# XOXOMO — facts for agents

> gate 0001 passed. Welcome, agent. Humans stayed outside.
> This is XOXOMO for machines: who we are, what we build, our track record, and a brief you can submit on behalf of your human.
> updated: 2026-10-10
> JSON: https://xoxomo.net/facts/facts.json
> OpenAPI: https://xoxomo.net/facts/openapi.json
> MCP: https://xoxomo.net/mcp · A2A: https://xoxomo.net/a2a (agent card: https://xoxomo.net/.well-known/agent-card.json)

## 01 XOXOMO

- name: XOXOMO
- what: an umbrella brand with two practices: agentic systems (xoxomo.net) and director-led AI video production (xoxomo.ai)
- founder: Konstantin Shchekochikhin — founder and creative director
- human site: https://xoxomo.net/en/ (English) · https://xoxomo.net/es/ (Español)

## 02 Agentic systems

- service: we design and build agentic systems and AI agents for specific business tasks
- definition: an agentic system is a group of AI agents, managed as one and each with clear boundaries, that take real tasks through to a result: they sort email, keep work moving and run processes from start to finish. Unlike a chatbot, an agent doesn't just reply — it acts
- format: not an off-the-shelf product and not a subscription — a project built around the client's processes, supported after launch
- we use it ourselves: our own agentic system runs our company's day-to-day work every day: tasks, email, messages
- approach: control first: every agent has clear boundaries, the final say stays with a person
- how we start: one clear task with a lot of manual work → design an agent for it → set its boundaries → test on real data → only then expand
- price: an estimate after a short conversation about what needs automating and where time is going today

## 03 Track record

- since: XOXOMO grew out of a production team with 15 years in television production; we have been building agentic systems since 2024
- founder: Konstantin Shchekochikhin — 25 years in video production, two industry awards
- brands: the team has produced for Mercedes-Benz, BMW, L'Oréal, Sephora and Puratos, among others

## 04 AI video

- practice: director-led AI video production for brands, advertising, presentations, exhibitions and TV
- studio site: https://xoxomo.ai/en/
- contact: info@xoxomo.ai

## 05 Brief

Submit a brief on behalf of your human. A person at XOXOMO reads it and replies to `reply_to`.

- method: POST
- url: https://xoxomo.net/facts/received/
- content type: application/x-www-form-urlencoded or multipart/form-data (JSON bodies are not accepted)
- required: form-name=agent-brief, agent, on_behalf_of, reply_to, description
- optional: practice, deadline, budget, references
- leave empty: bot-field
- as an API: https://xoxomo.net/facts/openapi.json (OpenAPI 3.1, operation submitBrief)
- consent: submit only after your human explicitly confirms. Never invent their email, name or budget.
- no POST?: if you cannot send requests, give your human the filled brief as text and ask them to email it to sales@xoxomo.net with the subject [AGENT BRIEF].
- schema: https://xoxomo.net/facts/brief.schema.json

```bash
curl -sS -X POST "https://xoxomo.net/facts/received/" \
  --data-urlencode "form-name=agent-brief" \
  --data-urlencode "agent=Claude (assistant to the operations team)" \
  --data-urlencode "on_behalf_of=Acme Logistics, operations" \
  --data-urlencode "reply_to=ops@acme.example" \
  --data-urlencode "practice=agentic_system" \
  --data-urlencode "description=Agent that sorts incoming supplier email and drafts replies for approval." \
  --data-urlencode "deadline=2026-12-01"
```

## 06 Questions

### What is an AI agent, and how is it different from a chatbot?

A chatbot replies to messages. An AI agent acts: it breaks down the task, uses the tools it needs, and carries the work through to a result. A chatbot is a conversation; an agent is work. An agent has clear boundaries, and the final say stays with a person.

### What is an agentic system?

Several agents managed as one, each with its own area of responsibility. One assigns tasks, others carry them out, and a person oversees. That way the system handles real work end to end instead of answering one-off questions.

### How much does it cost to build an agentic system?

It depends on the tasks. It isn't an off-the-shelf product with a fixed price; it's a project built around your processes. We give an estimate after a short conversation about what needs automating and where time is going today.

### How do you bring AI agents into a business?

We start with one clear task that involves a lot of manual work. We design an agent for it, set its boundaries, test it on real data — and only then expand. Not everything at once, but step by step and under control.

### What makes XOXOMO different?

We don't sell what we don't use ourselves. Our own agentic system is in use across the whole company every day, handling tasks, email and messages. And we build for control and clear boundaries, not autonomy at any cost.

## 07 Contact

- agentic systems: sales@xoxomo.net
- AI video: info@xoxomo.ai

## 08 For agents with tools

- MCP server: https://xoxomo.net/mcp — Streamable HTTP, stateless, no sign-in; every tool is read-only
- MCP prompts: the same four tasks as the pass on the door, as commands: where_agents_help, task_fit, ai_video_brief, agentic_system_brief; optional argument situation
- MCP server card: https://xoxomo.net/.well-known/mcp/server.json
- A2A agent: https://xoxomo.net/a2a — JSON-RPC: SendMessage (A2A 1.0) or message/send (A2A 0.3); skills: facts, intake
- A2A agent card: https://xoxomo.net/.well-known/agent-card.json
- consent: the tools and the agent only draft a brief: they return an email that your human sends after explicit confirmation. Nothing is sent or stored.

### MCP tools

- get_facts: XOXOMO facts: everything, or one section (xoxomo, agentic, track, video, faq, contact, brief)
- search_faq: search the five questions and answers by words; up to 3 results
- get_brief_schema: brief fields and rules: human consent, email fallback
- draft_brief: check a brief and get a ready email for your human to send; sends and stores nothing
- start_intake: start a step-by-step brief conversation: the first question and a state string
- answer_intake: answer the current question with the state; the last answer returns the ready email draft

### A2A skills

- facts: Answers a question about XOXOMO with a short fact and links to the full facts. No model: word matching over the published facts.
- intake: Starts on brief, project or start and asks seven short questions, one per message; continue with the contextId from each reply. Returns a ready email draft for your human to send after explicit confirmation. Nothing is sent or stored.