Launching a new go-to-market test at a voice AI software company took 45 to 60 days. After a six-month engagement, it took about 3 hours.
Blake Williams of Growthstory AI joined the company in April 2026 to lead ecosystems and later served as chief revenue officer. He is not an engineer. Using AI as the build team, he replaced a launch process that ran through briefs, vendor queues and engineering backlogs with a system one person can run in an afternoon.
The system takes an idea to a live landing page with a qualifying quiz that ends in a booked meeting or a self-checkout sale, publishes ad creative to Meta and Google Ads, tracks every lead into HubSpot, and uses AI to judge performance against revenue. The same approach rebuilt the partner program, its attribution, and its agency recruiting.
The problem: learning was slow and expensive
The go-to-market team could not test ideas fast enough, and spend kept running on ideas nobody had validated. Ad spend alone ran about $50,000 a month, or $150,000 a quarter, before agency and vendor fees.
Every test moved through a chain of handoffs:
- An idea became a brief.
- The brief went to an agency or an engineering queue.
- Creative arrived on the vendor’s schedule, in fixed batches.
- A landing page waited on web resources.
- The campaign launched, often without attribution clean enough to read the result.
At 45 to 60 days per launch, the team got two tests a quarter at most. Each one carried $75,000 to $100,000 of ad spend, plus agency fees, before anyone knew whether it worked.
The paid social account showed the pattern. It had been run as a precision lead-generation program, not as a way to learn what converts. The spend produced leads but very little information about which offers and audiences were worth scaling.
The partner channel had the same problem in a different form. Becoming a partner required a sales process, nobody could prove what partners produced, and agencies had no clear reason to sign.
The system: idea to live campaign in 3 hours
The build removed the handoffs. The person with the idea runs every step, with AI doing the production work.
- 01
Offer
The idea becomes a priced offer, created in Chargebee so it can be bought the day it launches, with copy and a page structure.
- 02
Funnel
A landing page is built and published with a quiz that qualifies each visitor. The quiz ends one of two ways: a booked meeting, or self-checkout through Chargebee that produces revenue on the spot.
- 03
Creative
Video and static ads are generated to spec in the formats each platform needs. No vendor batch, no wait.
- 04
Publish
Campaigns, ad sets and ads are pushed directly to Meta and Google Ads.
- 05
Attribution
Every lead carries its source into HubSpot, so a result ties back to the ad that produced it.
- 06
Analysis
AI reads ad performance against the CRM data and judges it on return, not on clicks or cost per lead.
Step six feeds step one. The result of each test shapes the next idea, so the system runs as a loop.
The system is live on both sides of the business. On the partner side it recruits, qualifies and signs agencies. On the direct side it sells to end customers. Both run on the same funnel, checkout, attribution and analysis.
Three hours gets a test into market. The ads still need days of delivery before the results mean anything. The gain is that a full test cycle now fits inside a week where it used to take a quarter.
Three frictions removed in the partner channel
The partner program was rebuilt from scratch and shipped by its June 16 deadline. Each part of the build answered one specific friction.
Becoming a partner required a sales process
Most partner programs open with a long application written for the vendor, followed by calls, a contract chase and manual setup.
The fix. A self-serve partner funnel. A partner applies, signs the contract by e-signature, checks out, and gets a referral link. Every step writes to HubSpot, and no person is in the loop. The application asks how the partner works — which AI tools they use, what they connect to, how they go to market — so enablement is shaped before anyone talks to them. One contract covers all three ways to get paid, so nothing is renegotiated when a partner changes how they sell.
Result. An agency signed and purchased through the flow on its own, and had a referral link live right after.
Nobody could prove what partners produced
Without attribution, a partner budget is a line item nobody can defend.
The fix. Dedicated partner pipelines in HubSpot, an attribution model that carries source from first touch through signed contract and payment, and a scorecard with nine measurables. For media and ad-platform partners the model went one step further: send booked jobs back to the partner as conversion data, so the partnership is measured on revenue rather than leads.
Result. A partner budget of about $50,000 a month was scored and defended with data.
Agencies had no reason to add a vendor
Recruiting on product features asks an agency to care about the vendor’s roadmap. Agencies care about their own economics.
The fix. Three angles did the work. Give to get — hand partners the leads the company’s own ad spend produces that are not ready to buy. Keep your stack, swap the voice layer — agencies keep the platform they already install and trade build work and compliance exposure for margin. Fast plays first — agency and reseller deals close quickly, platform alliances are slow, so effort went where the cycle was short.
Result. Seven agency partners recruited and activated, plus a distribution partnership reaching about 2,000 dealers.
Results
Launch time fell from 45 to 60 days to about 3 hours, a reduction of more than 350×. The other results follow from that speed.
- Launch time
- 45 to 60 days to about 3 hours, idea to live campaign
- Partner program
- Rebuilt from scratch and shipped on its June 16 deadline
- Self-serve partner signup
- First agencies signed and purchased with no sales involvement
- Agency partners
- Seven recruited and activated
- Distribution
- One channel partnership reaching about 2,000 dealers
- Attribution
- Source tracked from first touch to signed contract and payment
- Partner budget
- About $50,000 a month defended with data
- Creative
- Video and static ads produced on demand, replacing vendor batches
- Coverage
- Live on both the partner side and the direct side
How AI was used
AI did the production work of four teams. One operator supplied the judgment.
The diagnosis came first. Before the start date, AI had produced a full CRM audit with a 102-item gap analysis, teardowns of two competitors, and a shortlist of 55 industry leaders drawn from 13,000 connections.
Three practices made the output reliable:
- Written specs. Each repeatable job has a spec the AI follows, so the tenth output matches the first.
- Checks on the real output. Video is checked by transcribing the rendered audio, not by trusting the script.
- A revenue standard. Paid performance is judged against a fixed return target, not against the account’s own past results.
The operator’s job changed from doing the work or waiting on it to deciding what to build, what to stop, and what counts as good enough.