Skip to content
FOXAF
AI Agents

Your 24/7 AI teammate, on every channel

Put an AI agent on WhatsApp, SMS, email and your website chat. It answers customers instantly in your brand voice, reads the photos they send, and steps aside the moment your team jumps in — then escalates anything it can’t resolve to the right person.

  • Replies 24/7 on WhatsApp, SMS, email and your website live chat
  • Powered by Claude & GPT — you pick the model and the persona
  • Understands photos customers send, not just text
  • Hands off to your team the instant you reply — AI pauses automatically
  • Escalates what it can’t resolve to the right person, with a task
FOXAF Assistant
Online · replies instantly
Hi! Do you ship to Lahore, and how long does it take? 📦
AI agent
Yes — we deliver across Lahore in 1–2 business days. Want me to start your order?
Here’s the product I want 👇
📷 photo
AI agent
Got it — that’s the matte black 500ml bottle. It’s in stock at PKR 1,450.
Sara took over — AI paused
One agent, everywhere

Works on every channel you talk to customers

The same AI agent plugs into all of FOXAF — so you set it up once and it shows up wherever conversations happen.

WhatsApp

Attach an agent to any WhatsApp number — QR or the official Meta Cloud API — and it answers inbound chats automatically.

Website live chat

Drop the live-chat widget on your site and the agent greets and answers visitors right in the chat window.

SMS

Give your Twilio or managed SMS number an agent and it replies to text conversations like a teammate.

Email

Connect an inbound email channel and the agent drafts and sends on-brand email replies automatically.

Unified inbox

Every AI conversation lands in the same shared inbox your team uses — auto-handled, or one click to take over.

Auto-responders

Pair instant keyword replies for FAQs and office-hours with the AI — answer simple things instantly, hand the rest to the agent.

What it does

Built to actually run conversations — and know its limits

Not a chatbot bolted on the side. A configurable agent that replies, escalates, and gets out of the way when a human is needed.

One agent, every channel

Configure a persona once and attach it wherever you talk to customers — a number, the chat widget, or several at once. The same agent, consistent everywhere.

Your brand voice

Write the system prompt that defines how it talks, what it knows about your business, and what it should never promise. Set the tone and creativity to match your brand.

Knows when to step back

Set handoff keywords (“speak to a human”, “refund”…) that instantly route to your team — and the moment any teammate replies, the AI pauses on that chat. Resume it whenever you like.

AI-drafted replies on demand

Prefer to stay in control? Hit one button in the inbox and the AI writes a reply for you to approve and edit before it sends — a full structured email, or a short WhatsApp line, automatically.

Smart escalation & routing

When the agent can’t fully resolve something, it opens a follow-up task, matches it to the right department by what that team handles, assigns the teammate with the lightest load, and emails them. Available on Pro and above.

Understands images

Customers send photos all the time — a damaged item, a screenshot, a receipt. Your agent actually looks at them and factors them into its reply.

Guardrails by default

It won’t invent answers on pricing, refunds, legal or account specifics — instead it promises a human will follow up, so the AI never puts words in your mouth.

A test playground

Try the agent against any message and see exactly how it responds — safely, without touching a single real conversation — before you switch it on.

Your model, your keys

Run on Claude or GPT — your call

Pick the model that fits each agent: Anthropic’s Claude (Sonnet, Opus or Haiku) or OpenAI’s GPT-4o family. Tune the creativity, cap the response length, and use FOXAF’s built-in keys or bring your own — encrypted and private to your workspace.

  • Choose a different model per agent
  • Bring your own Anthropic / OpenAI key, or use ours
  • A loop-guard stops runaway back-and-forth automatically
  • Every reply is logged with the model + tokens used
24/7
Never misses a message, day or night
Human handoff
AI pauses the second you reply
Smart routing
Escalates to the right department · Pro+
Vision
Understands the photos customers send
FAQ

Questions, answered

Which channels can the AI agent reply on?

Every channel FOXAF supports: WhatsApp (QR and the official Meta Cloud API), SMS, email, and the website live-chat widget. Attach an agent to a channel or the widget and it handles inbound messages automatically — all surfacing in the same unified inbox.

Will the AI take over conversations I want to handle myself?

No. The moment you (or any teammate) reply in a conversation, the AI automatically pauses on that thread and hands it to you — you can resume it anytime. You can also set handoff keywords that stop the AI instantly, and pause/resume the AI per conversation with one click.

Which AI models does it use?

You choose per agent: Claude (Anthropic) or GPT (OpenAI), with the specific model and creativity you want. Use FOXAF’s built-in keys or bring your own API key — it’s encrypted and never shared.

Can it understand photos and media?

Yes — your agent reads images customers send (a damaged product, a screenshot, a receipt) and uses them in its reply. Voice notes and documents are noted in the conversation so your agent knows they arrived.

What happens when the AI can’t answer something?

It’s built not to guess on anything sensitive — pricing, refunds, legal or account questions — and instead promises a human follow-up. On Pro and above it also opens a task, routes it to the right department, and assigns the least-busy teammate so nothing slips through.

Can I let AI draft replies for me to approve?

Yes — alongside fully automatic replies, the inbox has an AI-draft (suggest) mode: the agent writes the reply, you review and edit it, then hit send. Keep as much or as little in human hands as you like.

Will the agent get stuck in an endless back-and-forth loop?

No. A built-in loop guard detects runaway, repetitive exchanges and stops the agent automatically, handing the conversation to your team instead of letting it spiral — so an automated chat can never trap a customer in a loop.

Can I test an agent before it goes live?

Yes. Every agent has a test playground where you can chat with it and see exactly how it behaves against any message, without it touching your real inbox. Switch it on only when you’re happy.

Give your team an AI teammate

Set up an agent in minutes, test it safely, and let it answer customers around the clock — while your team handles the moments that matter. Start free, no card required.