Accept reality and succeed

Accept reality and learn from it. If what we are doing right now isn’t working, we have to change our tactics.

Set high standards

Good enough is only the beginning. Set a high standard and stick to it even under pressure.

Give credit where credit is due

Recognition is a motivator that helps build trust and improve performance.

Just enough process

Process should serve us. We should not serve the process.

Problem

Time to demonstrating value to prospective customers is currently too high. It is estimated at more than 3 minutes before a prospect hears the product work.

Phone numbers are not verified. Anyone can enter any number and trigger at least one call.

Solution in brief

Verify the phone number with SMS, put a Try It button on the public site so a prospect hears a call in about 10 seconds, replace the guided tour with coach marks at the right moment, and spin up a trial campaign right after the demo so they see real outbound calls instead of filling in a long form.

Customer Story

I signed up for the demo account. There was a LinkedIn sign up screen, then an unskippable guided tour, a lengthy form filling process to create the first campaign… which didn’t work. Also, you currently do not verify the phone number with an SMS. That means I can enter any phone number that I want and make at least one call to it.

Definition of aha moment

Six turns by the AI, six turns by the customer.

Requirements

Verify phone number

Add SMS verification to phone number entry.

“Try It” button on the public site

• Put a Try It button on the public website.
• Click opens a pop-up asking for a phone number.
• Verify that number with an SMS.
• Once they enter the code, they receive a call immediately.
• This path should take about 10 seconds.

Replace the guided tour with context-driven coach marks

• Remove the current guided tour.
• Detect where the customer is in the journey.
• Highlight the next step when it is time — not a tour that talks over an empty product.

Pre-configured trial campaign immediately after demo call

• Guide them toward creating their own campaign. Do not interview them with a form.
• Spin up a predefined trial campaign.
• Ask for two more phone numbers — three outbound calls in total, including theirs.
• Those calls go out one after another so they can see calls actually leaving the system.
• Ask if they would like to edit the campaign.
• Let them edit it step by step.

Metrics

Objective metrics

• Lift in sign-up rate
• Lift in first call
• Lift in people who choose to upgrade

Guardrail metric

• Cost of prospect calls

Experimentation

We will run an A/B test for each requirement. 50% of prospects see the existing flow, 50% see the change. Each test is judged on the objective metrics, with cost of prospect calls as the guardrail.

• “Try It” button and call
• Pre-configured trial campaign
• Replacing the guided tour with context-driven coach marks
• Reducing the number of turns — can we shrink how many turns of conversation we need to get to aha without hurting the objective metrics?

Risks

The main risk is cost: people try the feature, burn telephony / voice / LLM minutes, and never purchase. Limit that risk by capping the length of the demo phone call.

Executive summary

Hurraay wants a marketplace like Mercor. Hiring managers will not pay into that system without a roster of workers. Interview practice is how we attract job seekers first, get some recurring revenue, and seed the roster with consent.

• Job seekers outnumber hiring managers, interviews are scarce, and not everyone has a coach. They might try, and even pay for, practice with an AI interviewer.
• This is not a crazy idea. Mercor, Interviewing.io, Ben Erez, and LinkedIn already show pieces of the model: AI interviews, paid practice, and charging both sides of the market.
• Even if this line does not make a lot of money, traffic can still cross-sell into the marketplace. We can reuse Nexora and Hurraay voice tech. Investment is mostly prompts plus a simple job-seeker web app.
• Go to market in phases: private MVP and high-signal trials (Phase 0), then LinkedIn ads and referrals (Phase 1), then product-led growth through sharing (Phase 3). Skip real payments at first. A false door is enough to see if people would pay.
• Watch unit economics, subscription pricing, niche scope, adverse selection on the roster, and the conflict of interest in scoring people we also train. Sunset it if a tight pilot does not produce consented, outcome-linked data, or if employers do not pull from that roster.

Current Situation

Robin mentioned that the current vision was to establish and move towards a marketplace like Mercor.

• Attracting hiring managers who will actually pay — both the platform and the workers — will be hard without an existing roster of workers.
• We need to attract workers and seed the platform with a roster of prospects before we can attract job posters or hiring managers.

