Real startup task
Beyond the Resume
Startups need help, students want startup work.
Applications are what make the match harder.
Talent reaches startups by referral.
- Pay is competitiveStartups compete on compensation. That is not what stops the match.
- The path is one address bookThe strongest people arrive because somebody already knew them.
- There is no process behind itNothing repeats it for a student nobody happens to know.
AI makes applications harder to trust.
- People overstate what they can doAnyone can now describe work they have never done, diluting the talent pool.
More students want
startup work.
More students want to contribute.
Quality over quantity
See the work.
Then decide.
A Covenda profile shows what a student built, the skills they can prove, when they can work, and what happened in a trial.
| AI and everywhere else | Covenda | |
|---|---|---|
| You receive | A résumé, often AI-written | A work profile with projects and availability |
| Filtered by | School and keywords | Skills, work type, stage, and time |
| Matched by | A generic job post and a model’s guess | The role and problem you describe |
| Before hiring | More interviews | A small paid trial with a clear finish |
| A strong trial | Another application | Can lead to an internship, project, or role |
| Who decides | The algorithm | The company and student |
Covenda makes the work visible.
Illustrative path. A trial creates evidence, not a guaranteed offer.
One project. Two sides win.
Referred talent, real work, one trial both sides can trust.
Scoped work trial
Referred student
Vouched by people who know them
For the startup
Useful deliverableFor the student
Credible proofMove through the trial
Let the company find you
Do not feed your resume to an AI agent and dilute your talent. Build a profile that lasts and grows.
Join without choosing
Join the Talent.
Pick a work area to see the exact kinds of work inside it.
Build the intern you need.
Describe the person, not a job post. We design a paid trial that does real work and shows whether the fit is right.
Tell us the problem
The intern you actually need
We turn it into a paid trial.
Hire vetted students
Hire from a vetted batch
Vouched undergraduates, reviewed before you meet them.
- Many arrive endorsed by a professor, club, or pro, and that endorsement is attached where you can read it
- Judged on proof of work, reviewed projects, GitHub
- Work-trial first, deeper access is earned
How it works
Vouched in. Matched out.
Talent reaches Covenda through people who put their name on it, then the engine ranks for the specific request in front of it.
Robotics startup
Talent request“Need 3 robotics students”
Robotics coach
Student A
Professor
Student B
Research mentor
Student C
Covenda talent pool
Compatibility engine
Top candidates
For technical recruiters
Shortlists that survive the manager.
Source, screen, build a pipeline, assess technical fit. Only the last one decides whether your shortlist holds, and it is the one a résumé cannot answer.
Résumé
- ClaimedListed it
- ClaimedListed it
- ClaimedListed it
Can they actually do it?
Covenda
- BuiltRepo they own
- VerifiedAgainst a published bar
- DefendedOn the record
The manager checks it
Scoped like your desk
Twenty-seven specialisations across software, finance, healthcare, retail and professional services.
The bar is written down
So "vetted" has an answer when a hiring manager asks what it meant.
Evidence states its own limits
Each piece says what it proves and what it does not. Nothing oversells on the way to your team.
No placements yet, so no time-to-fill to show you and none invented. No résumé database, no score to sort by.
Two ways in
Find a matched student. Or post one paid trial.
-
1
You have a problem.
Signups drop after onboarding.
-
2
Open Talent.
Covenda fills in what you need.
-
3
Illustrative matchMaya Chen
React product linked · contribution confirmed
-
1
Post one small task.
Audit the signup flow.
-
2
Qualified students apply.
You choose who starts.
-
3
Run the paid trial.
Clear finish. Named reviewer.
Illustrative flow. No placement or outcome is claimed.
Build credibility
through referrals.
Founders already ask you who is good. Today that stops at whoever has your email.
How it works today
- One founder emails two professors he happens to know.
- One professor names two students he happens to remember.
- One hire happens. Everyone else never hears about it.
It works because someone credible vouched. It doesn’t scale because it runs on one address book.
The same thing, systematised
- Your club vouches once. Members carry it into every application.
- Covenda vets again, against a published bar, so a company trusts it without knowing you.
- Every startup in the vertical reaches your students, not just the founder with your email.
- Your club keeps the record, and it outlives this year’s officers.
A selective club anywhere builds credibility a stranger will trust.
Ready to put names forward?
Add the students you would stake your name on. Nothing is shared until you send it.
Built by students, for students.
What we build on
Three pillars
Everyone else is building AI-powered hiring. We are building evidence-backed hiring. One predicts who might be good. The other shows you work they already did.
-
Evidence over claims
Demonstrate what you can do. Do not just state it.
-
Company-specific fit over generic matching
Learn what makes people succeed at each company, rather than ranking everyone by the same criteria.
-
High-agency talent over mass-market recruiting
Build a trusted network of people who consistently learn, create, and take initiative.
