GojiberryAI Sold Its AI Sales Agent Before It Built One

The public figures come from different dates and measure different things, so they are best read as individual snapshots rather than one continuous account.

GojiberryAI did not begin with a fully autonomous agent. In a founder interview published by FounderBase, co-founder Romàn Czerny described an early version that was closer to a service than a software product. The team presented the idea to prospects, took orders, and fulfilled them manually.

That approach let the founders test two separate questions: Would businesses pay for better prospects, and could visible intent produce better results than another large list of cold contacts?

The interview describes a simple internal comparison. One group received outreach based on random prospects from Apollo, while the other used people showing buying signals. According to Czerny, the intent-led group converted four times better, although the interview does not provide enough methodological detail to treat the result as an independent benchmark.

For the founders, the result pointed to a narrower product thesis. The valuable step was not generating another personalized message. It was choosing whom to contact and when.

This distinction explains the product that followed. GojiberryAI now monitors buying and social signals, scores prospects against a customer's ideal profile, enriches contact information, and conducts outreach through email and social channels. In other words, the product focuses first on deciding who is worth contacting, then on what to say to them.

Y Combinator lists three co-founders, all of whom arrived with relevant operating experience.

That background changes the origin story. The product was inexpensive to test, but the team already had sales experience, technical ability, previous exits, and audiences it could reach.

GojiberryAI's product looks for evidence that a prospect is already interested. Its founders used the same principle to distribute it. They posted in communities where founders and sales teams discussed outbound problems, watched who engaged, and joined those conversations. The strategy developed in two stages: Reddit helped validate the pitch, while LinkedIn and X provided a larger stream of potential buying signals.

The First Customers Came From Reddit

Founder accounts say detailed Reddit posts produced the first ten customers. The approach was straightforward: explain the problem in public, find readers who recognized it, and offer the result before building a complete platform.

LinkedIn and X Turned Engagement Into Sales Leads

LinkedIn and X extended the same system. The founders shared product experiments, revenue milestones, and outbound playbooks. Replies, profile visits, and requests for more detail could all indicate intent.

The company also paid for reach. In one X post, Czerny said GojiberryAI had hired more than 50 LinkedIn creators over three months and reported an average 1.5x revenue return. He later wrote that user-generated campaigns had produced more than two million Instagram and TikTok views.

Both performance figures come from the founder, not an independent advertising report. Even with that limitation, the figures illustrate an acquisition strategy that extended beyond the company's own accounts. Founder content, creator distribution, retargeting, and direct follow-up all fed into the same B2B marketing funnel.

GojiberryAI's public revenue trail contains several snapshots from different dates and sources. They don't reconcile cleanly enough to construct a precise month-by-month growth trajectory, but together they show a business that scaled rapidly during its first year.

Nine Months: $112,000 in MRR

The company's Y Combinator launch said it reached $112,000 in monthly recurring revenue within nine months, equivalent to an annualized run rate of about $1.34 million. It also claimed 44% monthly growth and more than 1,000 paying customers.

A later YC company profile said GojiberryAI grew from zero to $2.5 million in annual recurring revenue in ten months, was growing 30% month over month, was profitable, and served more than 2,000 customers. YC hosts the claims, but they remain company-supplied figures rather than audited accounts.

The two YC figures do not reconcile cleanly. Annualizing $112,000 in MRR produces roughly $1.34 million, while the later profile reports $2.5 million ARR only about a month later. The available disclosures don't provide enough detail to determine whether the difference reflects unusually rapid growth, different measurement dates, or another change in how the company reported revenue.

The Final TrustMRR Snapshot: $424,368 in MRR

TrustMRR presents a later view. Its last connected Stripe snapshot showed $424,368 in monthly recurring revenue, $407,199 collected over the previous 30 days, $1,801,629 in cumulative revenue, and 3,944 active subscriptions.

Three different numbers from the same snapshot. Only the first is plotted in the chart above.

The data stopped updating in mid-August 2026, after the Stripe API key expired. The $424,368 MRR snapshot annualizes to about $5.1 million, well above YC's earlier $2.5 million ARR figure. The public disclosures don't provide enough information to reconstruct the exact path between those figures. MRR is a recurring run rate, cumulative revenue is money collected over time, and subscriptions are not necessarily unique customers.