Business Goal

• Get recurring revenue for Hurraay.
• Strengthen and establish a marketplace.

Thesis

The job-seeker side of the market

• In any job market, job seekers will outnumber recruiters and hiring managers.
• Personal primary research has also indicated that it's really hard to get even a single interview, even after hundreds of applications.
• Not every job seeker has access to friends or coaches they can practice mock interviews with.

Why they might pay

Since it is really important to perform in the one interview call that a candidate gets after a hundred applications, candidates might be willing to try, and even pay for, interview practice run by AI interviewers.

Examples

This is not a completely crazy idea. There is already some proof behind it.

• Mercor is the marketplace shape Robin is pointing at: AI interviews plus a roster of workers that hiring managers actually pay for.
• Interviewing.io already showed that people will pay to practice interviews, and that practice can feed recruiting. They stayed smaller because the interviewer was still a human.
• Ben Erez is charging $500 for unlimited text interview practice. Job seekers will pay when the one interview after a hundred applications matters that much.
• LinkedIn already charges money from both sides: job seekers and hiring managers. LinkedIn Learning sits on the candidate side as courses that become credentials on the profile. Charging seekers for something that also helps the hiring side is a model the market already understands.

Supports the existing vision

Even if we do not see a lot of revenue from this new business line, as long as we have sufficient traffic we can still:

• Cross-sell practice users into the marketplace roster.
• Make those candidates available for hiring managers to pick from.

Flywheel

1. Job seekers join the interview-practice side of the business to try and improve their interviewing skills for jobs both on and off the platform.
2. A percentage of these job seekers become paid users. They use the product enough to give us quality signals.
3. A percentage of job seekers from step 2 are added to the roster with consent.
4. The total number of candidates on the roster increases. Hiring managers have more choice, and the perceived value of Hurraay increases in their eyes.
5. As more hires take place on Hurraay, other job seekers will also be attracted to the interview-practice side of the platform, which feeds back into step 1.

Where the money is

• Consumer revenue should cover its own acquisition.
• The real P&L is pay-by-hire, then employer seats, then a credential hiring managers trust.
• Expansion is every job where the work is talking — sales, CS, operators — which is also Nexora’s world.

Shared technology

We can reuse the underlying agentic voice technology that we have in Nexora and in the existing Hurraay to also power this new business line.

Investment can be limited to:

• Modifying the prompts to conduct interviews beyond the screening call.
• Building a web app to get sign-ups from job seekers and let them practice interviews.

Go to market

Phase 0 — Zero to one

What we do first is a zero-to-one phase. LinkedIn outreach, live events, and partnerships with organizations.

• Build a very simple MVP.
• It will not be on the public website to begin with.
• Reach out to job seekers privately, through networks or community events where a lot of job seekers are showing up to get help these days.
• Give out free or at-cost interviews to a few high-signal job seekers who can try it and give us good feedback.
• Once the trial is over, make a paid option available.
• Work on credits or a monthly subscription, ask people to pay, and see how many actually would.
• We do not need to build the payment feature yet. A false door is enough to validate the idea.
• Marketing and distribution will heavily depend on LinkedIn and effort from the company’s employees.
• Attend job-seeker meetups in Berlin.
• Partner with job training centers.
• Partner with social entrepreneurship organizations that are helping job seekers right now.

Phase 1 — Ads and referrals

Once the rate of learning from Phase 0 plateaus, we go to Phase 1.

• Run ads on LinkedIn.
• Referrals in the app: three free interviews for the person being invited, and one free interview for the person doing the inviting.

Phase 3 — Product-led growth

Build ways for candidates to share their achievements socially, and other features where candidates share something with their network. To participate, they have to create an account on Hurraay.

What has to be true

• Candidates complete the round and a share of them pay.
• A meaningful share consent onto the roster — without being coerced.
• Employers prefer consented, spoken-proof people over résumés.
• Those people get hired faster, or at higher quality, than the baseline.
• The eval trained on those outcomes beats a generic voice agent. That is the moat test.

Risks

Unit economics

We get the unit economics wrong. Mitigate this by giving out a limited number of free calls at first, measuring how much it costs us, and using that to estimate how much we should charge per call.

