Западный рынок: конкуренты и GTM без Telegram
Дата: 2026-05-03 10:06 UTC
Вопрос исследования
Пользователь спросил:
Что подобного есть на рынке? Если что-то сделает что-то похожее — достаточно моста туда. Как продвигать Trip2G на западный рынок без Telegram?
Короткий вывод
На западном рынке уже есть сильные продукты в отдельных слоях:
- agent memory;
- RAG/vector frameworks;
- enterprise knowledge search;
- docs-to-agent / support bots;
- PKM / second brain;
- MCP directories and tool integration layers.
Но почти никто не занимает ровно позицию:
MCP-native distributed knowledge hub where Markdown, docs, Obsidian, Notion, Drive and other sources become searchable, permissioned, federated memory endpoints for agents.
Для западного рынка не стоит начинать с Telegram. Telegram оставить как optional adapter. Главные западные входы:
- GitHub/docs → MCP documentation server.
- Obsidian → Claude/Cursor memory.
- Notion/Drive → team knowledge hub.
- Slack/Discord/community archive → answer engine.
- Webhooks/cron → safe content backend for AI agents.
Лучшее короткое позиционирование
English
Trip2G turns your Markdown, docs, and team knowledge into a distributed MCP memory network for AI agents.
Альтернативы:
Make your docs agent-readable.
Connect your knowledge to Claude and Cursor.
A federated knowledge layer for Claude, Cursor, Codex, and every MCP-aware agent.
Not another chatbot. A knowledge network your agents can actually use.
Russian working translation
Trip2G превращает markdown, документацию и командные знания в распределённую MCP-память для AI-агентов.
Карта рынка
1. Agent memory infrastructure
Mem0
- URL: https://mem0.ai
- GitHub: https://github.com/mem0ai/mem0
- Category: agent memory layer.
- Similarity: память для AI-агентов.
- Difference: больше SDK/API memory, меньше human-readable knowledge bases, Markdown publishing, federation.
- Trip2G angle:
Mem0 stores agent memory. Trip2G exposes human knowledge as agent memory.
Zep / Graphiti
- URL: https://www.getzep.com
- GitHub: https://github.com/getzep/graphiti
- Category: temporal knowledge graph / memory for agents.
- Similarity: evolving memory / graph for agents.
- Difference: developer infra, не продукт для human-owned knowledge publishing/federation.
- Trip2G angle:
Zep remembers conversations. Trip2G exposes your living knowledge base to agents.
Letta
- URL: https://www.letta.com
- GitHub: https://github.com/letta-ai/letta
- Category: stateful agents / agent memory OS.
- Similarity: agent memory and tool use.
- Difference: runtime/framework for agents, not knowledge publishing/proxy/federation.
- Trip2G angle:
Letta runs stateful agents. Trip2G is an external knowledge substrate those agents can use.
LangMem / LangGraph memory
- URL: https://github.com/langchain-ai/langmem
- URL: https://www.langchain.com/langgraph
- Category: agent memory inside LangChain ecosystem.
- Difference: ecosystem-specific.
- Trip2G angle: framework-agnostic memory via MCP.
Cognee
- URL: https://www.cognee.ai
- GitHub: https://github.com/topoteretes/cognee
- Category: AI memory / knowledge graph / RAG.
- Difference: more ingestion/graph infrastructure.
- Trip2G angle: editable, publishable, federated Markdown knowledge hubs.
2. RAG frameworks and knowledge pipelines
LlamaIndex
- URL: https://www.llamaindex.ai
- Category: RAG/data framework.
- Similarity: connects private data to LLM apps.
- Difference: developer framework; user still builds app/product.
- Trip2G angle: ready-to-use knowledge hub with MCP, Markdown publishing, permissions, federation.
LangChain
- URL: https://www.langchain.com
- Category: agent/RAG framework.
- Trip2G angle: do not fight; Trip2G can be a knowledge source/tool for LangChain agents.
Haystack
- URL: https://haystack.deepset.ai
- Category: production RAG pipelines.
- Trip2G angle: Haystack is infrastructure for engineers; Trip2G is a knowledge layer for users, teams, authors, agents.
