NPRODUCT
← Home
Step 3 · Sharpen your edge · pillars 01–04

Decide on evidence. Ship in weeks. Prove what it changed.

For CTOs and product leaders: how a 6–8 week pilot turns every customer conversation into ranked, shipped and measured product, what AI automates, what your people own, and what I need from your systems.

What changes

Four pillars, before and after.

01
Customer Signal
From
A sample of interviews
To
Every conversation, refreshed weekly
Reference case16,000+ calls a year → ~700 recurring signals
02
Decision System
From
Loudest voice in the room
To
Ranked by revenue at risk and bookings potential
Reference case~700 signals → 150 high-impact features
03
AI-Native Delivery
From
Quarters from idea to production
To
Weeks, behind human gates
Reference case4–6 months → 2–4 weeks; 4× throughput
04
Measured Impact
From
Releases shipped, outcomes assumed
To
Every release tied to its evidence and measured
Reference caseNRR +5 points in one quarter, based on a cohort analysis

Reference case: a $120M+ ARR B2B SaaS company, same team and budget. How I measured →

How it works

Each Edge pillar becomes a working stage.

Each of the four Edge pillars becomes a working stage of an AI-native workflow, from every customer conversation to released, measured software. AI does the work between the gates; your people make every call that matters.

AI automates Your people own
01
Customer Signal
Capture
AI automatesPulls in sales, implementation, success and support calls, tickets and CRM notes every week.
Your people ownChoose the sources, grant data access, set privacy rules.
Synthesize
AI automatesExtracts every product request with the exact quote and source link; groups them into recurring signals.
Your people ownPMs validate the signals and correct misreads.
02
Decision System
Prioritize
AI automatesRanks signals by bookings potential, ARR at risk, churn correlation and frequency.
Your people ownPMs and leadership decide what deserves investment.
03
AI-Native Delivery
Define
AI automatesDrafts the PRD and acceptance criteria from the evidence.
Your people ownThe PM edits and approves; engineering reviews the spec.
Design
AI automatesGenerates wireframes and prototypes from the PRD.
Your people ownDesigners own the experience.
Build
AI automatesBuilds from spec and design; opens a pull request for review.
Your people ownEngineers review security and edge cases, test and merge.
04
Measured Impact
Release & measure
AI automatesDrafts release notes and enablement; links each release to its evidence; tracks bookings, churn and adoption.
Your people ownApprove the release; judge the outcome and reprioritize.
Pilot6–8 weeks

Prove it on one revenue use case

Connect your data, baseline cycle time and output, and run all four pillars end to end on one revenue use case.

Scale & transfer3–6 months

Make it how every team works

Extend across teams, formalize the human gates and new roles, train your people and hand over the keys.

The pilot, week by week

Eight weeks, one revenue use case.

One pod runs all four Edge pillars end to end while everyone else keeps delivering. The readout compares the pod against its own baseline.

Weeks 1–2
Connect and baseline

Connect your call, ticket and CRM sources. Baseline cycle time, output and the revenue metric the use case targets. Agree how we will measure success.

Weeks 3–4
Signal and decide

First ranked signal list for the use case. Your PMs validate it; your leadership picks the bets.

Weeks 5–6
Build behind the gates

Spec, design and pull requests drafted by AI on the chosen bets. Your PM, designer and engineers approve each step.

Weeks 7–8
Release and read out

Ship, link each release to its evidence, measure against the baseline, and set the plan to scale.

What already exists, what I configure

A proven workflow, configured to your stack.

It is a method plus working components: the signal pipeline, the revenue-ranking model, the spec-to-pull-request flow and the human-gate framework, built and proven at a $120M+ ARR company.

For each client I connect your sources, tune the ranking to your revenue model, adapt the gates to your engineering standards, and train your pod. It runs on Claude, Cursor and/or Hermes Agent.

What I need from you
Customer conversationsCall recordings (e.g. Gong), support tickets, CRM notes
Revenue contextCRM opportunities, ARR and churn data to rank signals
Product toolsRoadmap and ticketing (e.g. Jira); your design system
EngineeringCode repository and CI; your review and test process
PeopleOne pod for the pilot: a PM, a designer and two or three engineers, plus an engineering lead to own the gates
Security and control

Human gates

PMs approve priorities and specs. Designers approve the experience. Engineers approve every merge and release.

Scoped access

Read-only where possible, governed by your security policies and reviewed with your security team before the pilot.

Traceability

Every requirement links to the customer quotes behind it; every release links back to its evidence.

A sharper edge needs a team built to use it. Strengthen your spine →

Ready to sharpen your edge?

Start with a leadership diagnostic. We pick the revenue use case worth piloting and the baseline to beat.

Plan your pilot