Subscription pricing

We get subscription pricing wrong. Hold off on subscriptions until we understand the unit economics. Then see how much people will use with unit pricing. Use unit economics and the cost of the 90th percentile of usage to set pricing until we find the balance between profitability and growth.

Scope

The job market is huge. Creating an AI coach for PM interviews is easy. Creating an AI coach that can handle marketing, operations, different industries like tech and automotive, and also other non-white-collar jobs is going to be a challenge.

• Pick a niche and stick to it in the beginning.
• Expand slowly into adjacent areas.

Adverse selection

There is a self-selection bias in who might sign up for the interview product. Job candidates who have trouble getting jobs, and who have trouble performing in interviews, will be more likely to try it out and then pay for it. But these are also the kind of candidates that hiring managers will not, or have not, hired so far. Moving a lot of them into the roster might affect its quality and reduce the trust we have from hiring managers.

Conflict of interest

Hurraay is creating a roster of “quality” candidates for hiring managers. Hurraay also trains candidates with practice interviews. Hurraay is placing candidates who use practice interviews on the roster as “quality” candidates. This participation in both sides of the marketplace might reduce the trust hiring managers have in the scores Hurraay gives to job seekers.

Gating and Sunset Criteria

After a tight pilot (one role, one market), stop building practice as a product if:

• Rounds do not produce consented, outcome-linked data, or
• Employers do not pull from that roster.

The question

What does Nexora / Hurraay do better than a smart person with an API key, or an enterprise company with deep pockets? In five years, how does the company still exist if frontier labs ship Voice Agent as a feature?

Data and scale: the usual answers

• Data. Generate your own data that nobody else has.
• Scale. Grow large and achieve efficiencies of scale.

These textbook answers are not applicable to baby companies just starting out. Of course, Nexora / Hurraay should focus on building these advantages early for winning in the long term. But what defends us until then?

Simpler defenses that can save us this week

• Local GTM advantage. Anthropic can’t run a meetup every week in Berlin.
• Relationships and references. Trust built from professional networks.
• Made in EU pride. Building third options beyond US and China giants.
• Local context. Compliance, interfacing with local works councils.

Going deeper made this section too long. Let’s discuss in person.

What is it?

One hour a week, everyone gathers to contribute ideas for growing the company. Together, we put as many ideas on the table as we can, including “dumb” ones. People are welcome to discuss half-baked ideas, get input and feedback, work on them, and come back with the same idea but in better health next week.

Why the company needs it

Hurraay / Nexora is early. Direction is still being set. Product-market fit is found by trying things and discarding them.

• We run this to put new ideas on the table for growing the company: product, GTM, ops, hiring, anything that could move us.
• Half-baked is welcome.
• A dumb idea is useful when it makes someone else think, and that person brings a better one.
• The more ideas we generate, the more shots we get.
• A small team that only hears from founders will under-sample the search space.

This session exists so the company does not starve itself of options.

Cadence

• Weekly, one hour. Not daily.
• Daily would pressure people into junk, and we cannot test ideas that fast.
• A week is long enough to bring something real, and short enough that the backlog does not rot.

Participants

• Founders — same rules as everyone else. No veto in the room.
• Engineering
• Product
• Sales and GTM
• Operations
• Everyone else on the payroll — no spectators

Facilitator

One facilitator rotates. They:

• Keep the clock
• Protect the “no dunking” rule
• Make sure names land on the log

Participation

This is all hands. It defines the future of the company, so no one sits it out. That is not a threat. We do not use discipline or coercion.

• We ask for cooperation, not compliance: show up because the room is safe, credit is public, and a sentence is enough.
• Bring something.
• Listen like it might be the one.
• Do not shoot ideas down.
• Park, prototype, or pick an owner before we leave.

Not every idea ships. Every idea is written down, with a name on it, forever.

Recognition — Wipe Code

Wipe Code is a small internal app: a public page that records who earned a slot in the meeting.

• Example slots: best idea of the week, most courageous.
• It is a leaderboard we can tally over time — T-shirt, hat, bragging rights — not a bonus plan.

No cash

Do not tie this to money.

• Cash makes people game the system.
• It costs the company.
• Extrinsic rewards have been shown to reduce creativity.

Status and craft stay. Pay stays off the table.