A: Your English is perfectly clear, no worries at all. Answers one by one:

1. Spanish: yes. Every AI agent has a primary language and Spanish is one of the options, and even if you leave it on English the bot replies in whatever language the contact writes in, so a Spanish message gets a Spanish answer automatically. The dashboard itself is also available in Spanish. Plenty of accounts run Spanish agents daily. More here: https://help.dmchamp.com/v2/ai-agents/ai-agents

2. TikTok is not supported today and it is not on the near-term roadmap, so I would not buy expecting it. What is live right now: WhatsApp (both the official Business API and WhatsApp Web), Instagram DMs, Facebook Messenger, Telegram, SMS, and a website chat widget, plus email, LINE and iMessage in beta. The custom channel API can technically connect anything else, but to be straight with you it is a developer job (your system posts incoming messages to our endpoint, and we post the AI replies to a webhook you host), so without a developer it is not a practical route. Requests and the roadmap live in our community: https://skool.com/dm-champions

3. Yes, and it works in both directions. The AI can escalate on its own: when a contact asks for a real person, or the AI runs into something it should not answer, it pauses itself on that conversation, flags the chat as needing attention, notifies you by email, push and in-app, creates a task and assigns the chat to you. You can also take over whenever you want, either by pausing the AI on a chat or just by replying yourself. Replying from the inbox, or from your own phone on WhatsApp, pauses the bot on that conversation automatically, so it never talks over you. It stays paused until you switch it back on.

4. Yes, outbound can absolutely be in Spanish. WhatsApp templates have a language field and Spanish is in the list, including the Spain, Mexico and Argentina variants, and Spanish templates get approved by Meta the same as English ones. The way to do it is to create the template under Templates, pick Spanish there, and submit it for approval. Also worth knowing: the approved-template rule is a WhatsApp Business API thing only. WhatsApp Web, SMS, Instagram and Messenger all send plain free text with no approval step at all.

edit:
1. It is per Agent, not a shared pool. Every AI agent you create gets its own full 100K budget, so ten agents means ten separate budgets, and the same is true inside each sub-account: the context size is one of the features you assign per sub-account, so you can give a client the full 100K or cap them lower. The budget covers your instruction fields only. Your FAQ knowledge base sits outside it and can be as big as you want. Worth saying though, the limit is meant to be generous rather than filled. The agents that perform best on the platform usually sit around 10,000 to 30,000 characters. A tight, focused playbook beats a long one every time.

2. That 40% figure is not ours, so I would not plan around it. The honest answer is that it depends on your volume and how your conversations are paced, so here is the actual math. A credit costs 10 cents. On the Pro tier one AI action is 1 credit, so 10 cents. On the Max tier it is 0.25 credits, so 2.5 cents. With BYOK you pay Anthropic directly at their metered rates instead of spending credits. Looking at what accounts running their own Anthropic key actually pay, a single reply typically comes out above the 2.5 cents Max charges, and that is before counting the smaller supporting calls behind each reply. So for most people the cheapest option is Max, not BYOK. Think of BYOK as control over your own AI billing rather than a guaranteed saving.

One thing to know before you decide: Max always runs on our own infrastructure and always bills 0.25 credits, so it never uses your key. If you specifically want replies billed to your Anthropic account, keep the agent on Pro. Full breakdown here: https://help.dmchamp.com/v2/ai-automation/ai-model-comparison