Continuously fetches listings from Adzuna, summarizes each listing with GPT-3.5, scores fit against candidate profiles, and logs results to Google Sheets for review.
The AI agent automatically discovers relevant job listings via Adzuna API. It uses GPT-3.5 to summarize descriptions and extract key requirements. It scores each listing against candidate profiles and organizes results in Google Sheets for easy review and action.
An outline of concrete actions the AI agent performs.
Fetches new listings from Adzuna API
Splits listings into individual items for processing
Summarizes each job description using GPT-3.5
Extracts key requirements, location, and company details
Scores listings against candidate profiles and preferences
Logs results to Google Sheets and highlights top matches for review
Before: manual job search wastes time triaging postings and juggling data across sources. Before: summaries require reading long descriptions and guessing fit. Before: inconsistent scoring makes it hard to compare opportunities. Before: data is siloed, scattered across spreadsheets and sites. Before: leads to delayed applications and missed chances. After: you receive curated, scored opportunities with concise summaries and a single, centralized log for review.
A simple 3-step flow that non-technical users can follow.
Collects user-defined job titles, locations, and candidate preferences to define the search parameters.
Calls Adzuna API to retrieve current jobs, splits results into individual items, and uses GPT-3.5 to summarize and extract details.
Applies matching against profiles, assigns scores, and writes results to Google Sheets with review flags.
A realistic scenario showing inputs, actions, and expected outcomes.
Scenario: A recruiter wants to surface 5 senior frontend roles in the US with React and TypeScript. Over about 20 minutes, the AI agent fetches listings, summarizes key details, scores each role against the candidate profile, and logs the results to Google Sheets, highlighting the top 5 matches for client outreach.
People and teams who want faster, objective job discovery and evaluation.
Saves time by automating discovery and providing concise role summaries.
Pre-screens roles for clients to speed up shortlisting.
Systematically identifies suitable opportunities to guide clients.
Creates a transparent, data-driven pipeline for multiple roles.
Builds ready-to-submit candidate lists with scoring.
Tests API integrations and iterates on prompts and prompts-based scoring.
Key tools that the AI agent works with and what it does inside each.
Fetches real-time job listings and details for processing.
Generates summaries, extracts requirements, and assesses fit.
Logs results, scores, and review notes for centralized access.
Orchestrates the AI agent workflow and data flow between services.
Practical scenarios where the AI agent adds clear value.
Common questions about setup, capabilities, and limitations.
The AI agent primarily queries Adzuna's job API to retrieve current postings and details. It uses GPT-3.5 to summarize descriptions and extract key requirements, then scores opportunities against candidate profiles. Personal data from users is only stored within the central Google Sheets log for review and is subject to your data handling practices. The architecture is designed to be extensible, so additional sources can be added by integrating new adapters. Access controls and credentials are managed by your configuration to protect sensitive information.
You provide your Adzuna API key and OpenAI credentials in the setup flow. The AI agent securely references these keys to fetch listings and perform prompts-based processing. You can rotate keys as needed and test connections through the onboarding steps. No credentials are embedded in the agent; they are accessed at runtime via secure storage within your environment.
The current implementation uses Adzuna as the primary source, but the design is modular. You can add adapters for additional sources by creating nodes that fetch data, normalize formats, and feed them into the same processing and scoring pipeline. Each added source will follow the same summary and scoring workflow to maintain consistency. This keeps a single, unified review surface in Google Sheets.
GPT-3.5 generates summaries guided by carefully crafted prompts to capture the most relevant details. While highly useful, summaries are not infallible and should be reviewed, especially for nuance like seniority or soft skills. You can tune the prompts and scoring thresholds to balance conciseness with completeness. The system is designed to flag uncertain cases for human review.
Scores are computed from a weighted mix of factors: essential skills, location alignment, seniority, and alignment with stated preferences. The weights are configurable so you can reflect your priorities. The results in Google Sheets include the raw scores and a narrative justification for transparency. This makes it easy to adjust the scoring as requirements evolve.
Data can be refreshed on demand or scheduled at regular intervals (e.g., hourly). The agent fetches fresh listings, re-generates summaries, re-scores, and updates the central log accordingly. Scheduling is configurable in the workflow, allowing you to balance freshness with API rate limits and cost. You can also trigger refreshes manually for urgent searches.
Yes. The primary repository for results is Google Sheets, which can be shared and exported. The integration writes structured rows with scores, summaries, and review flags for each listing. You can export to CSV from Sheets for use in other tools, or connect Sheets to dashboards for ongoing monitoring. If needed, you can implement additional export formats through the workflow.
Continuously fetches listings from Adzuna, summarizes each listing with GPT-3.5, scores fit against candidate profiles, and logs results to Google Sheets for review.