The problem
Hundreds of applications
nobody wrote.
Four hundred cover letters in the same voice. The signal is gone, so hiring falls back on the school name and who you know, and the person you actually needed is somewhere in the pile.
Applying costs nothing
One prompt makes a hundred applications. An afternoon of real work looks identical to nine seconds of it.
So the résumé stopped meaning anything
Nobody reads four hundred. They get filtered on the one line a model cannot fake: where you got in at seventeen.
And platforms hard-code it
Most sell access school by school, so the candidate you would have hired is filtered out before you ever see them.
What we do instead
- A model cannot sit for the interview. Every application carries an unscripted recording. Use AI to prepare, not to be there.
- Built by students, not bought. It grows by people vouching for work they have actually seen.
- Vetted per industry, not per school. Robotics and private equity are judged on different evidence, against a published bar.
- The output is finished work. The only application material a model cannot produce for you.
Why now
Talent exists.
Nothing pairs it.
Startups can reach students big employers stopped hiring. The gap is knowing who delivers.
- 7%New grads are now just 7% of Big Tech hires, down by half since 2019. That supply is unclaimed.SignalFire State of Talent, 2025
- 52%of new grads are underemployed, capable people paired to work that never tested them.Burning Glass Institute & Strada, 2024
- 45%are still underemployed a decade on. A bad first pairing compounds; it does not self-correct.Burning Glass Institute & Strada, 2024
- 50%of students would rather build something than climb a ladder. Startup work fits them, if it reaches them.Morning Consult for Samsung, 2023
These figures cite published research, not Covenda results. Covenda has no outcome data yet, the pilot is how we earn it.
Every ambitious student hits the same wall: “I could contribute here. I just can’t get in front of the founder.” What works is what always worked: someone credible stakes their name on you, and the work proves the rest.
The vetting system
Vetted twice, then proven.
Canopy · what a company sees
A record, not a résumé
Accepted work becomes employer-verified evidence in one vertical, with the reviewer, the date and the deliverable attached. Per-skill only, never a person score, never a ranking of students.
Employer-verifiedPer skillEvidence attached
Trunk · the second vet
Covenda vets again
A published bar per batch, checked against real evidence: connect the account that owns the work, have the file parsed, or defend it on record to two raters. Clearing it is a recommendation, an operator decides and records why.
Published barTwo ratersHuman decides
Roots · how proof is taken
Three ways to prove it
Ownership and timestamps where an API can reach, a connected GitHub account, months of commits. Programmatic parsing where a file can be read. A recorded walkthrough where neither applies, which is most fields.
Owned accountParsed artifactRecorded defence
Soil · the first vet
Someone screened first
A selective club admitting a few percent of its domain has already done vetting we could not write, and repeats it every year. Clubs lead; a lab PI or a professor is the same rail, carrying less on its own.
Verified clubsLab PIsFaculty
Student of the week
Meet someone who ships.
Vetted before the trial begins.
A referral starts the signal. Evidence has to confirm it.
01
Stand out
What you own, what you shipped, and you explaining it.
- Ownership
- Artifact
- Walkthrough
02
Evidence checked
Every claim is checked against the work behind it.
03
Admitted to a batch
Only students who clear the field-specific bar.
Qualified pool
04
Ready for a trial
Companies meet a small, evidence-backed pool.
Trial ready- 1
Tell us your skills
Personalise your account into a quality demonstration of what you can do, instead of a résumé and AI slop.
- 2
Join a batch
Let companies find you.
- 3
Keep the proof
Accepted work becomes a capability record, not a vague résumé bullet.
Illustrative example
ProfileWhat you can already do
FieldSoftware & AI, product engineeringStrongest skillFront-end, can work independentlyAttachedTwo repos you own, one deployed project, one endorsement
Illustrative example
OpenSoftware & AI batch
Published barOne project you own, defended live, with its limits statedReviewed byA practitioner in the field, not a keyword filterWhat companies seeThe batch first, then your profile inside it
Illustrative example
AcceptedWhy signups stall at step 3
CompanySeed-stage B2B SaaS, 6 peopleDeliverableFunnel walkthrough of 400 sessions, plus 3 fixes ranked by build effortScope14 hours over 2 weeks · $420 · public analytics export only
Vetted, ready
for a trial.
Covenda carries the scoping and review.
- 1
Company problem
You describe what is delayed and what useful looks like.
- 2
Covenda scopes
We isolate a safe work unit, inputs, and review plan.
- 3
You approve
Your team confirms the facts, boundary, deadline, and pay.
- 4
Student works
A qualified student works from approved materials.
- 5
Covenda reviews
Completed work returns to Covenda before delivery.
Start with bounded, reviewable, low-risk work.No production access or restricted records.
