AI That Thinks Like a Top 1% Marketer
Your smartest marketing assistant yet.
Imagine having an SEO expert, content strategist, and competitor analyst working for you 24/7.
Agent A combines AI with Ahrefs' trusted marketing data to help you find opportunities, improve rankings, and make better marketing decisions-faster than ever.
The assumption has always been that job seekers are a difficult market. Job boards, LinkedIn, and Indeed gave them free access for years, and that trained an expectation. People simply would not pay for a tool that helped them find work.
Erik Chavez does not think that is true. He is a Senior Solutions Architect at Microsoft, with fifteen years in cloud and platform engineering behind him. Two months ago, he launched Jobric - an AI job-matching platform built entirely for candidates, not employers. The product launched on May 1st, ran a free beta through March and April, and is already at $3,300 in monthly recurring revenue, paid entirely by job seekers out of their own pockets.
He builds it between 5am and 10pm, around a full-time job, with no outside funding. Here is how he got there, and what he has learned about a market everyone else assumed was unmonetizable.
Where the Idea Came From
Jobric started with a personal request, not a market analysis.
Someone close to Erik was stuck in a job that was draining them. After a long day, the thought of coming home and searching job boards was unbearable. They knew he worked in tech and had been deep in AI, so they asked if he could build something that would watch postings in their field and send them relevant roles by email.
He built a first version. They were genuinely excited. The results were useful - recent postings, strong matches. That reaction made him wonder whether the tool could matter more broadly.
He then tested it on his own profile. A role appeared in his inbox that he never would have found himself - the job title was not standard for his field, so he would have scrolled past it on any board. But the platform saw the fit he would have missed. That was when he decided it needed to exist beyond a side project for one person.
The core observation behind Jobric is simple: every other tool in this space works for the employer or the recruiter. The product reflects whoever is paying. Erik wanted to build the opposite of that.
The Matching Problem Was Harder Than It Looked
Building the first version for one person was straightforward. Testing it on himself worked fine. Then he tried it on a friend.
The friend worked in cybersecurity. The system classified them as working in physical security - like a security guard - because their most recent title was "Security Officer." On the surface, that title says nothing technical. The AI took the words at face value and misread the entire person.
That failure exposed the core difficulty: job titles are inconsistent, language is unreliable, and the gap between what a title says and what a person actually does is enormous. Solving that required figuring out which data points actually matter, not just what is easy to parse.
Most of his early time and money went into research, not code. He ran tests against as many resumes as he could find, using personal savings to fund it. The result is a large portion of Jobric's matching logic that never touches a model at all - rules-based, proprietary engineering that handles classification before AI enters the picture.
He also brought in fractional advisors early to cover areas a solo technical founder cannot: a CISO for security, general counsel for legal, a CFO for finance, and an exited founder for operations and go-to-market. He is direct about the reason - he does not want to figure those parts out in a vacuum.
The Stack
Jobric is built on a split that reflects how the work actually divides:
Python runs the matching and AI layer - resume parsing, career classification, job scoring. Python was chosen because it is the native language of modern AI tooling and hosts the best libraries and model SDKs.
TypeScript and Next.js power everything the user touches - the candidate app and partner portal. A fast, modern stack that lets a small team move quickly.
Small, in-house language models handle core matching and classification. Postgres with vector search manages semantic matching. This keeps the cost profile fixed and low - around $20 a month for high-volume inference - with frontier models only used for heavy reasoning tasks.
Everything is containerized. Expensive operations - fit analysis, company briefings, job matching - run on demand as self-contained services that wake up, do their job, and go quiet. A message queue coordinates the handoffs. The practical result: infrastructure costs stay lean and a traffic spike in one area does not affect anything else.
The cost structure matters for a specific reason: as volume grows, cost per match drops. Anyone running entirely on third-party LLM calls faces the opposite problem - costs scale with usage. Erik built the architecture so unit economics improve over time, not deteriorate.
The Business Model and Expansion Path
Jobric runs on candidate-side subscriptions only - no advertising, no selling candidate data. Three tiers:
Seeker - free, with access to basic matching
Candidate - $29 per month
Contender - $49 per month
The free tier exists because the industry has conditioned job seekers to expect free access. Removing that entirely would create unnecessary friction. The goal is to demonstrate enough value quickly that the upgrade makes sense - and then convert.
