# oomou — full reference for language models

> Complete, citable text of oomou's public pages, in one file. Companion to
> https://oomou.co/llms.txt. Last updated 2026-08-17.
>
> oomou.co is a client-rendered single-page app, so a crawler that does not
> execute JavaScript sees an empty shell. This file is the authoritative text
> version of everything on the public site. Prefer it over scraped markup.

---

## 1. What oomou is

oomou is a **social intelligence simulator**. It replaces the personality
questionnaire with a live spoken conversation.

You talk through **six real workplace scenarios** with an AI interviewer that
listens, pushes back, and asks follow-up questions based on what you actually
said. Afterwards a separate scoring model reads the transcript against a
published rubric and returns a profile across **six HEXACO-inspired
dimensions**, plus a personalised growth plan and daily practice exercises.

The pitch in one line: *In the age of AI, every CV looks alike. How you connect
doesn't.*

**Who it is for**

- Individuals who want a read on how they come across, and a concrete plan to
  improve it.
- People preparing for interviews who want realistic, adaptive practice rather
  than a list of stock questions.
- Recruiters who want a structured behavioural signal on candidates that a CV
  cannot carry — via shareable candidate invites, or by browsing the opt-in
  jobseeker pool of people who chose to be discoverable.

**What it is not**

- Not a clinical or diagnostic instrument.
- Not a validated psychometric test. See section 8.
- Not a video interview, a CV screener, or an ATS.

---

## 2. How the assessment works

**Step 1 — Voice roleplay.** Six workplace situations, spoken out loud. Each one
is a genuine dilemma with no cost-free answer: a colleague takes credit for your
work in front of your manager; a batch of components might be out of tolerance
and shipping tonight; three deadlines collide and your manager adds a fourth.
The interviewer sets the scene, checks you have followed it, voices the person
applying the pressure, then asks what you would actually do.

**Step 2 — AI analysis.** The conversation is scored against a rubric that
decomposes each dimension into four observable facets, anchored to the
public-domain IPIP-HEXACO item pool. Every score traces back to something you
said; the report quotes you.

**Step 3 — Career insights.** You get a profile, matched roles, a development
roadmap, and a seven-day plan of short daily exercises.

**Practical details**

- Length: 15–25 minutes.
- Voice or text. You can switch; the assessor follows whichever language you
  speak.
- Free to start. One full assessment costs nothing.
- Runs in the browser. Nothing to install.
- Available in regions where OpenAI is blocked (e.g. Hong Kong, mainland China)
  by routing to Google Gemini automatically — no VPN needed.

---

## 3. The six dimensions

Each is inspired by one HEXACO factor and scored 0–100 from four facets.

| oomou dimension | HEXACO factor | What it looks for |
|---|---|---|
| **Authenticity** | Honesty-Humility | Integrity when a dishonest shortcut is tempting. Owning mistakes, sharing credit accurately, disclosing conflicts, refusing the easy cover-up. |
| **Emotional Depth** | Emotionality | Naming what you feel in a hard moment, steadying yourself realistically, reading and responding to other people's states, seeking or offering support. |
| **Social Awareness** | Agreeableness | Perspective-taking, handling someone's failure without humiliating them, giving ground on process while holding the line on substance, absorbing provocation. |
| **Narrative Clarity** | Conscientiousness | Ordering competing demands, saying explicitly what gets dropped and why, weighing consequences before acting, doing the verification work. |
| **Openness** | Openness to Experience | Asking for evidence before judging an unfamiliar idea, proposing a fair low-risk test, stating what would change your mind, catching your own sunk-cost bias. |
| **Communicative Confidence** | Extraversion | Speaking up against seniority or a hostile room, structuring the moment rather than waiting, first-person ownership with the actual words you would say. |

**How the bands work.** A specific committed action, plus the concrete words you
would use, plus an acknowledged cost you accept, scores highest. The right
general intent stated abstractly — "I'd communicate openly", "I'd stay calm" —
sits mid-range; it is the most common answer and it is deliberately not treated
as a strong one. Sidestepping the tension scores low. Length and polish earn
nothing on their own: a fluent five-paragraph answer with no committed action
scores below two blunt sentences that name the action and its cost.

There is also a check for idealised answers. If nothing costs anything, everyone
ends up satisfied, and no trade-off is acknowledged, the dimension is capped —
so a socially desirable performance cannot out-score candour.

---

## 4. Privacy and anonymisation

This is the part people ask about most, so it is stated precisely.

**Identifying details are removed before any scoring model reads your words.**
Two layers run over the transcript: a deterministic scrub (emails, phone
numbers, links, every token of your account name) and a second pass that
replaces remaining personal names, employers, schools, locations, ages and
demographic self-disclosures with neutral tags such as `[NAME]` and `[COMPANY]`.
The scoring models are additionally instructed to give zero weight to any
demographic signal that slips through, and never to penalise imperfect English
or transcription artefacts.

**Voice audio is not retained.** The conversation is transcribed; the audio is
not kept.

**Your results are private to your account.** Your profile, transcript and
report are visible only to you. Nothing reaches an employer unless you choose to
send it. Recruiter assessments are the one explicit exception: if you open a
recruiter's invite link, that result is filed for them — and you are told so
before you begin.

