JOBO uses artificial intelligence to support companies in evaluating candidates. This page explains, plainly, what the system evaluates, what data it uses, who decides and what you can do if you disagree. It is not a courtesy text: what we state here binds us towards you (article 61 of the Spanish consumer protection act), so we only state what we can prove — and we also state what we do not have yet.

Its basis are articles 13, 14, 15(1)(h) and 22 of Regulation (EU) 2016/679 (GDPR) and Regulation (EU) 2024/1689 (AI Act), whose timeline changed in July 2026. We summarise it here and detail it in section 10: most of the AI Act high-risk regime will not be enforceable until 2 December 2027, and we apply it already on a voluntary basis; and what has been prohibited since 2 February 2025 — inferring a person’s emotions in the workplace — we comply with by design (section 5.bis).

1. Nature of the system

JOBO qualifies as a high-risk AI system under the AI Act because it supports hiring decisions. JOBO does not make automated decisions producing legal effects on the candidate: the outputs are decision-support tools for the hiring company, which must ensure meaningful human interventionbefore any final decision.

Who reviews what, and in which order. This is worth knowing precisely, because there are two human interventions and the first one is not the company’s:

1. The AI flows produce an evaluation from your answers.
2. A JOBO reviewer reviews it and decides whether it is published to the company. That review has three modes, and we tell you which is which because how much it weighs depends on it: detailed —the reviewer confirms the evaluation test by test and may re-evaluate one if they disagree, recording their reason in writing—; inherited tests —those already certified by a reviewer in an earlier process of yours are not reviewed again, and are recorded as inherited—; and fast acceptance —when an evaluation shows no alert signal, the reviewer may publish it in a batch together with others, and it is recorded that it was done that way—. If it is not published, the company never sees it.
3. Only once the evaluation is published does the company weigh it together with the rest of your application and decide on hiring. The hiring decision is always theirs.

In other words: the first human filter is JOBO’s, not the company’s. If your evaluation is not published and you want to know why, ask for a review or share your point of view, write to revision@joboapp.es (section 6).

“High-risk” does not mean “unreliable”. It is the label EU law puts on any AI system used to select staff, because of what is at stake for you, not a judgement on its quality. What it entails is that the system must be documented, monitored and explainable, and that you have concrete rights against it: they are the ones in section 6.

1.bis. Who is who: JOBO, the company and you

Almost every doubt about who to complain to is settled by this table. The roles are not interchangeable: each law allocates responsibilities differently over the very same process.

What this means in practice. If you want to contest an evaluation, always start with us: revision@joboapp.es or the in-app button (section 6). We review it and, where the decision belongs to the company because it is the controller, we pass your request on and confirm to you that we have. We are not going to send you from one door to another: the door is ours.

2. AI model and provider

Assessments are processed using Google Gemini 3.6 Flash via Google Genkit. Your data is not used to train public models of Google or of any third party. Outputs are stored on Neon (PostgreSQL) and Google Cloud (Firebase Storage) infrastructure.

Which model evaluated you is on record. Every evaluation records the exact version of the model that produced it, the flow that ran and the date (section 5). Changing model or reasoning level is, for us, a substantial change to the evaluation system, not a routine update: it is documented before it is applied.

The full list of providers involved and of the international transfers they entail is in Subprocessors and in section 6 of the Privacy Policy.

3. What is evaluated and how it is scored

Before the list of tests, the thing that matters most: how the score is set.

Everything is marked against a fixed rubric, not at the AI’s discretion. Every test has its scoring levels defined in advance, and what an answer has to show to reach each of them. The model does not pick the score it feels like: it has to fit your answer into one of those written levels and justify which one it applied by quoting what you said. It is the same mechanism as a public examination: the answer is checked against a written scale, not against an opinion. And a person then certifies it (section 1).

JOBO evaluates four pillars:

1. Soft skills. Job-contextual situational scenarios scoring professional competencies against a rubric with written anchors. Combined with a team-roles questionnaire computed with fixed rules, no AI.
2. Cognitive capabilities and simulation. A reasoning questionnaire and a resource-management simulation. The score is computed with fixed rules on our servers, with no AI; there is an explanatory text written by AI, but it does not change the score.
3. Technical knowledge. Questions tailored to the role and the experience level, marked against a rubric with fixed criteria and caps.
4. Visual interview. Six video questions, evaluated from the transcript of what you say: structure, clarity and pace of the discourse, with fixed levels per dimension. The model that scores neither sees the image nor hears the audio — it receives text only. Emotions and personality traits are NOT inferred (article 5.1.f AI Act).
   The company may add up to five extra questions of its own (for example, about your portfolio or your previous experience). Those questions are not evaluated by any AI and do not score: they are only transcribed and shown to the hiring manager, who writes them and is answerable for their content. Each one adds two minutes to the recording time.

