gigsup™
Join for free

The Science

Built for the human,
not the corporation

Every tool a student is handed was designed around what an employer or an institution needed to know. None of them were built around what the person actually is. That single decision is why the whole category falls short, and it shows up in four places.

The architecture

Four layers, and each one feeds the next

Data at the bottom, agents at the top. The record in the middle is the thing every layer reads from and writes back to.

Transcript · SIS Résumé · LinkedIn Monthly journal Labour market Data gateway Four axes · one scale Journal Retaken yearly Shape match Gigi Nine agents Outputs

The agent team

Gigi · eight specialists

Profile, programs, admission plan, scholarships, essays, applications, interview and evaluation. They hand work to each other rather than waiting to be asked.

The living record

Recursive · journal between

VISI is retaken every year and the journal captures what changed in between, so the profile ages with the student instead of freezing them at sixteen.

The instrument

VISI · 176 items

Values, interests, skills and identity, measured as dimensions. Every career profile carries the same four axes, so fit is a shape comparison rather than a label lookup.

Data

What the record is built from

Transcript and SIS, résumé and LinkedIn, the monthly journal, and regional labour market data. Everything arrives through one gateway, and everything a student builds belongs to that student.

01The assessment Fifty-year-old interest inventories that sort you into a type instead of measuring you. 02The fragmentation A test here, a course planner there, a scholarship site somewhere else. Nothing talks to anything. 03The memory An LLM has no record of you. No values, no lifestyle, no four years of how you changed. 04The guidance General models were never trained on what good career guidance looks like. Ours were.

where existing assessments fall short

Most career tools still run on interest inventories designed in the mid twentieth century for a labour market that no longer exists. They ask what you like, sort you into a handful of types, and hand you a list of job titles. Three problems follow.

01Self-report only

A student who was never given the chance to try something cannot report liking it. Interest inventories quietly reward exposure, and exposure tracks with privilege.

02One snapshot, treated as permanent

A single test at sixteen becomes a label carried for a decade. Nothing updates when the student changes, and students change constantly.

03Types instead of dimensions

Six or sixteen buckets throw away most of the information. Two students in the same bucket can want completely different lives.

what visi does differently

VISI measures four dimensions rather than assigning a type. Every career profile carries the same four axes, so fit is a shape comparison, not a label lookup.

Values

What the work has to be for.

Interests

What pulls your attention.

Skills

What you can actually do, self-report blended with measured aptitude from short timed exercises.

Identity

How you want to work and live. Your personality and preferences belong here.

Measured, not only asked. A student nobody called good at spatial reasoning can still discover that they are. Retaken every year. The journal captures what changed between takes, so the profile ages with the person instead of freezing them.

The instrument is a 176-item assessment built through a two-phase validation study as part of our founder's doctoral research, with results expected in early 2027. We will publish the psychometrics. Nobody should take a career tool's word for its own accuracy, including ours.

why an llm alone cannot do career guidance

It has no memory of youA chat model starts cold every session. Career decisions need a record that accumulates over years: what you tried, what you liked, what you quit and why.
It agrees with youLanguage models are trained to be agreeable. Good career guidance sometimes has to tell you the thing you did not want to hear about your average, your portfolio, or your plan.
It cannot measureNo amount of conversation produces a percentile. Aptitude has to be measured against a real population, not inferred from how you describe yourself.
It does not know local truthEntering averages, application deadlines, regional demand and program requirements change every year and vary by province and state. That is data work, not generation.
It will confidently inventAsk a general model for a scholarship deadline and you may get a plausible date that is wrong. A student misses the real one.

We use language models heavily, and we are glad they exist. They are the interface and the drafting engine. They are not the source of truth, and they are not the memory. That separation is the whole architecture.

one platform, end to end

Most tools do one slice: an assessment here, a job board there, an essay helper somewhere else, and the student is left to stitch it together. Gigsup is deliberately end to end: the same record that VISI builds powers the matching, the admission plan, the scholarship search, the skill-gap plans, the mentor outreach and the hiring. Every action improves the next one, because it is one system, not eleven tabs.

you should own your career data

Everything a student builds on Gigsup belongs to that student: the assessments, the journal, the skills evidence, the portfolio, the plan. Not to a school that they will leave in two years, and not to a platform that monetizes their attention.

That principle has a practical edge. Gigsup runs as an MCP server, which means a student can plug their own Gigsup record into Claude or any other MCP-capable assistant and use it for whatever they want: drafting an application in a different tool, prepping for an interview, reasoning about a decision with an assistant they already trust. Your record travels with you. That is what owning it actually means.