PodcastJune 2026
His Customers Come From ChatGPT
Hosted by Honest Wealth Builders (Abi Asija)
Not an interview so much as a live business teardown. Abi Asija pushes on segment focus, pricing, churn, and whether to bet the company on enterprise or mid-market, and Deepak answers with specifics. The useful half for practitioners is the first 35 minutes: what breaks in a legacy content operation when answer engines arrive, and how citation decisions actually get made.
Watch or listen
Also available on Apple Podcasts, Spotify, which have no episode-level player to embed:
Key takeaways
- The trigger event for a GEO budget is a traffic collapse. Customers arrive after organic blog traffic drops 40 to 50 percent to answer engines.
- Gated content is dead weight. Email walls and PDF-only assets are invisible to answer engines, and the question has to be answered in the first couple of paragraphs rather than buried below the fold.
- Authorship is machine-checkable. Engines map a byline to LinkedIn and other profiles to decide whether the author has real domain authority, and only then decide to cite. Content ghost-written by a marketer with no footprint does not clear that bar.
- There is no fast path. The quickest measurable movement he has seen is two to three months, and he tells customers that up front rather than selling a 30-day turnaround.
- LLM queries carry buying intent in a way Google queries often don't. A Google search may be pure research; someone asking an engine to recommend a solution is usually mid-purchase, which is why the conversion rate is higher.
- He recruits engineers into marketing and sales roles, not for the coding but for the analytical habit of reading the data behind a funnel and explaining a technical product accurately.
- Outcome-based pricing for AI visibility is tempting and he has brainstormed it, but guaranteeing a visibility score hands your revenue to a third party's algorithm. Both he and the host land on it being too fragile to build a company around.
- The numbers, stated on the record: a little over $1M ARR in the trailing twelve months, against a three-year target of $10M.
- Sixteen people, roughly 20 percent in the US and Canada and 80 percent in APAC, with engineers hired into every department including marketing and sales.
- Around 300 customers, split roughly 16 enterprise and 270 mid-market.
- Pricing is per prompt monitored rather than per seat. Self-serve runs $100 a month for 100 prompts and $500 a month for 500. Mid-market averages $3,000 to $4,000 a year; enterprise runs $35,000 to $40,000.
- 70 to 80 percent of leads come from AI search. The remainder comes from LinkedIn and X ads, plus early experiments on OpenAI's ad platform that have produced impressions but no attributable conversion yet.
- No enterprise customer has churned. The mid-market losses came from companies shutting down or being acquired, not from switching away.
- Enterprise deals average about two months to close, 30 days at the fastest, which is quick for anything sold into security. Getting to $10M ARR implies an average contract value above $100K, which is why he picks enterprise when the host forces a single lane.
In his words
“Their traffic has gone down 40 to 50 percent. That is the reason they come to us.”
“They have gated content. In the legacy model you put in your name and email and then you unlock the content. That is going away.”
“These AI engines are very smart. They know exactly why you are writing the content, who is writing the content, and how often you are doing it.”
“When people are searching in AI engines, they are actually looking to buy the solution.”
“In the last 12 months we did a little over a million ARR. In the next three years we are hoping to achieve at least 10 million.”
“The majority of our customers we get from AI engines. 70 to 80 percent.”
Quotes and chapter marks are transcribed from the episode audio and lightly edited for clarity. Nothing is paraphrased.
Jump to a section
Timestamps open the YouTube cut at that point.
- 0:47The real constraint: hiring
- 1:35The niche, and why it stays narrow
- 3:58Why every department hires engineers
- 7:10AEO and GEO, defined
- 8:46What changed since 2015-era SEO
- 11:12The trigger: traffic down 40 to 50 percent
- 12:02Ungate the PDFs, answer in the first fold
- 12:53How engines verify who wrote it
- 15:19How the product actually works
- 20:54How long results take
- 22:26Pricing, segments, and customer count
- 24:03Where the leads actually come from
- 31:19Why LLM intent converts better
- 57:48Outcome-based pricing, and why it's risky