Dify
- URL: https://dify.ai
- GitHub: https://github.com/langgenius/dify
- Category: LLM app builder / RAG app platform.
- Similarity: low-code knowledge bases and workflows.
- Difference: centralized app builder, not distributed MCP knowledge network.
- Trip2G angle: Trip2G should be the knowledge backend Dify-like apps can call, not another app builder.
AnythingLLM / Open WebUI / Khoj
- AnythingLLM: https://anythingllm.com
- Open WebUI: https://openwebui.com
- Khoj: https://khoj.dev
- Category: chat with docs / local AI / personal search.
- Trip2G angle: not chat UI, but agent-searchable knowledge network + website + automation layer.
3. Enterprise knowledge search and internal AI
Glean
- URL: https://www.glean.com
- Category: enterprise search / work AI.
- Similarity: search across company apps.
- Difference: enterprise-heavy, expensive, closed.
- Trip2G angle: small teams, indie experts, dev teams; open Markdown, self-hostable/simple, MCP-native.
Guru / Slite / Notion AI / Atlassian Rovo
- Guru: https://www.getguru.com
- Slite: https://slite.com
- Notion AI: https://www.notion.so/product/ai
- Atlassian Rovo: https://www.atlassian.com/software/rovo
- Confluence: https://www.atlassian.com/software/confluence
- Category: company wiki + AI answers.
- Trip2G angle:
They search one corporate workspace. Trip2G federates many knowledge hubs through MCP.
Microsoft Copilot / Google NotebookLM
- Microsoft Copilot: https://www.microsoft.com/en-us/microsoft-365/copilot
- NotebookLM: https://notebooklm.google
- Category: enterprise/personal document reasoning.
- Trip2G angle: persistent hub, automation, web publishing, MCP endpoint, open/federated sources.
4. Docs-to-agent / support bots / website RAG
kapa.ai
- URL: https://www.kapa.ai
- Category: AI support bot for developer docs.
- Similarity: answers technical questions from docs/community sources.
- Difference: mainly Q&A bot, not general knowledge network/memory layer.
- Trip2G angle:
Turn your docs into an MCP server for agents, not just a website chatbot.
DocsBot / Chatbase / Intercom Fin
- DocsBot: https://docsbot.ai
- Chatbase: https://www.chatbase.co
- Intercom Fin: https://www.intercom.com/fin
- Category: docs/support chatbot.
- Trip2G angle: avoid “another chatbot”; emphasize agent-native MCP and user’s own coding agent.
Mintlify / GitBook / ReadMe
- Mintlify: https://mintlify.com
- GitBook: https://www.gitbook.com
- ReadMe: https://readme.com
- Category: docs/developer portal with AI search.
- Trip2G angle:
GitBook/Mintlify publish docs for humans. Trip2G makes docs usable inside Claude/Cursor/Codex through MCP.
5. PKM / second brain
Obsidian
- URL: https://obsidian.md
- Category: local-first Markdown PKM.
- Relationship: strongest companion, not direct competitor.
- Trip2G angle:
Obsidian is where you write. Trip2G is how agents access and publish it.
Notion / Tana / Capacities / Logseq / Roam / Readwise / Mem.ai
- Notion: https://www.notion.so
- Tana: https://tana.inc
- Capacities: https://capacities.io
- Logseq: https://logseq.com
- Roam: https://roamresearch.com
- Readwise Reader: https://readwise.io/read
- Mem.ai: https://mem.ai
- Category: personal/team knowledge tools.
- Trip2G angle: these are sources and companions; Trip2G is access/proxy/publishing/federation layer for agents.
6. MCP ecosystem and tool marketplaces
Anthropic MCP
- URL: https://modelcontextprotocol.io
- Category: protocol standard.
- Relevance: Trip2G should anchor heavily on MCP as “USB-C for AI tools/data”.
Smithery / Glama
- Smithery: https://smithery.ai
- Glama MCP: https://glama.ai/mcp
- Category: MCP server registry/discovery.
- Trip2G angle: distribution partners, not direct competitors. Trip2G should be listed there.
Composio / Pipedream / Zapier MCP
- Composio: https://composio.dev
- Pipedream: https://pipedream.com
- Zapier MCP: https://zapier.com/mcp
- Category: tool/action integration for agents.
