PMAI PM Playbook

Customer Support Copilot

Medium risk33 min full case
The verdictPilot candidate

Pilot after blockers resolved

What would unlock the pilot
  1. 01Build 100-example golden eval set with labeled reference responses
  2. 02Implement accept/edit/reject event logging and build basic dashboard
  3. 03Implement low-confidence fallback behavior
  4. 04Confirm disclosure, retention, and tenant data handling requirements

The case in four facts

The problem

Agents spend 38% of handle time drafting replies from scattered knowledge base articles. Top 50 intents cover 62% of ticket volume.

The AI's job

Draft support responses using approved knowledge base articles for Tier 1 agents to review and send, subject to human approval before any customer-facing message.

Without AI

Agents search knowledge base manually, read articles, and write responses from scratch. Templates were tried and failed (adoption dropped from 22% to 11%).

Expected outcome

Reduce drafting time from 3.1 minutes to under 1.5 minutes for supported intents. 75% draft accept rate. Zero hallucinated policy claims.

OwnerMaya Patel, PMStageprototypeUserTier 1 support agents

Readiness at a glance

2.86/ 5
Pilot candidate

The problem, workflow, and AI job are strong enough to justify a pilot, but the team should not start that pilot until eval, observability, regulatory review, and fallback blockers close.

  • Problem fit
    4/5
  • Workflow fit
    4/5
  • AI job definition
    4/5
  • Data readiness
    3/5
  • Eval readiness
    2/5
  • System behavior
    3/5
  • Risk & safety
    3/5
  • Regulatory readiness
    2/5
  • Cost & business case
    3/5
  • Observability
    2/5
  • Launch & ops
    2/5

Where the score lands

4 dimensions to fix
  • Eval readiness2/5

    No golden dataset built yet. 5 examples are illustrative, not a scored eval set.

    NextBuild 100-example golden eval set with labeled reference responses. Run inter-rater reliability test.

  • Regulatory readiness2/5

    Prompt and output retention policy not confirmed.

    NextConfirm disclosure, retention, and tenant data handling requirements before pilot.

  • Observability2/5

    Accept/edit/reject tracking not implemented

    NextImplement accept/edit/reject event logging. Build dashboard with 8 key metrics. Configure alerts.

  • Launch & ops2/5

    No written pilot plan with success criteria or rollback triggers

    NextWrite pilot plan with success criteria and rollback triggers. Create agent training session. Assign review owner.

And before production

7 more gates after the pilot

  1. 01Pilot runs 2+ weeks with online metrics meeting targets
  2. 02No confirmed hallucination reports during pilot
  3. 03Agent feedback is net positive
  4. 04Formal risk register with incident response process
  5. 05250-example eval set passes at pilot thresholds
  6. 06Load testing at 2x expected concurrent usage
  7. 07Safety triggers tested against adversarial inputs
Alternative path

What the team could do instead

Invest in knowledge base quality and search UX. Better KB search with snippet previews may capture 40-60% of the drafting time savings without model risk. Evaluate whether this closes enough of the gap before committing to an LLM-based copilot.

A medium-risk AI drafting workflow for Tier 1 support agents. The team has a working demo, leadership wants to ship next quarter, and the right answer is "pilot after blockers" — not "launch."

The situation

Maya Patel is the PM for the Service Cloud at a 600-person B2B SaaS company. Tier 1 support handles 2,400 tickets a week, and a four-week time study showed agents spending 38% of their handle time drafting replies from a 400-article knowledge base. The top 50 intents cover 62% of volume. Templates were tried two years ago and adoption collapsed from 22% to 11% in three months because the templates were too generic to use as-is.

Eight weeks ago an engineer prototyped a draft generator on Claude Sonnet wired to the existing KB retrieval pipeline. The demo at the last leadership review was tight: a manager clicked an example ticket and watched a clean, on-brand reply appear in two seconds. The CRO asked if it could ship by end of quarter. Engineering said yes. The CFO asked what it costs. Maya said she'd come back with a real answer.

This case study is what Maya brought back.

What's working

  • Problem fit is real and well-measured. The time study is from twelve agents over a month, not a hand-wave. Top 10 intents alone cover 31% of volume — even a copilot that helps on those is meaningful.
  • The AI job statement is tight. "Draft support responses using approved knowledge base articles for Tier 1 agents to review and send, subject to human approval before any customer-facing message." Scope is limited to the top 10 intents in v1. The autonomy level is "suggest" — humans send everything.
  • The workflow is sensible. AI drafts, agent accepts/edits/rejects, agent sends. Integrates into the existing reply panel. Agents stay the decision-maker, which keeps blast radius bounded.
  • The cost model is in the right neighborhood. $0.0094 per draft baseline, ~$51/month for the full team at the v1 intent scope. Not free, but easily justified by even a 10% handle-time reduction on supported intents.

What's not yet ready

The PRD looks complete on paper. The operational scaffolding doesn't.

  • There is no eval set. The plan defines a rubric and lists five golden examples — but five examples are an illustration, not a regression suite. Without a 100+ example labeled set, "the draft looks good" is the only quality signal the team has. That's how you ship a copilot that hallucinates the wrong cancellation policy on a Tuesday and finds out from a Reddit thread on Friday.
  • Observability is barely there. Latency and API cost are logged. Accept / edit / reject events are not. No dashboard, no alerts. If draft quality silently drifts after launch, no one notices until the support manager mentions it in a 1:1 three weeks in.
  • Low-confidence fallback isn't implemented. The prototype shows a draft regardless of model confidence. A confidence threshold and a "no draft this time" fallback are specified in the PRD but not built. This is the single highest-blast-radius gap: the moments where the model should stay quiet are exactly the moments it currently speaks loudest.
  • Legal sign-off on the AI disclosure language hasn't started. Customer-facing messages need the right footer. Data retention and tenant handling for prompts and outputs aren't confirmed.

The verdict, restated

Pilot, after blockers close. Not now. The case is genuinely strong on the dimensions that take quarters to build (problem, workflow, AI job). It's weak on the dimensions that take weeks to close (eval set, dashboard, fallback, legal sign-off). A two-week sprint to close the four pilot blockers is the right unlock for an 8-agent, 2-week pilot on the day shift.

Saying "yes" today would borrow trust the team hasn't earned yet. Saying "no forever" wastes a strong problem fit. The PM job here is to say "yes, with these four things first," and to make those four things small enough to be obviously worth doing.

What would change the verdict

The team starts the pilot the moment all four of these are true:

  1. A 100-example golden eval set exists, with labeled reference responses and a 0.7+ pass rate at the proposed launch threshold.
  2. Accept / edit / reject events log to a dashboard, with an alert on a 7-day rolling accept-rate drop below 60%.
  3. The low-confidence fallback ships and is tested against the 100-example set.
  4. Legal confirms the disclosure language and the prompt/output retention policy.

If those four are not done by the pilot start date, the pilot moves. The decision criteria don't.

Link copied