Which AI Agents to Build by ARR Stage — A GTM Fleet Map | Bowtie Funnel
White Paper · GTM AI Orchestration

Which AI Agents Should You Actually Build?
A Fleet Map by ARR Stage.

Anyone can list fifty agents. The fleet that moves revenue is decided by two things — where you sit on the ARR curve and how your revenue splits between new logo and expansion. This paper maps both, and names what not to build yet.

Book the GTM AI Design Sprint → Read Dare to Orchestrate
Abstract

Most go-to-market AI conversations begin with a list of agents. That is the wrong starting point. The right fleet — the specific agents worth building — is a function of a company's ARR stage and its revenue mix: the split between new-logo acquisition and expansion of the installed base. Early-stage companies run almost entirely on new logo, so their agent investment belongs on the left wing of the bowtie funnel. As expansion overtakes acquisition past roughly $100M ARR, the investment flips to retention and expansion. Below is a stage-by-stage fleet map with an interactive selector, so you can locate your position and see which agents to build first — and which to defer.

Why this matters: Most vendors sell you every agent. The fleet you need is set by your stage and revenue mix — and knowing what not to build yet is as valuable as knowing what to build. That's a diagnosis, not a product pitch.
Key Takeaways
  • The number of agents you can name is irrelevant. Stage and revenue mix decide the fleet.
  • Below $20M ARR, ~90%+ of revenue is new logo — build the left-wing acquisition fleet only.
  • From $20M–$100M, expansion turns material; onboarding and retention come online and coordination starts to matter more than agent count.
  • Above $100M ARR, expansion leads (67%+) — retention and expansion fleets become the priority and the work shifts to governance at scale.
  • Knowing what not to build yet is as valuable as knowing what to build.
01

Agent count is a vanity metric

Ask a vendor how many AI agents they can deploy and you'll get a number in the dozens. Ask which of those a company at your stage should actually build, and the number collapses — often to three or four. Agent count is a supply-side figure. It describes what a vendor can ship, not what your revenue needs.

The useful question is narrower: given where our revenue comes from today, which agents compound — and which would we be staffing against volume that isn't there yet? A retention agent built on an installed base of forty accounts optimizes nothing. An acquisition fleet added at $150M ARR ignores the two-thirds of revenue now coming from existing customers. The map that follows answers the narrow question directly.

Build agents where the revenue is — not where the funnel diagram looks symmetrical.
02

The lens: revenue mix on the bowtie funnel

The bowtie funnel splits the customer lifecycle into two wings. The left wing is acquisition — Awareness, Education, Selection, Mutual Commitment — the motion that wins a new logo. The right wing is the post-sale lifecycle — Onboarding, Retention, Expansion — the motion that grows an account you already own.

Every company runs both wings, but never in equal proportion. A sub-$5M company books almost all revenue as new logo. A billion-dollar enterprise books most of it as expansion of a base built over a decade. Your revenue mix — the split between new-logo and expansion revenue — tells you which wing to instrument first, and therefore which fleet to build.

03

Find your stage

Select your ARR stage to see its revenue mix, the fleet to build first, and what to hold off on.