A block can drop out of the weighting. The 4 pillars do not always weigh the same: the split depends on the type of role and on which tests the process includes. And since 16 August 2026 it can also depend on something that happened to you: if you submit the video interview and we fail to assess it —for instance, if the recording cannot be processed— the behaviour block does not score 0: it leaves the split and its weight is shared among the others. Your score is then computed over the blocks that were actually measured, and the company is told that the test came out “not assessable” and how many blocks the score covers. A 0 and a “could not be measured” are different things and are not conflated. Not taking a test, by contrast, does count as 0. Today this rule applies to the video interview; for the other tests, a result that could not be measured still counts as 0.

3.bis. English level test

If the role requires it, you will take an English level test with a written part and a spoken part, and in that case it is compulsory: without it your application is not complete. It is evaluated by artificial intelligence against a written rubric, like everything else: content, organisation, vocabulary and grammatical accuracy are scored separately, each on a scale with defined levels, and from that comes a CEFR level (A1 to C2). Flows: evaluate-english-writing, evaluate-english-speaking. The spoken part is assessed from the transcript of your answer, not from the audio signal: your browser’s speech recognition transcribes what you say, JOBO only receives the text and JOBO does not record the audio.

What this result is and is not. It is an indicative estimate of the English level you demonstrate in that particular test, expressed on the six-band scale in common use across Europe (A1 to C2). It is not a qualification, nor an official certification, nor equivalent to one. No examination board issues it, JOBO is not accredited by any, and the result cannot be used to evidence your level to third parties.

It is independent from the selection score: your English level adds no points and subtracts none from any of the 4 pillars or from the final score. The company does see the result: the estimated level and, where the role sets a minimum, a “meets / does not meet” against that minimum. That comparison is a fixed check between two levels, not an extra judgement about you. If any of the test’s five skills was left unmeasured, the level is marked as partial and it is stated neither that you meet that minimum nor that you do not.

And it rejects nobody. Not meeting the minimum does not remove you from the process, does not lower your score and does not take you off the list: it is information shown to the company, and it is the company that decides what weight to give it. JOBO sets no candidate aside over their English level.

3.ter. Job fit check (no AI, but it can cancel the visual pillar)

You answer a few fit questions drawn from the requirements the company set in the job offer: availability to work in the location, work arrangement, contract type and equivalent profile requirements. No AI evaluates them: it is a deterministic check, with fixed rules, over what you yourself declare. They are not an assessment of you as a professional, but a check against the conditions of the role. You are asked them in every process: at the start of the visual interview when the process includes one, and when you apply when it does not.

We spell out its real effect because it matters: these are the job’s must-have requirements, not an assessment. If you answer “No” to any of them, your final score in that process is 0. The reason is simple: you do not meet a condition without which you could not hold the role —you cannot work in the location, you cannot start on the date, you do not hold the required qualification—, so there is no score to give. This happens whether or not the process includes a visual interview; and when it does, your video is still transcribed and evaluated, so that its assessment is on record in case that answer was a mistake and gets corrected — but that scoredoes not change the 0either.

If you answer “Yes, but…” nothing is cancelled: your comment is sent to the company and the application is flagged for a person to review and decide. And if you were never asked (processes predating this screen), you are not filtered for it. This filter is not an AI flow: it lives in JOBO’s scoring logic (src/lib/scoring), with fixed rules.

3.quater. Other automated processes

Besides the above, JOBO runs automated processes that do not score your application: the fit report delivered to the company, the explanatory texts of each section, the questions suggested to the hiring team for the final interview, salary and market insights, translation of the offer content and of the extra questions written by the company, offer autofill, and CV improvement suggestions, which are addressed to you.

Of the tests that do score, four are marked by AI against a rubric and three with fixed rules:

Cultural fit. It compares the preferences you declare with the values the company declares.The calculation is deterministic, without AI, and it does not score: it neither adds to nor subtracts from your mark. AI intervenes at a single point, and before you come in: when the company writes its values in its own words, a model proposes which catalogue values they match, anda person at the company accepts or rejects each proposal. It is not a validated psychometric test and must not be used to reject you.