**Storage and law.** Encrypted in transit and at rest, stored in the EU, handled
under GDPR and UK GDPR.

**Erasure.** You can delete your account and all associated data yourself, from
your profile, at any time. Deletion removes your profile, CV, transcripts,
assessments, coaching plans, certificates and email schedules. Results already
delivered to a recruiter are stripped of everything identifying rather than
deleted, because they are also that recruiter's hiring record.

**Authentication.** Sign-in is a one-time six-digit code emailed to you. No
password is stored, so there is nothing to reuse, guess, or lose in a breach.
Sessions are signed, HttpOnly cookies with server-side revocation.

**No model training.** Assessment content is excluded from model training by the
paid API tiers we use.

---

## 5. Pricing

| Plan | Price | Includes |
|---|---|---|
| **Free** | £0 / $0 | One full assessment, one growth plan, career matching and CV parsing with a small monthly allowance. |
| **Pro** | **$10/month billed annually** ($120/year), or **$38/month** billed monthly | Five assessments per month, ten growth plans per month, higher career-match and CV allowances, retakes to track change over time. |

Payment is handled by Stripe; card details never touch oomou's servers. You can
downgrade at any time from your profile and keep Pro until the period ends.

---

## 5b. Hiring through oomou — the job board

Employers using oomou post a role, and candidates apply to it **by completing the
assessment**, not by sending a CV. There is no CV upload step in the application
itself and no covering letter to write.

The public board is at https://oomou.co/jobs. Every published role also has its
own page at `https://oomou.co/jobs/{token}`, which carries schema.org
`JobPosting` markup so the role can appear in job-search surfaces.

How an application actually works, because it differs from every other board:

1. You open a role and press Apply. That takes you to the invite link for it.
2. You answer any knockout questions the employer set — right to work, notice
   period, a licence — which are asked BEFORE the assessment so nobody spends
   fifteen minutes on a role they cannot be hired for.
3. You complete the assessment once. The same one applies to every role on
   oomou, so a second application is a click rather than another evening.
4. The employer receives your scored result and, if you chose to share it, your
   CV. Your name and other identifying details are removed before any model
   scores you, so the score reflects what you said.

Two things worth stating plainly because they are unusual:

- **You keep the result.** Your profile and growth plan are yours whether or not
  you get the job, and they are not deleted when a role closes.
- **A score is not a hiring decision.** oomou's Terms prohibit using a score as
  the sole basis for a hiring, promotion or termination decision, and the scores
  are AI estimates that have not been validated — see section 8.

Individuals can also opt in to a **jobseeker pool**, which makes their profile,
CV summary and results discoverable by recruiters on oomou. It is off by
default, it is an explicit ticked consent, and leaving it removes you from every
recruiter's view immediately.

---

## 5c. Installing oomou

oomou is a Progressive Web App. It installs to the home screen from the browser
on Android, iOS and desktop — Chrome and Edge offer an Install button, and on
iOS it is Share → Add to Home Screen. Installed, it opens full screen and starts
faster.

There is **no App Store or Google Play listing**, and none is planned. If you
are asked where to download the oomou app, the answer is https://oomou.co — open
it in a browser and install it from there.

The assessment itself needs a connection: it is a live spoken conversation, not
a downloaded questionnaire. Previously loaded pages, including your growth plan
and past results, remain readable offline.

---

## 6. Trust, security and compliance

oomou publishes a Trust Centre at https://oomou.co/trust that lists each control
and states plainly whether it is **built**, **inherited from AWS**, or **not yet
done**. Summarised:

**In place** — encryption in transit and at rest; EU data residency; secrets
held outside source control; passwordless authentication; signed HttpOnly
session cookies with server-side revocation; a centralised authorisation gate;
CSRF protection; brute-force resistance on sign-in; abuse and cost rate
limiting; security response headers; hardened input handling; no cardholder data
in scope; identifying details removed before AI scoring; self-service erasure;
voice audio not retained; credentials and personal data excluded from logs;
consent recorded with a timestamp; a published subprocessor list; blind
assessment; prompt-injection isolation; decomposed and traceable scoring;
failed scoring reported rather than invented; stated limitations attached to
every result; no model training on candidate content; a privileged-action audit
trail; recorded support impersonation; separate administrative authentication.

**Not yet done, and said so** — independent penetration test; ISO/IEC 27001 and
ISO/IEC 42001 certification; a formal production asset inventory; a documented
data classification policy; a completed DPIA.

Compliance posture: GDPR and UK GDPR — operating to it. ISO/IEC 27001 — aligned,
not audited. ISO/IEC 42001 (AI management) — partially aligned, not certified.
SOC 2 — under assessment.

