Sourcing candidates for an open role can require recruiters to move between LinkedIn searches, spreadsheets, job descriptions, candidate databases, and internal communication tools. They need to define the role requirements, find suitable candidates, assess fit, compare profiles, and maintain accurate records throughout the process.
This use case explores how a connected AI automation can turn a plain-language recruiting request into a structured candidate-sourcing workflow. The agent can interpret hiring requirements, source candidates, assess match quality, refine the search when needed, store results in a database, and return a ranked candidate list through Telegram.
The Problem
Sourcing candidates for a role often means a recruiter has to manage multiple tools at once. This can include searching LinkedIn, tracking candidates in spreadsheets, comparing profiles against job requirements, and updating a database as new information becomes available.
Doing this conversationally and on demand from a phone is not something most applicant tracking systems are designed to handle. As sourcing volume increases, recruiters may spend more time switching between systems, manually reviewing profiles, and organising candidate information than focusing on the quality of the hiring process.
The Solution
The Recruiting Orchestrator Agent turns a Telegram chat into a recruiting assistant.
A single orchestrating AI agent receives a plain-language request, interprets the hiring need, creates a structured Ideal Candidate Profile, and delegates work to specialised sourcing, scoring, and database agents. It can refine searches when candidate match quality is low, store information in a Postgres database, and return a ranked candidate list in the same Telegram conversation.
How the Agent Works
- Receive a hiring request. A recruiter can send a message through Telegram describing the role, required skills, experience level, location, or other candidate requirements.
- Interpret the request and create an Ideal Candidate Profile. The orchestrator identifies whether the user wants to source candidates, rank existing candidates, or search the database, then converts the request into structured hiring criteria.
- Source suitable candidates. The workflow sends the job requirements to a dedicated sourcing process that returns structured candidate information.
- Score candidate fit. A separate AI agent compares each candidate against the Ideal Candidate Profile and assigns scores based on role match, skills, location, and experience.
- Refine searches when needed. When the average candidate match quality is low, the system can improve the search criteria and run the sourcing process again.
- Store and return results. Candidate information, scores, and relevant records are saved in the database, while the recruiter receives a ranked summary through Telegram.
Technical Workflow
1. Chat entry point
A Telegram Trigger listens for incoming messages. Every message is handed straight to the Orchestrator Agent, which is the brain of the whole system.
2. ICP parsing and intent routing
The Orchestrator, running on Google Gemini, is given a detailed system prompt: interpret the user’s intent (source candidates, rank existing candidates, or query the database), parse an Ideal Candidate Profile out of free text, and decide which tools to call.
It keeps a 15-message conversation buffer, keyed by Telegram chat ID, and has access to a Think tool for reasoning through query refinements.
3. Candidate sourcing (sub-workflow tool)
When sourcing is needed, the Orchestrator calls a Candidate Sourcing Workflow Tool. This invokes a separate n8n workflow (“Divverse P2 – Google Search Jobs”), passing a job description and desired lead count, and gets back structured candidate objects.
This keeps the actual web-sourcing logic decoupled from the conversational agent.
4. Iterative quality control
If scored matches come back below an approximately 80% average, the Orchestrator is instructed to use the Think tool to refine the search query and re-invoke sourcing, up to three iterations, before settling on a final candidate set.
5. Candidate scoring (sub-agent tool)
Sourced candidates and the parsed ICP are passed to a Candidate Scoring AI Agent (GPT-4.1-mini, with its own memory buffer) exposed as a callable tool.
It runs an explicit, deterministic-style scoring rubric, including role match, skill-weight normalization, location match, and experience threshold, returning each candidate with a roleScore, skillScore, locationScore, experienceScore, and a normalized finalScore out of 100, sorted descending.
6. Database operations (SQL sub-agent)
A third layer, the SQL Agent, also Gemini-powered with its own memory, is exposed as a tool for anything touching the Postgres/Supabase database.
It has three Postgres tools of its own: list tables, fetch column definitions/foreign keys, and execute arbitrary SQL built dynamically via $fromAI. The prompt instructs it to always check schema before writing.
7. Response delivery
Once sourcing, scoring, and storage are complete, the Orchestrator formats a detailed, human-readable summary and a Telegram node sends it back to the same chat.
Technology and Integrations
Built with: n8n, Telegram, Google Gemini, OpenAI GPT-4.1-mini, Postgres or Supabase, and a multi-agent architecture with dedicated sourcing, scoring, and database agents.
Outcome
The Recruiting Orchestrator Agent creates a more flexible and structured candidate-sourcing process.
Instead of moving manually between recruiting tools, recruiters can submit a request through a Telegram conversation and receive a ranked candidate list based on defined hiring requirements. The system can source candidates, assess match quality, refine searches, and store results in a central database while maintaining conversational context throughout the process.
Explore Custom AI Automation for Your Business
The Recruiting Orchestrator Agent is one example of how AI automation can support more structured recruitment and talent-sourcing workflows.
We design and build custom AI agents, automation workflows, internal tools, and connected systems for a wide range of business processes.
From sales, marketing, recruitment, customer support, and reporting to finance, operations, and internal team workflows, each solution is designed around the way your business works.
Request a Quote for a Custom AI Automation Solution to get started.