Question 1

What is Kopern?

Accepted Answer

Kopern is an AI agent builder platform that lets you create, test, and deploy production-grade AI agents. You can configure custom tools with sandboxed execution, connect GitHub repos, build multi-agent teams, validate with deterministic grading pipelines, and deploy agents via embeddable widgets, webhooks, Slack bots, and MCP endpoints.

Question 2

Which AI models does Kopern support?

Accepted Answer

Kopern supports Anthropic Claude (Opus 4.6, Sonnet 4.6, Haiku 4.5), OpenAI GPT (GPT-4o, GPT-4o-mini), Google Gemini (2.5 Flash), Mistral AI (Mistral Large, Codestral, Nemo), and local models via Ollama. Each agent can be configured with a different model, temperature, and thinking level.

Question 3

What is agent grading?

Accepted Answer

Agent grading lets you create test suites with specific criteria to automatically validate your agent's behavior. Kopern supports 6 criterion types: output match, schema validation, tool usage, safety checks, custom scripts, and LLM-as-judge evaluation. Results are persisted for tracking improvement over time.

Question 4

How do I deploy an agent on my website?

Accepted Answer

Use the embeddable chat widget. Add a single <script> tag to your website with your API key. The widget renders in a Shadow DOM for CSS isolation, streams responses via SSE, supports markdown, and is mobile-responsive. Configure welcome message, position, and allowed origins from the dashboard.

Question 5

Can I connect my agent to Slack?

Accepted Answer

Yes. Install the Slack bot from the Connectors dashboard using OAuth. Once connected, your agent responds to @mentions and direct messages in Slack. It supports threaded conversations and adds reaction checkmarks to processed messages.

Question 6

What are webhooks used for?

Accepted Answer

Inbound webhooks let external services (n8n, Zapier, Make, custom apps) send messages to your agent via REST API and get JSON responses. Outbound webhooks fire when your agent completes actions (message sent, tool called, session ended), letting you trigger downstream workflows. HMAC-SHA256 signing is supported for security.

Question 7

Can I connect my own GitHub repositories?

Accepted Answer

Yes. Connect your GitHub repos via OAuth and your agents get built-in read_file, search_files, and github_write tools, plus the repo tree and README injected into context. Agents can read, search, and write to your codebase during conversations.

Question 8

What is MCP (Model Context Protocol)?

Accepted Answer

MCP is an open standard for connecting AI agents to external tools and data sources. Kopern lets you expose any agent as an MCP server with API key authentication, compatible with Claude Code, Cursor, and any MCP client. Other applications can send messages to your agent via the standardized protocol.

Question 9

What is AutoResearch?

Accepted Answer

AutoResearch is Kopern's optimization lab with 6 modes: AutoTune (auto-optimize prompts), AutoFix (diagnose failing tests), Stress Lab (adversarial testing), Tournament (prompt comparison), Distillation (transfer to cheaper models), and Evolution (genetic prompt optimization). It automatically improves your agent's performance.

Question 10

How does multi-agent orchestration work?

Accepted Answer

Kopern supports agent teams with three execution modes: parallel (all agents run simultaneously), sequential (agents run one after another with shared context), and conditional (routing based on input rules). Each agent in a team keeps its own tools, skills, and configuration. Pipelines chain agents sequentially with input/output mapping.

Question 11

Is Kopern free to use?

Accepted Answer

Kopern offers a free Starter tier with 3 agents and 100K tokens per month. Pro ($79/mo) adds grading, teams, and connectors. Usage-based billing is available for pay-per-token needs. Enterprise ($499/mo) provides unlimited everything with priority support.

Question 12

What languages does Kopern support?

Accepted Answer

The Kopern interface is available in English and French. Agents themselves can be configured to respond in any language supported by the underlying LLM model.

Question 13

What is the Workflow Quality Monitor?

Accepted Answer

Kopern Monitor runs 18 standardized test prompts across 6 quality criteria (reasoning depth, instruction following, consistency, latency, edge cases, output quality) against any LLM model or API endpoint. It compares scores against baselines for 29 models and generates actionable improvement recommendations.

Question 14

Can I test my own AI endpoint?

Accepted Answer

Yes. Both the Grader and Monitor support endpoint mode — connect any HTTP endpoint that accepts messages and Kopern will send test prompts, evaluate responses, and score quality. Supports OpenAI, Anthropic, n8n, and custom API formats with authentication (Bearer, API key header/query).

Question 15

Does Kopern support social media?

Accepted Answer

Kopern includes a Bluesky AT Protocol connector with 9 tools for social media management: create posts, threads, read feeds, track metrics, search mentions, reply, delete posts, get profiles, and list connected accounts.

Blog

Why We Built a Workflow Quality Monitor (And What We Found)

Making AI Agents Accessible: Our MCP Journey from 0 to 32 Tools

How We Built a Production-Grade AI Agent Grading System

Building Kopern: From Zero to 121 Features in 3 Weeks