- Trip2G angle:
They connect actions. Trip2G stores/version-controls/serves knowledge.
Главное стратегическое следствие
Если западный игрок делает похожую часть, Trip2G не обязан конкурировать напрямую. Часто достаточно сделать bridge/adapter:
- Mem0/Zep/Letta: Trip2G can be a human knowledge source feeding agent memory.
- LangChain/LlamaIndex/Dify: Trip2G can be a retriever/tool/source.
- Glean/Notion/Drive/Slack: Trip2G can index/proxy selected sources.
- Smithery/Glama: Trip2G should use them for discovery.
- Composio/Zapier/Pipedream: Trip2G can integrate as a knowledge/action endpoint.
- Mintlify/GitBook/ReadMe: Trip2G can expose docs as MCP for coding agents.
Как идти на запад без Telegram
Не вести с Telegram
Telegram оставить как optional adapter. Для Western market в hero лучше использовать:
- Obsidian;
- GitHub/Markdown docs;
- Notion;
- Google Drive / Google Docs;
- Slack;
- Discord;
- Linear / Jira;
- Readwise;
- Web/docs crawler;
- MCP clients: Claude Desktop, Claude Code, Cursor, Codex, Windsurf, GitHub Copilot.
Рекомендуемые wedges
Wedge 1: GitHub/docs → MCP docs server
Best first developer wedge.
Hero:
Make your docs agent-readable.
Subheadline:
Connect your GitHub docs folder and give Cursor, Claude Code, and MCP clients grounded access to your real documentation.
Why strong:
- Developers already feel Cursor/Claude hallucinate APIs.
- Docs are often Markdown already.
- Easy public demos with open-source docs.
- Clear distribution through GitHub, HN, DevRel, MCP directories.
Killer demo:
User in Cursor: “Implement OAuth refresh using this SDK.”
Cursor calls Trip2G MCP:
- search("OAuth refresh token SDK")
- note_html("docs/auth/oauth.md")
- similar(...)
Then writes grounded code.
CTA:
Create free docs MCP server.
Paid expansion:
- more docs repos;
- private docs;
- custom domain;
- usage analytics;
- hosted sync;
- premium support;
- docs QA automation.
Wedge 2: Obsidian → Claude/Cursor memory
Hero:
Talk to your Obsidian vault from Claude.
Subheadline:
Trip2G turns your vault into a safe MCP server for Claude, Cursor, and other AI agents.
Why strong:
- Obsidian users understand local-first Markdown.
- They have real notes.
- They like plugins and technical workflows.
- Good early adopters for “knowledge network”.
CTA:
Connect first vault free.
Paid expansion:
- multiple vaults;
- private endpoint;
- hosted sync;
- public/private knowledge split;
- web publishing;
- automation/webhooks.
Wedge 3: Notion / Drive → team knowledge hub
Hero:
One MCP endpoint for your team knowledge.
Subheadline:
Expose selected Notion pages, Drive folders, docs and decisions as a secure MCP knowledge base for Claude, Cursor and team agents.
Why strong:
- Small teams already have knowledge scattered across SaaS.
- Glean is too enterprise/heavy.
- Notion AI is locked inside Notion.
CTA:
Connect one source free.
Paid expansion:
- multiple sources;
- teams/access control;
- scheduled refresh;
- Slack/Discord connectors;
- private hosted hub.
Wedge 4: Slack/Discord archive → answer engine
Hero:
Your community has already answered this. Let agents find it.
Subheadline:
Turn Slack or Discord discussions into curated, AI-queryable knowledge for support, onboarding, and docs.
Why strong:
- Western communities use Slack/Discord more than Telegram.
- Repeated questions are an obvious pain.
- Can generate FAQ/digests from archive.
CTA:
Turn one channel into an AI FAQ.
Paid expansion:
- hosted archive index;
- scheduled refresh;
- moderation workflows;
- community support bot;
- private/public access.
Wedge 5: Safe content backend for AI agents
Hero:
A safe content backend for AI agents.
Subheadline:
Versioned Markdown, scoped tokens, webhooks, cron, MCP tools and human review.
Why strong:
- Agent builders need safe writes, not just vector DB.
