When you need to hire developers in Latin America, you can point the AI tool you already use — Claude, ChatGPT, Codex, or Cursor — at Get on Board’s Talent Database through the MCP Server. Your agent searches real, opted-in profiles from a plain-language brief, compares safe candidate summaries, and hands you Get on Board profile links. You review, unlock contact details with credits, and run the process in Get on Board — the offer and the contract are yours to close, on whatever payroll or contract setup you already use. This article is for recruiters and company teams sourcing tech talent in Chile, Colombia, Mexico, Peru, Argentina, and the rest of Spanish-speaking Latin America.
Three ways companies use AI to recruit, and how they differ
Most “AI recruiting” offers for Latin America fall into one of three models:
| Model | How it works | Trade-off |
|---|---|---|
| Closed vendor agents | A proprietary agent searches the vendor’s own pool and makes decisions inside a black box | You can’t choose your tool, inspect the reasoning, or take the workflow with you |
| DIY enrichment chains | You wire together scrapers and enrichment services (Apollo, Lusha, ZoomInfo) to guess emails for scraped profiles | Contacts are guessed, not given; bounce rates and spam complaints are yours to manage |
| Open protocol + opted-in marketplace | Your own agent connects through an open standard (MCP) to a pool of professionals who chose to be found | The agent only reads what candidates agreed to share; contact data stays gated |
Get on Board takes the third path: you bring the AI tool you prefer, and it works with an opted-in LatAm talent pool — profiles written and maintained by the professionals themselves, with verified GitHub and LinkedIn accounts — through an open protocol. There is no proprietary agent to adopt and no enrichment chain to orchestrate. It is the same principle the platform has always run on: remove the intermediary, agency or enrichment chain alike, between you and the candidate. For a deeper comparison of enrichment and scraping tools versus a marketplace, see Tools to reach tech candidates in Latin America.
Step 1: Write a hiring brief your agent can act on
An agent is only as good as the brief you give it. Describe the role in the same plain language you would use with a colleague:
- Role and seniority (e.g. senior backend developer).
- Stack and must-have skills (e.g. Python, AWS, event-driven systems).
- Country or time zone constraints, and remote preference.
- English level, salary range, availability.
You don’t need filter IDs or query syntax — the Talent Database search resolves requirements written in natural language, in English or Spanish.
Step 2: Connect your AI tool to the MCP Server
Paste https://www.getonbrd.com/mcp into any compatible MCP client and approve access with your Get on Board recruiter account through the browser. Setup is URL-only: no API keys, no tokens to copy. Follow the guide for your client:
Access tokens last 2 hours, so re-authorizing during a long session is normal. The connection is available to recruiter team members with an accepted company membership.
Step 3: Let the agent search, filter, and shortlist
With the connection active, ask your agent to search Talent Database from your brief. A practical loop looks like this:
- Search: the agent runs
search_talentwith your brief and gets back candidate summaries — country, seniority, years of experience, skills, English level, and a recruiter-oriented summary. No contact data is included. - Post-filter: ask the agent to discard results that miss a hard requirement and explain why each remaining candidate fits.
- Read deeper: for the strongest candidates, the agent calls
talent_profileto review professional and academic background, skills, and quiz traits before you open a single tab. - Anchor on a job: if the role is already published on Get on Board, the agent can read its description and questionnaire with
list_jobsandjob_detailsand use them as the brief.
Example prompt:
Search Get on Board for senior React developers in Colombia or Peru with
English B2+, at least 5 years of experience, and fintech background.
Discard anyone without recent frontend work, then rank the top 5 with reasons.
Step 4: Unlock, invite, and hire in Get on Board
Every result includes a Get on Board profile link. From there, the workflow continues in the web app, where your team’s process and history live:
- Review the full profile and any fields not included in the agent’s view.
- Unlock contact details with Talent Database credits — unlocking always happens in the web app, never through the agent.
- Invite the professional to one of your open processes.
- If candidates are already applying to your job, Superpower AI adds match scores and AI summaries to speed up applicant review.
What the agent never sees
Professionals own their data on Get on Board, and they opted in to be discoverable. The MCP Server enforces that ownership: it never returns emails, phone numbers, WhatsApp, CV files, or social profile links to any AI tool. Your agent works with summaries and profile links only; the decision to access contact information stays with a person, in the web app, gated by credits. Agents also cannot unlock candidates, send invitations, or apply to jobs — searching and reading is the full scope today.