- Deliverable
- Public customer research summary
- Approved inputs
- Public reviews, product pages, approved sources
- Information boundary
- No internal data, customer lists, or private documents
- Acceptance criteria
- 12+ sources, themes traced, clear recommendations
- Deadline
- Agreed before publication
- Fixed pay
- Confirmed before student assignment
What you actually do.
- 1
Add who you vouch for
Name, school email, interest area. Paste a whole list at once.
- 2
Covenda routes the work
Scoping, safety, and review are ours. You do none of it.
- 3
They keep the proof
Accepted work becomes a record the student controls.
What Covenda is
Can you decide what to execute?
Most hiring asks whether you can execute. Startups care far more about the question above, and Covenda is built to answer it.
A profile you own
Built from what you have actually done, with what each piece proves and what it does not.
Agency, measured
What you do when the obvious path closes. A transcript is blind to it.
Reaches the team
Your work goes to the people who would work with you, not into a queue.
Three ways in.
One ladder up.
Any of the three gets you a start. What you do after is the same for everyone.
Ways in
- A verified clubStrongest, carries to every member
- ResearchYour PI answers for the work
- Vetted workNo club, no lab, no contacts
- Faculty endorsementStill counts, lightest of the four
Then the same ladder, for everyone
- 01EndorsedA start
- 02Role-qualifiedA reviewed work sample
- 03Employer-verifiedAccepted, paid work
- 04ProvenInterview, referral or offer
Useful without being risky.
Start where context can be made portable, the finish can be reviewed, and the student does not need access to critical production systems.
Good first fitPublic-source research, data cleanup, QA, documentation, workflow mapping
Needs redesignUnclear deliverables, heavy live supervision, undocumented context
Not eligibleProduction access, client or patient records, regulated decisions, restricted information
A pipeline of proof.
The students you vouch for get bounded, paid, reviewed work-trials, and your vouch gains weight as they deliver. No marketplace to run, no accounts to manage.
Good fitProfessors, club officers, and industry pros who can speak to a student’s real ability
How data is usedNames and school emails go to the Covenda pilot team only, no public listing
Not includedAccounts, rankings, automatic placement, or a guarantee of paid work
Trials + Batches
Trials work.
Batches shortlist.
Two different things, and they only make sense together, one earns the evidence, the other is what companies pay to search.
A trial
Paid work you actually do
A company posts bounded, paid work. You do it; a named reviewer accepts or asks for changes. No batch needed.
What you get: paid, plus accepted work on your record.
A batch
The vetted list companies search
A small cohort in one field. Companies spend credits to open it and read the evidence. You get in by clearing a published bar, accepted trial work is the strongest way to clear it.
What you get: found by companies who never advertised the role.
No evidence yet? Start with trials. A batch is where finished work gets you seen.
The batches
Select the ones you’re considering.
See it for your industry
What this looks like for you.
Vetting differs by field. Pick yours.
Illustrative walkthrough, no real company, no real student, and no claimed outcome.
What a referral becomes
One vouch, six months on.
You add a name. This is what it turns into.
- 1You vouch
A name and an interest area. Thirty seconds.
Day 0 - 2They clear a batch bar
Your name carries them to review, not past it.
Week 3 - 3A paid trial is accepted
A company accepts the work. That is employer-verified evidence.
Week 9 - 4It compounds
An interview, another project, or an offer, recorded against your vouch.
Month 6
University robotics club
14students vouched
9cleared a batch
6employer-verified
3hired or interviewed
No student names, just the numbers. Yours to show a department, whether or not the club keeps referring.
Illustrative example. Covenda has no outcome data yet, this is the shape of the record, not a claim about one.
For school clubs
Register free.
Verification is earned.
Anyone can register. Verification comes from what your members deliver.
Verify your club
1 Register
Free, and commits you to nothing.
2 Earn verification
Standing opens doors, it does not walk through them. Members still apply with their own recorded work and are judged on it.
Referrer track record
Your endorsements are a reputation asset.
Every student you refer builds your track record, something you can show a department. Counts below are illustrative until real outcomes accrue.
About us
For students,
built by students.
Covenda is building the operating system for startup hiring: evidence-backed talent profiles, combined with an understanding of how innovative companies identify, evaluate, and develop exceptional people.
We are Tyler and Dylan. We are 19, and we built this because getting a first real internship was harder than the work turned out to be once we had one.
The process rewarded people who already had a way in. A name on a referral, a parent in the industry, a school on a list. What it did not reward was being good at the work, because there was no way to show that before you were hired.
So we built the thing we needed. A referral opens the door, and finished work decides the rest.
Start with the work you own.
Submit an interest profile. We’ll use it to shape the first pilot group, not to promise a job or automatic match.
Bring the work that slips.
Book 20 minutes to walk through a real problem, or write it up. We assess whether it can become a safe, scoped, paid project.
Vouch for students you back.
Add the students you’d stake your name on. Covenda routes paid work-trials as they open, and your vouch gains weight as they deliver. Founder-led, no placement guarantee.