Anti-cheating signals: indications, not proof. While you take the tests, the platform records objective patterns from your device —tab switches, leaving the window, attempts to paste text, implausible typing rhythms, a covered camera, multiple screens— and shows them to the company. You are told about thisbefore each test starts, and your acceptance is recorded. Three things worth knowing:

• They are computed with fixed rules, not AI, and they are not an accusation: almost all of them have innocent explanations —a notification, a second monitor, someone who types very fast—.
• They do not reject you. No signal discards anyone automatically: the system is advisory by design and the decision is made by a person at the company.
• They do not enter your score. They are shown separately, and their intended use is to ask you how you took a test, not to reject you outright.

What your score means

Each test produces a score per dimension. That score informs the company about your fit for the role, together with the rest of your application. It does not reject you automatically: before the company sees anything, a JOBO reviewer certifies the evaluation and decides whether it is published; the hiring decision is then taken by a person at the company (see section 1). If you believe an evaluation does not reflect your profile, you can request a review, share your point of view or ask for its erasure (see section 6).

4. What is evaluated and what is not

This is probably the most important section on this page, because it contradicts what almost everyone assumes about a hiring process: the only thing evaluated is your answers to the tests.

Goes into the evaluation:

• Your answers to the tests (text, options and simulation decisions).
• The transcript of what you say in the video interview.
• Your answers to the English level test: the written text and the transcript of what you say in the spoken part (the audio is not recorded).
• Your answers to the job-fit questions (what you yourself declare about availability and conditions).
• The job context provided by the company (description, requirements, competencies sought), which is what makes the questions and the marking specific to the role rather than generic.

Does NOT go into the evaluation, and this is a rule of the system, not an intention:

• Your CV. Neither the file, nor its text, nor a summary of it. You may upload it and the company may read it, but no score is computed from it.
• Your age and date of birth. The date of birth is asked at sign-up for one thing only: to check that you are of legal age. It reaches no evaluation system.
• Your nationality and origin. They are not asked and do not exist as data on the platform.
• Your education, qualifications and previous experience. The company sees them if you publish them; the score does not use them.
• Your name, your photo and your image. The system that scores the interview receives text, not video or audio.
• Your sex, marital status, health, religion, beliefs, sexual orientation or trade union membership. They are neither asked nor processed.

In one sentence: your score comes from what you do in the tests, not from who you are or where you come from. What the company looks at separately —your CV, your track record— is its business and its decision, not part of our evaluation.

4.bis. How long each thing is kept

A retention period is a safeguard, not an administrative detail: it is what stops an evaluation of yours from years ago still going around. These are ours (article 5(1)(e) GDPR).

The full detail, category by category, is in the Privacy Policy. You may request erasure before those periods at dpo@joboapp.es.

5. Implemented safeguards

Non-verbal analysis framework (AI Act art. 5.1.f compliance). JOBO does not infer emotions, personality or intent from candidates. In the video interview we describe only observable discursive indicators, grounded in published academic literature. Each dimension carries an explicit instruction not to label affective states. The human reviewer integrates the observation with substantive content analysis before any effect.

• A written, closed rubric: no test is marked at the model’s discretion. The scoring levels and what must be shown to reach each of them are written in advance, the model must fit the answer into one of them and justify which one it applied by quoting what you said (section 3).
• Inference log: each time an AI process evaluates your answers it leaves one row, the one for its final result, with the process that ran, the model version, the date and a cryptographic digest of the input and the output. The log is immutable — it cannot be rewritten after the fact — and is kept for 4 years. It is what makes it possible to reconstruct, years later, what evaluated you and with what. It is the practice the AI Act will require in 2027 and that we apply already.
• Human certification before publication: no evaluation reaches the company without a JOBO reviewer reviewing it and deciding to publish it, in any of the three modes in section 1 —detailed, inherited or fast batch acceptance—. Each certification is recorded with its author, its date and the mode used.
• AI Act filter: the video evaluation includes a banned-terms list (emotion, personality, honesty, etc.) and a post-scan flagging any suspicious output for review.
• Assessment language locked: the language you are assessed in is set by the role and cannot be altered from the interface, so that two candidates in the same process are not assessed under different conditions.
• Input limits and defence against text manipulation: answers have size caps and go through checks that prevent a text from trying to instruct the model to alter its own score.
• Reusable results (90 days): the results of the reusable tests —cognitive, team roles, simulation and English level— are retained for 90 days and reused across the positions you apply to during that period, so you do not have to retake them and the same candidate is not re-tested repeatedly; while they remain valid you cannot retake them at will, and normally you will only have to sit a test again if the position you apply to uses a different version of it. They are deleted once the period ends.
• Expired session closure: an automated process finalises pending tests so no orphan data remains.
• Sensitive change and deletion logs, kept for 5 years.

