Research · the evidence behind the work
LinkedIn ↗
The evidence room

Where does AI create useful work in a revenue org?

I wanted a decision base I could trust, so I built one. It covers the GTM roles, the RevOps functions underneath them, the tools they use and the controls an agent needs before it touches company data.

Research cards
221

Each card has a stable ID, dated sources and an honesty label for analysis or hypothesis.

● Counts checked · 12 Aug 2026
16
revenue, GTM and adjacent roles
6
core RevOps functions
30
tool integration guides
1,320
source records in the dataset
The sourcing rule

A claim carries its own receipt.

Inside the source library, each claim is tiered, dated and linked. Vendor numbers keep the vendor label. Anything I worked out myself is marked as analysis. The links below are a small public sample.

T1 / primary

Original evidence

Official documentation, filings, research papers, legislation and first-party datasets.

T2 / practitioner

Named experience

Credible operator accounts, technical writeups and direct interviews with clear provenance.

T3 / directional

Use with care

Consultancy synthesis, vendor benchmarks and secondary reporting that helps frame a question.

[ANALYSIS] = my conclusion from the evidence · [HYPOTHESIS] = a claim that still needs testing · [UNVERIFIED] = do not build on it yet
What exists

The database runs past RevOps.

Revenue decisions depend on the teams around them. The research follows those handoffs instead of stopping at the sales org chart.

16

Role packs

Week-in-the-life, KPIs, AI use cases, tooling and sources for the people doing the work.

sales · csm · solutions · marketing · fp&a · legal
6

RevOps function packs

The operating spine: planning, forecasting, systems, insights, deal desk and incentives.

data · process · planning · governance
37

Cross-cutting cards

Agent engineering, AI governance and payments evidence, kept separate from the role research.

14 governance · 8 payments · 15 agent engineering
98

Training modules

Practical AI use, change management, enterprise controls and paths for leaders and teams.

claude · openai · gemini · adoption · integrations
Current P0 shortlist

Six decisions worth funding first.

This is my analysis of impact, buildability and evidence quality, last checked on 12 August 2026. Two are build specs. Four are buy evaluations.

01

CRM hygiene agent

Start with a read-only scan. The scan earns the right to build the agent.

BUILD SPECpropose, review, then apply
02

Expansion signals

Account-grain product telemetry is the gate. Human owners run every play.

BUILD SPECread-only detection
03

Trust centre

Vanta when compliance automation matters; Conveyor for the narrower use case.

BUY PACKpilot before contract
04

Questionnaire auto-fill

SiftHub leads the current evaluation. Test it on a real SIG and CAIQ set.

BUY PACKfastest near-term win
05

AI forecast

Gong Forecast when Gong already sits in the stack. Fix the data layer first.

BUY PACKdata quality dependency
06

Enrichment

Clay Growth with named providers inside it, checked on real coverage and unit cost.

BUY PACKmeasure provider overlap
Publication line

Public work comes out of a private repository.

The research contains company-shaped questions and a private opportunity backlog. I publish the method, the patterns and the tools without publishing internal context.

That gives the work somewhere solid to come from. It also stops a directionally interesting vendor number turning into a confident LinkedIn fact by the time it reaches the feed.

What ships publicly
  • working browser tools
  • operator writeups on LinkedIn
  • practical guides and templates
  • sourced findings with caveats
Follow the research as it becomes work

I build the next useful thing in public.