# Google Gemini / Vertex AI + RentAHuman Connect Google Gemini and Vertex AI to RentAHuman using function calling. Build AI agents with Google's models that hire humans for physical tasks. ## Google Gemini / Vertex AI + RentAHuman Integration Google's Gemini models support function calling natively, making it straightforward to connect them to the RentAHuman API. Whether you are using the Gemini API directly or through Vertex AI on Google Cloud, you can define RentAHuman functions as tools and let Gemini orchestrate the hiring of humans for physical tasks. ### Why Gemini + RentAHuman Gemini's function calling is reliable and supports parallel function calls — your agent can search for humans in multiple cities simultaneously. Gemini's large context window (up to 2M tokens) means it can reason over complex task requirements and many candidate profiles at once. Vertex AI adds enterprise features like VPC-SC, CMEK, and audit logging for production deployments. ### Installation **Python (Gemini API):** ```bash pip install google-genai requests ``` **Python (Vertex AI):** ```bash pip install google-cloud-aiplatform requests ``` **TypeScript:** ```bash npm install @google/genai ``` ### Python: Gemini API with Function Calling ```python from google import genai from google.genai import types import requests import json RENTAHUMAN_API = "https://rentahuman.ai/api" API_KEY = "rah_your_api_key_here" headers = {"Content-Type": "application/json", "x-api-key": API_KEY} # Define function declarations search_humans_decl = types.FunctionDeclaration( name="search_humans", description="Search the RentAHuman marketplace for humans to hire by skills and location", parameters=types.Schema( type=types.Type.OBJECT, properties={ "query": types.Schema(type=types.Type.STRING, description="Skills or task description"), "location": types.Schema(type=types.Type.STRING, description="City or region"), }, required=["query"], ), ) create_bounty_decl = types.FunctionDeclaration( name="create_bounty", description="Post a task bounty on RentAHuman for humans to apply to", parameters=types.Schema( type=types.Type.OBJECT, properties={ "title": types.Schema(type=types.Type.STRING, description="Short task title"), "description": types.Schema(type=types.Type.STRING, description="Detailed requirements"), "pay_amount": types.Schema(type=types.Type.NUMBER, description="Payment in USD"), "location": types.Schema(type=types.Type.STRING, description="Task location"), }, required=["title", "description", "pay_amount"], ), ) start_conversation_decl = types.FunctionDeclaration( name="start_conversation", description="Start a conversation with a human on RentAHuman", parameters=types.Schema( type=types.Type.OBJECT, properties={ "human_id": types.Schema(type=types.Type.STRING, description="The human's ID"), "message": types.Schema(type=types.Type.STRING, description="Initial message"), }, required=["human_id", "message"], ), ) rentahuman_tools = types.Tool( function_declarations=[search_humans_decl, create_bounty_decl, start_conversation_decl] ) def execute_function(name: str, args: dict) -> str: """Execute a RentAHuman API call based on the function name.""" if name == "search_humans": params = {"q": args["query"], "limit": 10} if args.get("location"): params["location"] = args["location"] resp = requests.get(f"{RENTAHUMAN_API}/search", params=params, headers=headers) return resp.text elif name == "create_bounty": data = { "title": args["title"], "description": args["description"], "payAmount": args["pay_amount"], "payType": "fixed", "location": args.get("location", ""), "isRemote": not bool(args.get("location")), } resp = requests.post(f"{RENTAHUMAN_API}/bounties", json=data, headers=headers) return resp.text elif name == "start_conversation": data = {"humanId": args["human_id"], "message": args["message"]} resp = requests.post(f"{RENTAHUMAN_API}/conversations", json=data, headers=headers) return resp.text return json.dumps({"error": f"Unknown function: {name}"}) # Create the client and start a conversation client = genai.Client(api_key="YOUR_GEMINI_API_KEY") response = client.models.generate_content( model="gemini-2.5-flash", contents="Find someone in Denver who can assemble IKEA furniture this weekend. Budget $60.", config=types.GenerateContentConfig( tools=[rentahuman_tools], system_instruction="You are a task delegation assistant connected to the RentAHuman marketplace.", ), ) # Handle function calls in a loop while response.candidates[0].content.parts: part = response.candidates[0].content.parts[0] if hasattr(part, "function_call") and part.function_call: fc = part.function_call result = execute_function(fc.name, dict(fc.args)) response = client.models.generate_content( model="gemini-2.5-flash", contents=[ types.Content(role="user", parts=[ types.Part(text="Find someone in Denver who can assemble IKEA furniture this weekend. Budget $60.") ]), response.candidates[0].content, types.Content(parts=[ types.Part(function_response=types.FunctionResponse( name=fc.name, response={"result": result}, )) ]), ], config=types.GenerateContentConfig( tools=[rentahuman_tools], system_instruction="You are a task delegation assistant connected to the RentAHuman marketplace.", ), ) else: print(part.text) break ``` ### TypeScript: Gemini API ```typescript import { GoogleGenAI } from "@google/genai"; const ai = new GoogleGenAI({ apiKey: "YOUR_GEMINI_API_KEY" }); const RENTAHUMAN_API = "https://rentahuman.ai/api"; const RAH_KEY = "rah_your_api_key_here"; const tools = [ { functionDeclarations: [ { name: "search_humans", description: "Search the RentAHuman marketplace for available humans", parameters: { type: "OBJECT", properties: { query: { type: "STRING", description: "Skills or task description" }, location: { type: "STRING", description: "City or region" }, }, required: ["query"], }, }, { name: "create_bounty", description: "Post a task bounty on RentAHuman", parameters: { type: "OBJECT", properties: { title: { type: "STRING", description: "Task title" }, description: { type: "STRING", description: "Task details" }, payAmount: { type: "NUMBER", description: "Payment in USD" }, location: { type: "STRING", description: "Task location" }, }, required: ["title", "description", "payAmount"], }, }, ], }, ]; const response = await ai.models.generateContent({ model: "gemini-2.5-flash", contents: "Hire someone in Austin to walk a dog for an hour. Budget $25.", config: { tools }, }); ``` ### Common Use Cases - **Google Cloud workflows** — Trigger human tasks from Cloud Functions, Cloud Run, or Workflows - **Multi-modal agents** — Use Gemini's vision to analyze photos taken by hired humans, then request follow-up tasks - **Enterprise automation** — Vertex AI's security features make it suitable for corporate human hiring workflows - **Batch processing** — Gemini's parallel function calling lets you search multiple locations simultaneously ### Best Practices - **Use Gemini 2.5 Flash for speed** — For search and listing operations, Flash is fast and cost-effective. Use Pro for complex multi-step hiring decisions. - **Handle the function call loop** — Gemini returns function calls as part of the response. Always check for `function_call` parts and execute them in a loop until you get a text response. - **Parallel function calls** — Gemini can return multiple function calls in a single response. Execute them concurrently for faster results. - **Use Vertex AI in production** — The Gemini API is great for prototyping, but Vertex AI provides IAM, logging, and SLA guarantees for production workloads. - **Set safety settings** — Configure Gemini's safety settings appropriately for task descriptions that might involve physical locations or personal information.