Select your ARR stage
Sub-$5M ARR
95% New Logo
5% expansion
Almost all your revenue comes from winning new customers. You don't have the installed base to make retention agents worth building yet — and a full implementation is likely too expensive to justify.
Build first — your money is here
  • Awareness fleet — content, SEO, social, demand signals
  • Education fleet — lead qualification, chatbot, research
  • Start with one Wedge stage, not the whole funnel
Don't build yet
  • Retention & Expansion fleets — installed base too small
  • Full lifecycle coverage — staffing stages with no volume
Where to start
Skip the implementation. The honest entry is the book, Dare to Orchestrate and a focused GTM signal sprint — prove the motion at one stage before spending on a full build.
$5M – $20M ARR
90% New Logo
10% expansion
Still a new-logo business, but the first signs of expansion are showing. This is where an agent build starts to make sense — focused on the left wing, with the first onboarding agents coming online.
Build first — your money is here
  • Awareness fleet — content, competitor intel, SEO
  • Education fleet — qualification, enrichment, FAQ
  • Selection fleet — outreach, meeting prep, AI sales
Light or not yet
  • Onboarding — first agent only, mostly manual still
  • Retention & Expansion — not in scope yet
Where to start
The Agent Build entry point — deploy the core left-wing fleet at the single stage leaking the most, prove the return, then expand.
$20M – $50M ARR
70% New Logo
30%
expansion rising
Expansion is now nearly a third of revenue — but new logo still dominates, so the left wing is still where the agent investment goes. Even with ten agents deployed, most belong on the acquisition side.
Primary investment
  • Awareness — ABM orchestration, intent signals
  • Education — qualification, call coaching, personalization
  • Selection — deal risk, pricing, proposals
  • Mutual Commit + Onboarding — now active
Light, not primary
  • Retention — health score only, early
  • Expansion — not the priority yet
Where to start
A multi-stage left-wing build with the first handoffs — this is where the Command Tower begins to coordinate, not just isolated agents.
$50M – $100M ARR
60% New Logo
40%
balanced
Revenue is becoming balanced. The left wing still leads, but expansion is now a real investment — retention and expansion fleets are no longer optional. This is where full lifecycle coordination starts to matter.
Both wings active
  • Full left wing — awareness through selection
  • Onboarding — intake, activation, stall detection
  • Retention — health score, churn risk, engagement
  • Expansion — upsell, opportunity detection
The shift to watch
  • Command Tower coordination becomes the real value, not agent count
  • T1 orchestrator should be coming online now
Where to start
The Command Tower build — balanced fleets across both wings, coordinated by an orchestrator, with governance maturing as agents earn autonomy.
$100M – $200M ARR
33%
67% Expansion
expansion-led
Two-thirds of revenue now comes from existing customers. The investment flips: the right wing — retention and expansion — is the priority, and the left wing is optimized rather than the gap. This is a full Command Tower with enterprise governance.
Primary investment — right wing
  • Retention — health, churn prediction, renewal, win-back
  • Expansion — upsell, cross-sell, multi-BU, advocacy
  • Mutual Commit + Onboarding — enterprise-grade
Optimized, not the gap
  • Left wing — running well; agents here are optimization, not foundation
  • The work is governance at scale, not new acquisition agents
Where to start
A full Command Tower with enterprise governance — right-wing fleets as the priority, a steering committee, and the orchestrator coordinating the whole lifecycle.
Strategic Advisory · Retainer
$200M – $1B ARR
25%
75% Expansion
expansion-dominant
Three-quarters of revenue is expansion. You already have a working acquisition machine, internal platform teams, and likely transformation partners in motion. What you need isn't a builder — it's the specialist who owns the GTM agent architecture and governance specifically.
Where the value is
  • Right-wing at scale — retention, churn prediction, renewal, multi-BU expansion, advocacy
  • Governance architecture — the orchestration and autonomy model across a large fleet
  • The GTM Orchestrator function — standing up the role and operating model
The engagement shifts
  • Not a one-time install — an ongoing advisory relationship
  • Bowtie Funnel as the GTM specialist alongside your broader program, not the sole implementer
The model at this stage: a strategic advisory retainer, not a project. You own the GTM Orchestration layer — architecture, governance, and the orchestrator role — while internal teams and broader partners run the surrounding transformation.
Specialist Partner
$1B+ ARR
15%
85% Expansion
expansion-led enterprise
At this scale you almost certainly have a Chief AI Officer, an internal AI function, and a primary transformation partner already in place. The honest answer: Bowtie Funnel isn't your primary partner here — and saying so is the point.
The fit, if there is one
  • GTM Orchestration specialist within a larger program — a specific deliverable, not the whole engagement
  • Designing or auditing the GTM agent governance layer alongside your prime
  • The independent expert view on orchestration architecture
Who leads at this scale
  • Your internal AI function and Chief AI Officer
  • A primary transformation partner (the large consultancies)
  • Bowtie Funnel contributes the GTM-specific layer, not the program
Why we tell you this plainly: the same honesty that tells a $20M company what not to build yet tells a $1B+ company we're a specialist contributor, not the prime. Knowing exactly where we fit — and where we don't — is the diagnosis working as intended.
04

