Enterprise buyers asking about AI workforce redesign always ask the same question eventually: "What will my sales team actually do differently?" This post answers it with a single role from a live engagement at a Singapore digital media platform — the before state, the AI surface introduced, the after state, the skills delta, and the measured outcome.
The before state: a prospector paying a human salary for machine work
The role we inherited was a cold-call prospector. The week-one task audit — a 90-minute interview plus two weeks of time-logging — produced the split: roughly 70% of hours on list research, manual enrichment, and sequence maintenance; about 15% on Anchor work, meaning actual consultative conversations with advertisers; the remainder on internal reporting.
Read that split as an economics problem: the company was paying a human salary, in one of the most expensive labour markets in Asia, to compile lists a machine compiles better. The consultative work — the only part of the role that closes and retains revenue — was starved to 15% of the week.
The AI surface: the Prospecting Assistant
The Prospecting Assistant is a CRM-and-enrichment pipeline with three functions. It compiles and enriches prospect lists from CRM data and public sources; it drafts first-pass outreach in the house voice; and it scores interest signals as replies arrive, routing the scored queue to the human. It needs CRM access and an email domain — nothing more — and every outbound message goes only after human confirmation.
Operationally, the prospector's Monday morning changed from list-building to queue review: a ranked list of interested advertisers with context, instead of a spreadsheet of cold names.
The redesigned role: Advertiser Solutions Consultant
The redesigned task split: 10% oversight of the automated pipeline — reviewing drafts, tuning scoring thresholds; 25% AI-assisted work — research-supported conversation preparation, AI-drafted proposals under human judgement; 65% Anchor — consultative selling, relationship strategy, renewal and expansion conversations.
The job description was rewritten around that split, and — critically — the targets changed with it. The role is no longer measured on dials made; it is measured on qualified conversations held and revenue retained. Behaviour follows measurement.
The skills delta: from list research to consultative selling
The delta was larger than expected. List research skills do not transfer to consultative selling; they are different disciplines. The training approach: a one-day workshop on consultative technique and AI-signal interpretation, followed by two weeks of shadowing senior commercial staff on live advertiser conversations, assessed against the skills matrix in the Job Redesign Workbook. Skills-assessment closure for the role's cohort is reported on the workforce analytics dashboard like every other role family.
The hardest part of the redesign was not the AI. It was giving someone who had spent years compiling lists the permission — and the training — to stop.
The measured outcome
The workforce analytics dashboard reports the movement of hours: hours reallocated from rote to Anchor work per role, monthly, against the week-one baseline. For this role the line moved from 15% Anchor to 65% Anchor across the first two quarters — and qualified conversations per week rose with it, because the pipeline produces scored interest continuously rather than in weekly list-building bursts.
Revenue-per-FTE for the role family is reported quarterly against baseline. It is the slowest of the three dashboard metrics and the only one the board asks about twice.
What this answers for an enterprise buyer
The question — "what will my sales team actually do differently?" — now has a concrete answer: they stop compiling lists, start working a scored queue, and spend two-thirds of the week in conversations that close and retain revenue. The AI surface is a pipeline, not a chatbot; the redesign is a job description, not a training day; and the proof is a dashboard line with a baseline under it.
If your sales team is spending more time on research than selling, email hello@qlabshq.com.
Week one: what the task audit actually found
The audit's surprises mattered as much as its headline split. List research was not one task but nine — sourcing, deduplication, enrichment, contact verification, territory mapping, sequence selection, and three flavours of manual CRM hygiene — and only two of them involved anything a prospector would call selling. The 70% figure understated the fragmentation: the role switched context roughly forty times a day, which is why time-logging, not interviews alone, was needed to see it.
The second finding shaped the redesign: the 15% Anchor time was not evenly distributed. A minority of conversations produced most closed revenue, and those conversations were the ones most often cancelled by list-research overrun. Protecting Anchor time was therefore not a fairness measure; it was a revenue measure.
Pipeline governance: where the human sits
The Prospecting Assistant drafts and scores; it does not send. Every outbound message passes a human confirmation point, and scoring-threshold changes are a governed amendment — proposed by the consultant, approved by the commercial lead, dated in the workbook. This is deliberate: an ungoverned pipeline optimises for volume within a quarter and for reputation over years, and the two are not the same objective.
Governance also answers the compliance question before it is asked: a lawful basis is recorded for every data source the pipeline touches, and personal data is minimised to what scoring requires. In a PDPA-first jurisdiction, that record is not overhead; it is the artefact that lets the pipeline exist.
Targets and compensation: the part everyone skips
The redesigned role would have failed without a targets rewrite. A consultant measured on dials made will make dials; the AI queue does not change that arithmetic. The redesign moved measurement to qualified conversations held, Anchor-hours share, and retained revenue — and compensation followed the measurement, not the other way round.
This is the least glamorous and most load-bearing part of any role redesign: the task mix moves when the measurement moves, and the measurement moves when someone edits the compensation spreadsheet. Everything upstream — the framework, the surfaces, the training — creates the possibility; the targets create the behaviour.
The first ninety days in the redesigned role
Week by week, the transition looked like this. Weeks one to four: the consultant worked the scored queue while shadowing the pipeline — learning to read interest scores, to override drafts, to escalate edge cases. The time-logging from the diagnostic continued, so the dashboard's first reads had a continuous series rather than a gap. Weeks five to eight: first cohort training — the one-day consultative workshop plus shadowing — ran alongside queue work; Anchor share crossed 40%.
Weeks nine to twelve: the pipeline's scoring thresholds were amended twice on the consultant's proposal, each dated in the workbook; the first renewal conversations run entirely on Anchor time closed. By the month-six re-audit the role's week looked nothing like its diagnostic: queue review on Monday morning, consultative conversations through the week, pipeline governance on Friday — the 10/25/65 split, observed rather than aspirational.
Frequently asked questions
Will the Prospecting Assistant replace the sales role entirely?
No — and that was never the design. The surface automates list research and first-pass drafting; the consultative work that closes revenue stays human by design. The role's Anchor share rose from 15% to 65%; the headcount did not change.
What CRM systems does the Prospecting Assistant work with?
The pipeline is CRM-agnostic at the integration layer: it reads and writes through the CRM's API and needs an email domain for outreach drafting. In this engagement it ran against the platform's existing CRM with no replacement project.
If your sales team is spending more time on research than selling, email hello@qlabshq.com.