- Trip2G webhooks/cron/scoped tokens are rare and valuable.
Messaging:
- trigger agents when notes change;
- schedule recurring agent jobs;
- prevent infinite loops with depth limits;
- let agents write safely with expected hashes;
- use MCP for retrieval.
CTA:
Build your first agent workflow.
Paid expansion:
- hosted bots;
- custom adapters;
- workflow templates;
- high automation limits;
- support.
Где продвигать
Highest priority
- Hacker News: Show HN / technical essays.
- r/ObsidianMD.
- Obsidian Forum / Discord.
- Cursor community/forum.
- Claude / Claude Code communities.
- Anthropic MCP discussions.
- MCP server directories: Smithery, Glama, PulseMCP-like directories.
- GitHub open-source projects with Markdown docs.
- DevTools Twitter/X.
- AI Engineer / Latent Space communities.
- LangChain / LlamaIndex Discords.
Medium priority
- r/Notion.
- Notion consultants.
- Indie Hackers.
- Product Hunt after demos.
- r/selfhosted.
- r/LocalLLaMA.
- r/AI_Agents.
- n8n / Make / Zapier communities.
- DevRel communities.
- Discord admin/community-builder groups.
Lower priority initially
- generic SaaS communities;
- generic productivity communities;
- enterprise KM communities;
- SEO/content marketing groups.
Reason: Trip2G needs technical/agent-aware early adopters first.
30-day Western GTM plan
Week 1
Ship 2 focused landing pages:
/docs-to-mcp/obsidian-mcp
Create public demos:
- one open-source docs MCP endpoint;
- one Obsidian/demo vault MCP endpoint;
- one webhook AI workflow demo.
Create 3 short videos/GIFs:
- Claude asks Obsidian vault;
- Cursor uses docs MCP;
- Webhook AI edits note safely.
Week 2
Outreach:
- 20 Obsidian users;
- 20 OSS maintainers;
- 20 DevRel/API startups.
Submit to MCP directories.
Post in MCP/Claude/Cursor communities.
Week 3
Run white-glove onboarding.
Publish technical essay:
Your docs need an MCP server.
Alternative essays:
- “Knowledge bases are becoming APIs for agents.”
- “RAG is not enough: agents need instructions, permissions, and safe writes.”
- “Markdown is the shared interface between humans and AI agents.”
Week 4
Double down on highest activation segment.
Add paid expansion prompt:
- second source;
- private/team access;
- webhooks/cron;
- hosted sync;
- custom adapter.
Collect 3 case studies.
First 10 experiments
- Obsidian white-glove onboarding: 20 users, first vault free.
- GitHub docs MCP for OSS projects: 20 docs-heavy repos, 3–5 demo endpoints.
- “Make Cursor stop hallucinating your API”: DevRel/API startup outreach.
- MCP directory launch: submit hosted knowledge MCP examples.
- AI docs maintainer tutorial: Trip2G webhooks + Claude.
- Notion consultants partnership: “AI-ready Notion workspace”.
- Discord community FAQ pilot: last 30 days → searchable FAQ/MCP.
- First source free activation funnel: measure second-source conversion.
- Public demo knowledge bases: suggested questions + visible source notes.
- Founder-led technical essays for HN/dev audience.
Outreach drafts
Open-source maintainer
Hey — I noticed your docs are Markdown-heavy and users likely ask Cursor/Claude about your API.
I made a small MCP endpoint concept that lets Claude/Cursor search and read your docs directly instead of guessing.
Would it be useful if your README had: “Use these docs in Cursor/Claude via MCP”?
Happy to set up the first version free.
Obsidian user
I’m testing a tool that turns an Obsidian vault into an MCP server for Claude/Cursor.
The idea: keep writing in Obsidian, but let your AI assistant search and read your notes safely instead of copy-pasting.
Looking for 10 power users to try it free and tell me what breaks.
DevRel/API startup
A lot of your users are probably coding with Cursor/Claude now. The problem is those agents hallucinate APIs unless your docs are available as tools.
Trip2G can expose your docs as an MCP server with search, section retrieval, similar docs, and usage instructions.
Want a quick demo with your actual docs?