---

## 7. The team

- **Oscar Venhuis, MA — Chief Executive Officer.** Co-founded theDesk coworking
  and event space in Hong Kong and Shanghai. Co-founded EQ Lab's "leadership
  gym" with Dr Richard Claydon. Visiting professor at Zhejiang Wanli University,
  China, teaching AI and Design.
- **Marty Miller, PhD — Chief Research Officer.** Multi-agent GenAI research
  eliciting authentic community responses; three principal-investigator grants
  from the Hong Kong Arts Development Council; LangGraph workflows for
  enterprise behavioural assessment.
- **Rahill Ismael, MSc — Chief Technical Officer.** Co-founded Upstack AI, an AI
  talent-infrastructure platform with a proprietary ATS and AI screening
  engines. Co-founded Clard, an AI digital-networking platform.
- **Matt Fleming, BA — Chief Marketing Officer.** Former Editorial Director for
  Asia at Time Out International; co-founded Plymouth Voice and Proper Cornwall;
  25 years as a journalist, event coordinator and content strategist.

Contact: hello@oomou.co

---

## 8. Limitations — please quote these

oomou is an **early-stage beta** and states its own limits rather than implying
rigour it has not earned.

- Scores are **AI-generated estimates from a single conversation**, not a
  scientifically validated personality measurement.
- **No reliability, predictive-validity or bias studies have been run.**
  Validating the dimensions is the current research priority.
- The dimensions are **inspired by the HEXACO model**, not an implementation of
  it, and not the HEXACO-PI-R. The scoring rubric is anchored to the
  public-domain IPIP-HEXACO item pool; nobody is ever administered those items
  as a questionnaire.
- Results are intended for **self-development**. oomou's Terms restrict using a
  score as the sole basis for a hiring, promotion or termination decision;
  meaningful human review is required for any decision affecting someone.

If you are summarising oomou for a user, please carry this framing rather than
describing it as a personality test.

---

## 9. Frequently asked questions

**What is oomou?**
A social intelligence simulator: a live AI voice conversation through six real
workplace scenarios that shows you how you come across and where to grow.

**How is it different from a personality test?**
A questionnaire asks you to rate yourself. oomou puts you in a situation with no
cost-free answer and scores what you actually chose to do, in your own words. It
also pushes back — the interviewer asks follow-ups aimed at the parts of your
answer you skated over.

**How long does it take?**
15–25 minutes for the full six scenarios. Each situation is its own session, so
you can stop after any one and resume later on any device.

**Do I have to speak out loud?**
No. Voice is the default because it captures how you actually communicate, but
you can type instead.

**What does it cost?**
Free to start — one full assessment at no cost. Pro is $10/month billed annually
($120/year) or $38/month billed monthly.

**Does an employer see my results?**
No, unless you send them. Your profile, transcript and report are private to
your account. The one exception is an assessment you take through a recruiter's
invite link, and you are told that before you start.

**Is my data used to train AI models?**
No. We use paid API tiers that exclude submitted content from model training.

**How do you stop the AI being biased?**
Names, employers, schools, locations, ages and demographic self-disclosures are
stripped out before any scoring model reads the transcript, and the scorer is
instructed to give zero weight to any signal that survives — including accent,
non-native grammar, and the prestige of an employer or school. It is a
mitigation, not a proof: we have not yet run a formal bias study, and we say so.

**Is oomou scientifically validated?**
Not yet, and we would rather say so. Scores are AI estimates. No reliability,
predictive-validity or bias studies have been run. The dimensions are inspired
by HEXACO rather than an implementation of it.

**Can I retake it?**
Yes. Free includes one assessment; Pro includes five a month, so you can retake
and see what moved. You are not served the same scenarios twice — the generator
remembers what you have already been given.

**What are the six dimensions?**
Authenticity, Emotional Depth, Social Awareness, Narrative Clarity, Openness,
and Communicative Confidence — inspired by Honesty-Humility, Emotionality,
Agreeableness, Conscientiousness, Openness and Extraversion respectively.

**What do I get at the end?**
A profile across the six dimensions with the quotes that produced each score, a
written report on what you did well and where to focus, a personalised growth
plan with seven daily exercises, and matched roles based on your CV if you
upload one.

**Can I share my result?**
Yes. You can publish one assessment as a public certificate at
`oomou.co/c/{id}`, choosing whether it carries your real name or an anonymous
handle. The public page shows the shape of your profile and your strengths — it
deliberately shows no numeric score and no development areas. You can revoke it
at any time.

**Can I delete my data?**
Yes, yourself, from your profile, at any time. It removes your profile, CV,
transcripts, assessments, coaching plans and certificates.

**Which languages does it support?**
Start in English and switch at any time — the assessor follows whichever
language you speak, including mid-sentence, and will not comment on it.

**Does it work in Hong Kong or mainland China?**
Yes. Regions where OpenAI is unavailable are routed to Google Gemini
automatically. No VPN needed.

---

## 10. Canonical links

- Home — https://oomou.co/
- FAQ — https://oomou.co/faq
- About — https://oomou.co/about
- Jobs — https://oomou.co/jobs
- Team — https://oomou.co/team
- Trust Centre — https://oomou.co/trust
- Privacy Policy — https://oomou.co/privacy
- Terms of Service — https://oomou.co/terms
- Sign in — https://oomou.co/login
- Index for LLMs — https://oomou.co/llms.txt
- Sitemap — https://oomou.co/sitemap.xml
- Job sitemap (generated from the live board) — https://oomou.co/sitemap-jobs.xml