5.bis. What we do NOT do

Negative commitments are the ones that can actually be checked. These bind us, and any of them we stopped honouring would be a breach, not a change of product:

• We do not infer emotions, mood, personality or intent. It is prohibited in the workplace by article 5.1.f of the AI Act since 2 February 2025, and it is the only part of the AI Act on high-risk systems that is already fully enforceable today.
• We use no biometrics: no facial recognition, no voice identification, no biometric categorisation of any kind.
• We do not score your face or your voice. The model that scores the interview receives only the transcribed text: it neither sees the image nor hears the audio.
• We do not process your nationality or origin: they are not asked and do not exist as data on the platform. Your date of birth is used solely to check that you are of legal age at sign-up, and reaches no evaluation system.
• We do not evaluate your CV, your education or your previous experience. The score comes only from what you do in the tests (section 4).
• We do not leave the score to the AI’s discretion. Everything is marked against a rubric written in advance and a person certifies it afterwards.
• We do not train models with your data, neither our own nor third parties’.
• We do not reject you automatically. No score closes a process on its own.
• We do not score protected characteristics —age, sex, origin, health or disability, pregnancy, religion, sexual orientation, trade union membership—, and companies are contractually barred from asking about them in the questions they write.
• We do not build a file that follows you from one company to another. Each company sees the results of its own process; it does not see which other processes you applied to or what you were told in them.
• We do not change an evaluation because the company asks. It is corrected if a review concludes it is wrong, not because someone dislikes the result.
• We do not sell your data nor share it for advertising purposes, and we do not connect web analytics to the evaluation system (section 8).

One nuance we would rather tell you. JOBO has built a second system that does look at the video, in order to describe —without judging you— observable indicators of your discourse; what it produces is a description, not a score. It is currently switched off, so no AI looks at the image of your interview: what you say is transcribed and that text is what gets evaluated. Were it switched back on, it would still not enter your score and would remain subject to every prohibition in this section, starting with not inferring emotions.

5.ter. Fairness: what we do so the evaluation does not discriminate

A system that evaluates people has to be able to prove it does not discriminate. The strongest defence is not a statement of intent: it is that the data one could discriminate by never enters the system. And it does not.

• Your nationality and origin do not exist as data on the platform: not asked, not stored, cannot be used.
• Your date of birth serves only to check that you are of legal age at sign-up. It reaches no evaluation, neither directly nor summarised.
• Neither your sex, nor your health, nor your family situation, nor your religion, nor your sexual orientation, nor your trade union membership are asked or processed.
• Your CV, your education and your experience stay out of the calculation (section 4): what is evaluated is what you do in the tests.
• All candidates for the same role sit the same tests, in the same assessment language, against the same rubric and the same criteria.
• The criteria are written down and are the same for everyone, and the marking has to be justified by quoting your answer — an unsupported judgement does not count.
• Every evaluation goes through human review before the company sees it.

Fairness audit: it is already done. The systematic bias examination required by article 10 of the AI Act does not become binding on us until 2 December 2027 (section 10), and itwas carried out on 15 and 16 August 2026 on the AI-based evaluators, with thresholds fixed before looking at the results (article 9(8) of the AI Act) and in a reproducible way. The report itself states which protected grounds still lack measurement: it is not presented as a complete picture.

And what it still lacks, because keeping quiet about it would be the opposite of an audit. The report is signed by the responsible person (17 August 2026), as Annex IV requires, but the check of real outcomes by group could not be run yet: it is waiting for a sufficient volume of candidates. That is why you will not read «audited and bias-free» here. A result of this kind also expires with the model and the criteria: changing them requires repeating the measurement, and the report records exactly which version was measured. If you want the consolidated report, ask for it at legal@joboapp.es.