Multiple public snapshots nevertheless show a business reaching hundreds of thousands of dollars in monthly recurring revenue within roughly a year. The exact trajectory and current figure remain unclear.

Czerny's account of joining Y Combinator shows that the accelerator arrived after traction, not before it.

1. First application: According to Czerny, the team was at about $20,000 MRR when it pitched an "AI LinkedIn outreach tool" and was rejected.
2. Second application: Czerny says the founders reapplied at about $1 million ARR after repositioning the product as a learning GTM system and joined YC's Spring 2026 batch.
3. During the batch: Czerny reported that the company grew 2.5 times in six weeks. He also said YC's standard $500,000 investment arrived after a delay related to the company's international structure.

The sequence matters because YC did not create the initial demand. Instead, it accelerated a growth system that had already taken the company from a spreadsheet to seven-figure ARR.

GojiberryAI packages parts of an outbound workflow that lean sales teams might otherwise assemble from several tools. Its $99 monthly plan includes two AI agents and outreach to as many as 1,800 prospects each month.

The trade-off is straightforward: less flexibility than a larger GTM stack, but fewer tools to connect before a team can begin outreach.

This is also where the company is moving beyond its original position. An "AI LinkedIn outreach tool" competes on messaging and automation. A self-learning GTM system makes a broader promise: learning which customer profiles, signals, and conversations are most likely to lead to conversion before deciding whom to contact.

That is a more ambitious business. It is also harder to prove. GojiberryAI says its agents track results, adjust targeting, and benchmark campaigns. Public sources do not reveal how much proprietary conversion data the company holds, how effectively learning transfers between customers, or whether performance improves over long periods.

GojiberryAI overlaps with competitors across four parts of the outbound sales stack:

Together, these competitors create two pressures: well-funded companies can reproduce parts of the workflow, while platforms such as LinkedIn and Gmail can constrain how outbound tools operate.

Well-Funded Competitors Can Rebuild the Workflow

Some rivals have far more capital. TechCrunch reported that 11x secured roughly $50 million in Series B funding after a $24 million Series A. Artisan announced a $25 million Series A in 2025. Incumbents such as Salesforce and HubSpot can add similar agent features to systems customers already use.

GojiberryAI's current advantages are more immediate: it bundles a broad workflow at an accessible price and has founders with a demonstrated ability to attract attention. Those strengths can produce fast acquisition. They do not automatically create high switching costs or proprietary technology.

LinkedIn and Email Create External Dependencies

A signal-led outbound platform also depends on channels it does not control. LinkedIn says it prohibits unauthorized bots, scraping tools, browser extensions, and software that automates activity on its website. That does not establish that GojiberryAI violates LinkedIn's rules. Instead, it illustrates a category-wide risk for products that depend on LinkedIn data and automated outreach.

The risk has already touched a close competitor. TechCrunch reported that LinkedIn temporarily restricted Artisan's company and employee accounts while reviewing concerns that included the use of data brokers accused of scraping the platform without permission. Artisan was later reinstated.

Email has its own constraints. Gmail requires sender authentication and imposes additional unsubscribe and spam-rate requirements on high-volume senders. For an AI sales product, compliance, account health, and deliverability are fundamental to whether the product works. Better copy has little value if the message never reaches an inbox.

The most memorable detail in GojiberryAI's story is not the $424,368 MRR snapshot. It is the spreadsheet.

The founders first performed the work manually because that was the fastest way to test what customers valued. They learned that intent-led prospects appeared to outperform random lists. Then they automated the surrounding workflow and used the same logic to find their own buyers.

That approach helps explain the early growth. It also defines the company's next test. The software components around modern outbound are becoming easier to reproduce. Language models can write messages. Data providers can enrich contacts. Sequencers can send follow-ups.

A more defensible moat would come from identifying which signals predict a sale and improving that judgment with every campaign. If GojiberryAI can turn the interactions generated across its customer base into better prospect selection and timing, it may become part of the infrastructure behind lean sales teams.

If it cannot, the company will remain an efficient bundle in a market full of other bundles. GojiberryAI sold the result before it built the agent. Its next challenge is proving that what the agent learns is harder to copy than the pitch that sold it.