A build from the PromptMates community, and a preview of where recruiting is headed
For decades, the front door to a company has been the same: a career page, a list of open roles, a resume upload, and a form. Whether you’re applying to a startup or a Fortune 500, the mechanics haven’t really changed: collect an application, screen it, move the best ones forward.
Stephanie, a member of the PromptMates community, just rebuilt that front door from scratch. And the way she did it is worth paying checkin’ out because it’s not a new tool bolted onto the old process. It’s a different process entirely.
What She Actually Built
Stephanie replaced her company’s traditional career page and application workflow with an AI-guided conversational screening system. There are three real changes stacked on top of each other here:
1. The entry point changed from role-based to function-based.
Instead of browsing a list of job postings and applying to a specific one, candidates select the function they’re interested in (say, product, engineering, or ops). If there’s a current open role in that function, it surfaces automatically. If there isn’t, candidates can select “Future opportunities” and stay in the pipeline for what’s next.
This alone reframes the relationship. The old model asks: “Which of our open jobs matches you?” The new model asks: “What kind of work are you good at?” and lets the company match that answer to current or future need.
2. The resume and application form were replaced with an AI conversation.
Instead of uploading a resume and answering static application questions, candidates go through a roughly 10-minute AI-guided conversation (typed or spoken) about what they’ve built and how they think. The landing page frames it explicitly: “This is a conversation about who you are across everything you have done. Not an interview, and no resume to live up to.”
That framing matters. A resume is a performance: optimized keywords, curated bullet points, etc. A conversation is much harder to stage. It’s designed to surface reasoning and problem-solving in real time, rather than a polished summary of past reasoning and problem-solving. Which, I absolutely love this.
3. The conversation itself is dynamically generated from real job and company context, via MCP.
This is the part that makes it more than a chatbot. On the backend, Stephanie connected MCP to Gem, her recruiting CRM, and Metaview. That connection lets the system pull the actual job details, the live job posting, and notes from the hiring team’s Metaview kickoff call, and use all of that context to build the conversation a candidate actually has.
In other words, the AI isn’t asking generic behavioral questions. It’s asking questions shaped by what the hiring manager said matters most in the kickoff call for that specific role, at that specific moment.
Put those three changes together, and the career page stops being a place to collect applications and becomes a system that actively conducts a structured, context-aware conversation with every candidate. That’s not a new vendor swapped in for an old one. It’s a re-platforming of what the top of the recruiting funnel actually does.
Why This Will Spread, and Where It Won’t
It’s tempting to say “every company will do this soon.” That’s too broad a claim, and not a very useful one. The more accurate, and more useful, version is: this becomes standard for knowledge-work hiring at companies optimizing for quality over volume.
Why it spreads
- Resumes are losing their signal value. AI-written resumes and ATS keyword-stuffing have made “credentials on a page” a weaker proxy for capability than it used to be. Companies are actively looking for a replacement signal, and “watch someone reason through something in real time” is a strong candidate.
- LLMs just made this cheap to build. Dynamically generating a tailored conversation from live job context wasn’t practical a couple of years ago. Now, as Stephanie’s build shows, it’s something one person can put together and ship.
- The plumbing already exists. MCP-style connections mean recruiting tools (ATS platforms, CRMs like Gem, interview intelligence tools like Metaview) can be stitched together without a custom integration project. That’s what makes “pull real job and company context automatically” feasible for a small team, not just a large one with an engineering budget.
- Candidate experience is a real differentiator. In a tight talent market, “have an actual conversation” versus “fill out a form” is a noticeable difference for candidates, especially ones who feel like resumes undersell what they can actually do.
Why it won’t be universal
- Volume economics don’t work everywhere. Stephanie flagged this herself in her rollout notes: her organization doesn’t manage high-volume inbound applications, so trialing a 10-minute conversational format doesn’t create a bottleneck. That math changes fast for hourly, retail, or high-volume roles, where a form still processes candidates far more efficiently.
- Compliance and legal risk are real and growing. Any system where AI is making evaluative judgments about candidates sits inside an increasingly active regulatory space: NYC Local Law 144, the EU AI Act, and EEOC guidance on algorithmic hiring tools are all relevant here. Companies in more regulated hiring environments will move slower, or need audit and bias-testing layers before they can adopt something like this.
- It only works where “how someone thinks” is the actual hiring signal. This approach fits roles where capability and reasoning matter more than a checklist of certifications or credentials. For roles where credentials genuinely are the qualifying signal, a conversation isn’t obviously better than a form. It’s just different.
- It shifts effort earlier in the funnel. A 10-minute conversation is a bigger ask than a 5-minute form. Stephanie was candid about this tradeoff too: this approach could create a barrier to entry for candidates unwilling to navigate a nontraditional process. For some companies, that’s an acceptable filter. For others, especially those competing hard for high volumes of applicants, it’s a real risk.
The Honest Tradeoff, In Her Own Words
What stands out in how Stephanie shared this internally is that she didn’t oversell it. From her note to the team:
“We are a small organization and do not manage high volume inbound so trialing it will not present a massive negative impact; so I recognize that this approach could create a barrier to entry for some that are unwilling to navigate a different application process.”
That’s a useful model for anyone considering something similar: know exactly which conditions make the experiment safe to run, and be upfront about who it might exclude. This isn’t a “resumes are dead” pitch. It’s a scoped bet, made by someone who understands both what she’s gaining and what she might be trading away.
The Bigger Pattern
Zoom out, and this build is one instance of a broader shift in how AI-native teams are approaching hiring: using live company and job context to actively evaluate candidates, rather than passively collecting information and hoping to screen it well later.
That shift, from collect and screen to engage and evaluate, is the part I think is worth watching. The specific tools (Gem, Metaview, MCP) will change. The underlying move, of replacing static application collection with dynamic, context-aware candidate conversations, is likely to show up in more recruiting stacks over the next year, especially at companies that are small enough to experiment and clear-eyed enough to know exactly where the tradeoffs are.
This build was shared by Stephanie Happ inside the PromptMates community. If you’re experimenting with something similar, or thinking about where AI belongs in your own hiring funnel, the conditions Stephanie laid out (volume, role type, and candidate expectations) are a solid checklist to start from. Thanks for sharing Stephanie, I absolutely love this one!!