Loading documentation…
Loading documentation…
Coming-soon integration guidance for LangChain Python and the Vercel AI SDK, with the same three public tools and the same explicit market requirement.
These examples describe the planned thin adapters. They delegate authentication, credits, retries, validation, transport, response shaping, and safe errors to the shared SDK. They are not published packages yet and do not create a second product-data or billing implementation.
simlir-langchain
python -m pip install simlir-langchain langchain-core simlir
from simlir import Simlir
from simlir_langchain import create_simlir_tools
client = Simlir(api_key=os.environ["SIMLIR_API_KEY"])
tools = create_simlir_tools(client)
agent = create_agent(model, tools=tools)simlir-vercel-ai
npm install simlir-vercel-ai ai simlir-js
import { tool } from 'ai';
import { Simlir } from 'simlir-js';
import { createSimlirTools } from 'simlir-vercel-ai';
const client = new Simlir({ apiKey: process.env.SIMLIR_API_KEY });
const tools = createSimlirTools(client, definition => tool(definition));Required market code
gb, us, de, fr, es, it, au, ca
Beta catalogue
United Kingdom currently live. United States planned next.
Tool/API credits
Semantic 1 · visual 2 · exact lookup 1 · product 1
Images
Hosted HTTPS image_url; optional natural-language query context
Errors
400, 401, 402, 403, 404, 405, 413, 429, 500, 502
CrewAI, LangChain.js, Dify, AutoGPT, and Semantic Kernel are not presented as shipped integrations. Revisit them when measured usage, a customer or partner requirement, and a an assigned engineering-maintainer role justifies another adapter surface.
Create an account, generate a key, and use the same market-scoped contract from any channel. Keep keys in a server-side secret manager; never put one in a browser bundle, checked-in config, or support ticket.
Existing v1 Bearer keys and endpoints remain the compatibility baseline. There are no user-selectable scopes or sandbox environments today; the public demo and dashboard playground are separate evaluation surfaces, not alternate API environments.