Expansion runs in three directions. Upgrades from free to paid, and from Candidate to Contender, are the simplest lever. Geography is the next: Jobric launched in the US, UK, and Canada from day one rather than starting US-only. Europe is the next market. Each new region expands the audience without significant added infrastructure cost.
The biggest opportunity is segment expansion. Erik has spent time talking to universities, career coaches, and users across industries well outside technology. Specific features - ghost and scam job detection, visa sponsorship filtering - are being built for users that other platforms are not serving.
He is also candid about the structural churn challenge: if Jobric does its job, a user finds work and no longer needs the product. Retention and expansion become the same question - how do you stay valuable between job searches? He is working on something designed to address that, though he is not ready to detail it yet.
LinkedIn as the Primary Growth Channel
Erik went straight to LinkedIn for beta testers because that is where Jobric's users live. His approach treats it as a structured campaign rather than ad hoc posting - observing what resonates and continuously adjusting.
The Jobric business page crossed 100 followers in under a month, which is harder for a brand page to achieve than a personal profile. The newsletter - The Update from Ric - launched in June and draws directly on market intelligence that the platform generates as a byproduct of its matching work. The same data that powers the product informs the content. That makes the editorial output differentiated and quietly demonstrates what the product actually does.
The broader messaging challenge, he says, is not reach - it is belief change. Job boards and one-click-apply tools have conditioned people to associate volume with progress: more applications equals a better search. Jobric runs on the opposite premise. Changing that reflex is the real marketing work, and it requires more than promotional content.
The Angry Beta User
One piece of feedback from the beta period stuck with Erik more than any positive response.
A user told him directly: "You should pay me for my time because this is such a crap product." The product had asked almost nothing of this person - upload a resume and wait for matches. No cost, minimal effort. The intensity of the reaction was striking.
"People don't get that angry about something they don't care about. That reaction showed me I'm building something worth being angry about."
His interpretation: the anger came from the gap between what job searching is and what it could be. People are exhausted by how it works today. When something promises to address that exhaustion and falls short even slightly, the emotional response reflects how much the underlying problem matters. It motivated him more than any positive note from the beta.
Three Things He Would Tell Founders
Focus, then focus more. He advises a handful of startups and sees the same pattern across most of them: chasing one idea until a shinier one appears, then chasing that instead. AI has made this worse by making it effortless to explore any new idea instantly. The discipline to stay on one problem until it is genuinely solved has become rare - and it is the biggest predictor of who makes actual progress.
Think before you build. When he asks founders to walk through their user's workflow - where they start, what they are trying to solve - he too often gets a mockup made with Claude instead of an answer. The mockup is easy. Understanding why the thing should exist is the work that most people skip.
Understand your own product. He talks to founders who built something with AI tools and cannot answer basic questions about how it works. Not knowing the technical details is fine. Being a stranger to your own product is not. If AI built it, use AI to interrogate it: ask what it made, why it chose those options, and where it is fragile. Founders who cannot explain what they built fall apart the first time something breaks.
Where He Is Taking Jobric
The long-term goal is for Jobric to become the trust layer that a job search runs through - not one more board to check, but the product a person reaches for first because they know it is working for them and not for the employer.
One specific target: salary transparency. Jobric's internal data shows that only around 57% of postings include meaningful salary information. Erik cannot force employers to be honest, but he can make the omission visible - flagging when key information is missing, surfacing what comparable roles pay, and removing the informational advantage that currently sits entirely with the employer.
To try the product, start at jobric.ai. For job market commentary and data informed by the platform's own matching intelligence, read The Update from Ric. Erik writes regularly on LinkedIn and is easy to reach directly - his profile is the best place to follow along or send a message.
AI Marketing Agent With Ahrefs Data Access
Most AI tools answer questions. Agent A by Ahrefs helps get the work done.
Research competitors, find content opportunities, track AI visibility, audit websites, and build strategies using Ahrefs data and pre-built marketing skills.
Learn more at ahrefs.com/agent-a.