AI automation agency
We’re building a safe content backend for AI agents: versioned Markdown, scoped temporary tokens, webhooks, cron jobs, and MCP retrieval.
Useful when agents need to read/write client knowledge without getting full workspace access or overwriting data.
Want to try it on one client workflow?
Pricing / packaging hypothesis
Free
- 1 source.
- Hosted MCP endpoint.
- Basic search/read tools.
- Public or personal private endpoint.
Pro
- Multiple sources.
- Private sources.
- Larger index.
- Custom instructions.
- Webhooks/cron.
- More frequent sync.
- Priority indexing.
Team
- Team permissions.
- Scoped source access.
- Audit/version history.
- Slack/Discord/Drive/Notion connectors.
- Shared MCP endpoints.
- Admin controls.
Builder/API
- Agent write access.
- Scoped temporary tokens.
- Webhook workflows.
- Cron jobs.
- Higher automation limits.
- Custom domains/endpoints.
Важное: messaging hierarchy
Не начинать с
- distributed knowledge base protocol;
- federated knowledge network;
- content infrastructure;
- programmable platform.
Это точные, но слишком абстрактные фразы для acquisition.
Начинать с
- Make your docs agent-readable.
- Turn your Obsidian vault into Claude/Cursor memory.
- Hosted MCP for your knowledge base.
- Safe content backend for AI agents.
- Connect Notion/Drive/docs to Claude via MCP.
Потом объяснять архитектуру
- MCP tools;
- vector search;
- note_html;
- similar;
- instructions;
- scoped tokens;
- webhooks/cron;
- depth protection;
- versioning;
- first source free;
- paid expansion.
Итоговая рекомендация
Для западного рынка лучший старт:
- Primary wedge: GitHub/docs → “Make your docs agent-readable.”
- Secondary wedge: Obsidian → “Your vault as Claude/Cursor memory.”
- Builder wedge: “Safe content backend for AI agents.”
Trip2G не должен выглядеть как Telegram-centric publisher. В западной упаковке это:
Hosted MCP knowledge infrastructure for humans and AI agents.
Или ещё проще:
Connect your knowledge to Claude and Cursor.
Sources / URLs
- Mem0: https://mem0.ai
- Zep: https://www.getzep.com
- Graphiti: https://github.com/getzep/graphiti
- Letta: https://www.letta.com
- LangMem: https://github.com/langchain-ai/langmem
- LangGraph: https://www.langchain.com/langgraph
- Cognee: https://www.cognee.ai
- LlamaIndex: https://www.llamaindex.ai
- LangChain: https://www.langchain.com
- Haystack: https://haystack.deepset.ai
- Dify: https://dify.ai
- Flowise: https://flowiseai.com
- AnythingLLM: https://anythingllm.com
- Open WebUI: https://openwebui.com
- Khoj: https://khoj.dev
- Glean: https://www.glean.com
- Guru: https://www.getguru.com
- Slite: https://slite.com
- Notion AI: https://www.notion.so/product/ai
- Atlassian Rovo: https://www.atlassian.com/software/rovo
- Confluence: https://www.atlassian.com/software/confluence
- Microsoft Copilot: https://www.microsoft.com/en-us/microsoft-365/copilot
- Google NotebookLM: https://notebooklm.google
- kapa.ai: https://www.kapa.ai
- DocsBot AI: https://docsbot.ai
- Chatbase: https://www.chatbase.co
- Intercom Fin: https://www.intercom.com/fin
- Mintlify: https://mintlify.com
- GitBook: https://www.gitbook.com
- ReadMe: https://readme.com
- Obsidian: https://obsidian.md
- Notion: https://www.notion.so
- Tana: https://tana.inc
- Capacities: https://capacities.io
- Logseq: https://logseq.com
- Roam Research: https://roamresearch.com
- Readwise Reader: https://readwise.io/read
- Mem.ai: https://mem.ai
- Model Context Protocol: https://modelcontextprotocol.io
- Smithery: https://smithery.ai
- Glama MCP: https://glama.ai/mcp
- Composio: https://composio.dev
- Pipedream: https://pipedream.com
- Zapier MCP: https://zapier.com/mcp
- Cloudflare Agents: https://developers.cloudflare.com/agents