Muchas personas usan un asistente de empleo con IA solo para pulir el currículum. Sirve, pero deja mucho valor fuera.
El flujo más útil es filtrar vacantes primero y adaptar el CV solo para las que valen la pena.
Prepara tres tipos de material
Antes de reescribir, prepara:
- perfil personal;
- biblioteca de experiencia;
- preferencias de postulación.
El perfil puede incluir rol objetivo, años de experiencia, ubicación, stack, CV actual y límites que la IA no debe cruzar.
La experiencia conviene organizarla por proyecto:
1
2
3
4
5
6
7
8
9
10
11
12
|
## Project: Internal knowledge-base search system
- Role: backend developer
- Time: 2025.03 - 2025.09
- Tech: Python, FastAPI, PostgreSQL, vector search, Docker
- Work:
- designed document parsing and ingestion
- added permission checks
- optimized retrieval latency
- Results:
- average query latency dropped from 3.2s to 1.1s
- used by 5 business teams
|
Preferencias:
1
2
3
4
5
6
|
## Application Preferences
- Prefer: remote/hybrid, AI applications, platform tools
- Acceptable: SaaS, internal systems, data platforms
- Avoid: outsourcing, long-term on-site roles, sales-heavy roles
- Weekly limit: 15 high-quality applications
|
Paso 1: pedir que analice la oferta
No preguntes primero si eres buen candidato. Pide estructura:
1
2
3
4
5
6
7
8
9
10
|
Please extract structured information from this job description:
1. Must-have requirements
2. Nice-to-have requirements
3. Actual responsibilities
4. Implied capabilities
5. Likely interview focus
6. Keywords my resume should address
Do not evaluate my fit yet.
|
Paso 2: filtrar con una matriz de evidencia
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
|
Based on my profile and experience library, evaluate this role.
Output a table with:
- job requirement
- my supporting evidence
- match level: strong / medium / weak / none
- whether more evidence is needed
- how to reflect it in the resume
Finally give an application recommendation:
A: strongly apply
B: apply with tailored resume
C: pause
D: do not apply
If there is no evidence, write "no evidence". Do not invent experience.
|
Una matriz de evidencia vale más que un porcentaje de coincidencia.
Paso 3: clasificar vacantes
| Nivel |
Señal |
Acción |
| Alto |
requisitos clave coinciden con proyectos reales |
adaptar CV y carta |
| Medio |
dirección correcta con algunas brechas |
ajuste ligero |
| Bajo |
palabras clave similares pero rol distinto |
omitir o guardar |
El objetivo no es postular más, sino postular mejor.
Paso 4: reescribir solo lo relevante
Evita reescribir todo el CV.
1
2
3
4
5
6
7
8
|
Based on this JD, rewrite only the resume sections related to the role.
Rules:
1. Do not add experience I did not have.
2. Do not change project facts.
3. Strengthen relevant keywords.
4. Prefer action + method/tech + result.
5. Output before / after / reason.
|
Antes:
1
|
Participated in internal knowledge-base development, responsible for backend APIs and data processing.
|
Después:
1
|
Built backend components for an internal knowledge-base search system, including document parsing, permission checks, and retrieval APIs with FastAPI and PostgreSQL; reduced average query latency from 3.2s to 1.1s for cross-team knowledge lookup.
|
Paso 5: completar la cadena de evidencia
Un buen CV muestra:
1
2
3
4
5
|
qué pide la vacante
dónde hice algo parecido
qué hice
qué resultado produjo
por qué importa para este rol
|
Pide revisión:
1
2
3
4
5
6
7
8
|
Check whether this tailored resume has a clear evidence chain.
For each project, output:
- related job requirement
- whether current evidence is enough
- vague wording
- missing data or context
- anything that sounds exaggerated
|
Paso 6: carta enfocada en el rol
1
2
3
4
5
6
7
8
9
10
11
12
|
Based on the JD and my tailored resume, write a short cover letter.
Structure:
1. why I am interested in this role
2. which two experiences match best
3. what problem I can help the team solve
Rules:
- under 300 words
- do not exaggerate
- avoid empty corporate phrases
- sound like a real candidate
|
Si la carta queda genérica, quizá la vacante no merece prioridad alta.
Paso 7: registrar y revisar postulaciones
| Field |
Example |
| Company |
Example AI |
| Role |
AI application engineer |
| Source |
company site / LinkedIn |
| Tier |
A / B / C / D |
| Match points |
RAG, Python backend, internal system |
| Gaps |
limited Kubernetes experience |
| Tailored resume |
yes |
| Date |
2026-07-10 |
| Status |
applied / interview / rejected / no response |
| Notes |
interview focus and next improvement |
Revisión semanal:
1
2
3
4
5
6
7
|
Review my application log:
1. Which roles got more responses?
2. Which keywords appear in high-match roles?
3. Which gaps repeat?
4. What should I prioritize next week?
5. What small resume changes should I make?
|
Errores comunes
1. Inventar experiencia
Si no hay evidencia, no se escribe.
2. Rellenar de keywords
Las keywords deben estar dentro de acciones reales.
3. Personalizar demasiado cada vacante
Personaliza fuerte solo nivel A; ligero nivel B; omite C/D.
4. Cambiar CV pero no estrategia
La falta de respuestas puede venir del canal, ciudad, seniority o salario.
5. Ignorar ATS
Evita tablas complejas, texto como imagen, encabezados confusos y bullets no copiables.
Prompt reutilizable
1
2
3
4
5
6
7
8
9
10
11
12
13
14
|
You are my AI job assistant. Based on my profile, experience library, and the job description, help me screen this role and tailor my resume.
Output:
1. JD breakdown
2. Matching matrix
3. Application recommendation
4. Resume rewrite
5. Risk check
Important:
- write "no evidence" where needed
- do not fabricate experience, data, company names, or responsibilities
- preserve my own voice where possible
|
Resumen
Un asistente de empleo con IA no sirve solo para pulir CV. Sirve para decidir qué vacantes no postular, adaptar con evidencia real y revisar qué estrategia funciona. Así se convierte en un flujo de búsqueda laboral, no en un generador de frases.