6. Your rights

As a candidate you may exercise the following rights at any time:

• Information and explanation: receive meaningful information about the logic applied and the foreseeable consequences of the processing (articles 13(2)(f), 14(2)(g) and 15(1)(h) GDPR).
• Human intervention: request that a person review an evaluation you consider inaccurate or unfair.
• Express your point of view: provide additional context or contest an evaluation.
• Access, rectification, erasure, restriction, objection and portability: over all your personal data (articles 15 to 21 GDPR).
• Not to be subject to a decision based solely on automated processing with legal or similarly significant effects (article 22 GDPR). At JOBO that situation does not arise: marking is done against a written rubric, a human reviewer certifies every evaluation before it is published, and the hiring decision is taken by a person at the company. The right is stated all the same because it is yours, not because we fall short of it.
• Explanation of the individual decision: article 86 of the AI Act will give you the right to a clear explanation of the role the AI system played in the decision affecting you. That article will not be enforceable until 2 December 2027 (section 10); we give you that explanation already, if you ask.

Your score is not shown to you in the app, and that is a deliberate decision. In your dashboard you see the status of your process, not the score. The reason is twofold: to stop the tests from being gamed by repetition, and to stop a decontextualised figure from reading as a judgement on you. That does not cancel your right of access: if you ask for it, your result is provided along with a plain-language explanation of what was evaluated and how it was arrived at.

What that explanation includes and what it does not. It includes which tests you sat, what was measured in each, what result you obtained and what the marking relied on. It does not include the internal detail of the system —the code, the calculation formulas, the weightings between tests or the full text of the rubrics—, which are JOBO’s trade secrets and are covered by article 15(4) GDPR, which prevents the right of access from adversely affecting the rights of others. That limit will never be used to deny you the explanation of what was evaluated and how your result was reached: it protects the “how it is built inside”, not the “what happened to you”.

When it makes sense to ask for a review. You do not need to have seen a score. These are the real cases, and any of them counts:

• A test failed technically: the recording cut out, a question did not load, time ran out because of a connection problem.
• You believe the transcript of your video does not capture what you said — the most frequent case, because everything evaluated from the video goes through it.
• You answered a job-fit question by mistake, or your circumstances have changed since.
• You sat the test in conditions that put you at a disadvantage and want that taken into account, or you need an accessibility adjustment.
• The company has not selected you and you want to know what your evaluation said, share your point of view or contest it.
• You want to see your result (paragraph above) or to have it erased.

How to ask. Write to revision@joboapp.es stating which process you mean. Use that address for anything concerning your evaluation; for all other data protection rights (general access, rectification, erasure, etc.) use dpo@joboapp.es. You may also lodge a complaint with the Spanish Data Protection Agency (aepd.es).

Asking for a review does not count against you. Requesting human intervention, contesting an evaluation or exercising any of these rights does not penalise your application nor is it reported to the company as a negative data point about you. On receiving your email we enter your request into the review queue with the date it arrived, which is the date the deadline in section 6.bis runs from. In My processes you will see the status of each application, but the review is not requested from there.

6.bis. How we handle a challenge, and how fast

A right with no deadline and no procedure is half a right. This is ours, and the deadline binds us:

1. Your request comes in by email to revision@joboapp.es, stating the process. The deadline starts running from the date your email arrived.
2. It is entered into the review queue, with the date your email arrived and the specific evaluation it concerns.
3. A person reviews it, with access to your answers, to the evaluation and to the record of which model produced it. Whoever resolves it is always a person: there is no automated system deciding on your challenge.
4. We resolve in writing and with reasons: the evaluation is confirmed with an explanation of why, corrected, or withdrawn. If it is corrected or withdrawn, the company receives the corrected version or stops seeing the previous one.
5. If you are not satisfied, we tell you who to turn to —the company as controller, and the Spanish Data Protection Agency— and give you whatever you need to do so.

Deadline: one month from your email reaching our inbox — the one set by article 12(3) GDPR. If the case is particularly complex, that period may be extended by two further months, but only by telling you within that first month and why. And it runs from the moment the email arrives, not from the moment someone opens it.

