student in roster:
        matches = match(student.profile, listings)
        for role in matches:
            schedule_nudge_ladder(            # T-14, T-7, T-3, T-1
                student, role,
                framing = behavioral_trigger(role.days_left)
            )

02 / Student Signals Dashboard

You have a large student dataset. Build a dashboard that watches it. When a student hits an action moment or trips a red flag, the system messages them to bring them in for corrective action. The dashboard is the cockpit, the agent is the outreach.

# nightly scoring -> dashboard refresh -> outreach on red
on schedule("0 3 * * *"):
    for student in warehouse.students:
        score = signal_engine(student)        # attendance, grades, inactivity
        flag  = bucket(score)                 # green | amber | red
        dashboard.upsert(student.id, score, flag)

        if flag == "red" and not recently_contacted(student):
            outreach_agent.run(
                student,
                goal = "book corrective-action meeting",
                tone = "supportive, specific, low-friction"
            )

Live triage simulation

Pick a scenario and watch the agent work a real student ticket end to end. Tier 1 resolves itself and logs it. Tier 2 drafts a reply and waits for a human yes. Tier 3 recognizes when it should step back and route the student to a person. Tap a card to run it.

Live demo · Internship agent

The student-facing side of Workflow 01. The agent maps employer deadlines and submission steps, matches them to the right student, and once the student opts in, runs world-class behavioral nudges. Action earns points and bookstore swag. Tap around, it is live.

Live demo · Student signals

The counselor side of Workflow 02. The agent watches the data nightly. When a student trips a pattern, like missing the same Friday 8am class two weeks running or a grade slide, it surfaces them and engages their assigned counselor. Run the scan.

Platform picker

One table to settle the build decision. Match the workflow to the platform that owns its pattern, then let the orchestrator route between them.

Rollout plan

The path from zero to live for 120 staff and 10,000 students. Build it once at the ServiceNow source, prove it in the dark, pilot a slice, then scale. Tick items off as you go and your progress is saved.

Meet the team

The two people you work with directly on this engagement. One builds the system, one keeps it human. Both answer their own email.

Tom Carter

Founder & CEO · Connected Consulting

Tom builds AI systems that take the repetitive work off a team's plate so the people can spend their time on people. Before Connected Consulting he spent his career in enterprise sales, ranked the number one rep out of more than 3,800 worldwide at LinkedIn and closing over $100 million across accounts like Google, Microsoft, and JP Morgan. Today he is recognized as an agentic workflow thought leader, building AI-native systems that run real operations end to end.

He is also a Hoosier. Tom came to Indiana University as a Big Ten track and field athlete, and when his family lost their home during his junior year he kept working, finished his degree, and later earned his MBA. That blend of enterprise rigor and ground-level grit shapes every build: start with the real workflow, automate what never should have been manual, and keep a person on the decisions that matter.

✉tom@connectedconsulting.io ☎317-213-2437 inlinkedin.com/in/tomcartermba

Amber Dyckman

Founding Partner & Placement Executive · Connected Consulting

Amber leads career and outcomes strategy as a placement executive. She came up inside the enterprise technology world and built the kind of network that opens doors, and she is fluent in translating raw talent into the language hiring managers and committees actually read.

On a student services engagement that lens is the whole point. The same system that triages tickets also surfaces internships, maps them to the right students, and nudges them to follow through. Amber designs that human layer, the part that turns an automated alert into a student who actually takes the next step. She keeps the work honest about one thing: the technology should serve the person on the other end, never replace them.

✉amber@connectedconsulting.io ☎704-740-2463 inlinkedin.com/in/amberdyckman