What the map means for your roadmap

The map yields one sequencing rule: instrument the wing that carries the majority of your revenue, at the single stage that leaks the most, and prove the return before widening. Everything else is premature. A sub-$20M company building a retention fleet is optimizing a base too small to matter. A $150M company still pouring agents into acquisition is ignoring the two-thirds of revenue now coming from its customers.

Agent count never appears in that rule. Coordination does — as the mix balances, the value moves from individual agents to the orchestration layer that hands work between them. That's the Command Tower, and it's why the later stages read as governance problems, not build problems.

05

Frequently asked

How many AI agents does a GTM team actually need?

Far fewer than vendors imply. The right number is set by your ARR stage and revenue mix — typically three to four focused agents at the single stage leaking the most revenue, expanded only after the return is proven.

What is an AI agent fleet?

A coordinated set of AI agents mapped to stages of the customer lifecycle — for example an Awareness fleet for demand generation and an Onboarding fleet for activation — rather than a collection of standalone, disconnected bots.

Which agents should a company under $20M ARR build first?

Only the left-wing acquisition fleet: Awareness agents (content, SEO, demand signals) and Education agents (lead qualification, research). Retention and expansion agents should be deferred until the installed base is large enough to justify them.

When do retention and expansion agents make sense?

Around $50M ARR, when expansion approaches 40% of revenue, retention and expansion fleets stop being optional. They become the primary investment above roughly $100M ARR, where expansion leads the revenue mix.

What is the bowtie funnel?

A go-to-market model that gives equal weight to acquisition (the left wing: Awareness, Education, Selection, Mutual Commitment) and post-sale growth (the right wing: Onboarding, Retention, Expansion), reflecting that in most SaaS businesses expansion eventually outproduces new-logo revenue.

What is a "Wedge" stage?

The single lifecycle stage you instrument first — the one point in your funnel leaking the most revenue. Instead of building agents across the whole funnel at once, you start with one wedge, prove the return, then widen. It's the smallest place to enter that still moves the number.

What is the Command Tower?

The orchestration layer that coordinates individual agents — handing work between them across the lifecycle — rather than running them as disconnected bots. As your revenue mix balances, the Command Tower (not the number of agents) becomes where the value sits.

What is a GTM Orchestrator, and what's a "T1" agent?

The GTM Orchestrator is the top-tier coordinating agent — and the human role that owns it — that directs the fleet and hands work between stages. "T1" (tier one) refers to that top orchestrator layer, which comes online once you're running balanced fleets across both wings and coordination matters more than agent count.

New-logo vs. expansion revenue — what's the difference?

New-logo revenue comes from winning brand-new customers (the left wing of the bowtie). Expansion revenue comes from growing accounts you already own — upsell, cross-sell, and renewals (the right wing). The split between the two is your revenue mix, and it decides which wing to instrument with agents first.

That's the diagnosis. The AI Design Sprint is the prescription.

This paper is the surface. The GTM AI Design Sprint maps your actual lifecycle, quantifies exactly where you're leaking revenue, and names the specific fleet to deploy first — with the return attached. You don't buy agents. You start by finding out which ones you need.

Book the GTM AI Design Sprint → Read Dare to Orchestrate
Resources
  1. Dare to Orchestrate — the Bowtie Funnel book on building and coordinating GTM agent fleets. bowtiefunnel.com/book-dare-to-orchestrate
  2. The GTM AI Design Sprint — a quantified, stage-specific fleet roadmap for your lifecycle. bowtiefunnel.com/ai-design-sprint-cross
Scroll to Top