7. Known limitations

No evaluation is infallible, and we would rather say where its real limits lie. They are not in the AI “deciding on its own”: it does not, because it marks against a written rubric and a person certifies the result before the company sees it (sections 3 and 5). They are elsewhere:

• Applying a rubric well is still a judgement. Two markers —human or automated— may place the same answer in adjacent levels. That is why the marking must be justified by quoting your answer, why a person reviews it, and why you can ask for it to be reviewed again.
• Everything evaluated from the video goes through a transcript, and a transcript can get things wrong: with accents, with noise or with technical terms. If you believe yours does not capture what you said, that alone is reason enough to ask for a review (section 6).
• A test measures what it measures. It describes how you answered specific questions, at a specific moment and for a specific role. It is not a portrait of you and it does not predict what you will do on the job.
• The format can penalise. The tests are taken in writing and on camera, and that can disadvantage anyone with technical, connectivity or accessibility difficulties. If that is your case, write to revision@joboapp.es before or after the test: adjustments are handled through that channel and are described in the Accessibility Statement.

This section does not limit your rights. Nothing said here releases JOBO from its obligations or restricts the rights the law gives you as a consumer and as a data subject (article 86 of the Spanish consumer protection act). The hiring company remains responsible for the decision it takes, and JOBO for the system doing what this page says it does.

8. Web analytics (does not apply to evaluation)

JOBO uses web analytics and marketing tools (audience and campaign measurement) only on the public pages of the site. This analytics is not an automated individual decision within the meaning of article 22 GDPR: it does not decide or influence your selection, it does not profile you as a candidate and it is independent of the AI evaluation system described on this page. As an express limit, JOBO does not connect web analytics or advertising pixels to the candidate-evaluation pipeline: no candidate behaviour during the tests is sent to these platforms. Details of these cookies and their legal basis (your consent) are in the Cookies Policy and the Privacy Policy.

9. Behavioural Profile Test (does not apply to evaluation)

In the public resources section we offer a free Behavioural Profile Test, open to anyone. It describes behavioural styles and is not an automated individual decision within the meaning of article 22 GDPR, nor part of the high-risk system described on this page: it does not score, does not rank anyone, plays no part in any selection, and its result is not shared with companies. It also uses no artificial intelligence: the calculation is deterministic — a fixed formula, always the same, with no generative model — and runs on our servers. Its basis is your consent, requested before you answer anything, and your answers are deleted automatically after 90 days. As an express limit, JOBO does not connect this test to the candidate-evaluation pipeline.

Do not confuse it with the behavioural profile inside the platform. If you are enrolled in a process and take the test from your dashboard, that result is seen by the company running the process (levels per axis, dominant axes, consistency indicator, ticked words and your self-description). It is also voluntary, also deterministic and it likewise neither adds to nor subtracts from your selection score, but its recipient is different. It is described in the Privacy Policy.

10. Applicable legal framework and from when

This is where most pages of this kind turn vague. Ours does not: the AI Act changed its timeline in July 2026. Regulation (EU) 2026/1744, published on 24 July 2026 and in force since 27 July 2026, postponed from 2 August 2026 to 2 December 2027 the obligations for Annex III high-risk systems, which is where JOBO sits. That changes neither what we are —we remain a high-risk system— nor what we do: it means that much of what is described on this page we apply by our own decision before it is required of us.

Why we tell you this. Because the usual alternative is to cite the AI Act with no dates and let it be understood that everything is already complied with. We would rather you knew exactly what is an obligation in force, what is a voluntary commitment of ours and what is still outstanding (section 5.ter).

11. What this document is, how it changes and who answers

What it is. An informational document: you are not asked to accept it and it is not a contract. That said, what it states binds us as pre-contractual information (article 61 of the Spanish consumer protection act). It does not replace the Terms of Service or the Privacy Policy. Should anything on this page contradict them, whichever wording is more favourable to you prevails.

How it changes. Every revision raises the version and the date you see in the header. The full history is available on request at legal@joboapp.es, and the status of the fairness audit in section 5.ter is updated right here whenever it changes.

Who answers (article 10 LSSI). JOBO PEOPLE TECHNOLOGIES, S.L. (sole-shareholder company), NIF B88736665, Calle Barrau 1, Bloque H, Portal 6, 1.º A, 41018 Seville, Spain. Full details in the Legal Notice.

Where to complain, in order. First to us: revision@joboapp.es for your evaluation, dpo@joboapp.es for everything else. If our answer does not satisfy you, before the Spanish Data Protection Agency (aepd.es) and, once Spanish law designates the AI Act market surveillance authority, before it as well. As a consumer, you also keep whatever consumer complaint routes apply to you.