About AI Smarty

AI Smarty offers businesses highly intelligent ai agents that can perform multiple roles in a company from customer service, sales leads and more. AI Smarty agents never sleep, are infinitely scaleable and use multiple cutting edge LLMs to converse, monitor and guide every conversation!

Contact Info

Follow Us

questions, answered

Common questions about AI customer support agents.

Pricing, GDPR, integrations, security, support. If we missed yours, drop us a line at [email protected].

Product

Q1How is this different from a regular chatbot?

AI Smarty agents are trained on your specific business documents and use a hybrid RAG pipeline that cites sources from your knowledge base. They reason about context, capture leads conversationally, and hand off to humans when sentiment changes. Scripted chatbots can’t do any of that.

Q2What does the agent do that humans can’t?

It works 24/7, never forgets a doc you trained it on, replies in 50+ languages natively, and answers in seconds whether you have 1 visitor or 1,000 at once. It also captures leads mid-conversation without breaking the flow.

Q3Will it hallucinate?

The hybrid RAG pipeline grounds every answer in your documents. When the agent doesn’t know, it says so and offers to escalate or take a message — it doesn’t invent. Higher plans bring sharper chat models for more nuanced reasoning, but the same grounding rules apply on every tier.

Q4What happens when a customer needs a real human?

Sentiment-aware escalation triggers a handoff with a pre-summarised conversation. You’re notified by email or webhook in real time. Slack and WhatsApp notifications are on the roadmap.

Q5What languages does it support?

50+ out of the box. The agent detects the visitor’s language per message and replies natively — no separate setup per locale.

Q6Can it capture leads, or just answer questions?

Both. Conversational lead forms with smart triggers detect buying intent (pricing questions, demo requests, complaint sentiment) and capture contact info mid-conversation without breaking the flow.

Q7What’s an "AI agent" vs a "chatbot" in your terminology?

We say AI agent because it does more than match keywords to scripted replies. It reads your documents, reasons about context, decides when to capture a lead, decides when to escalate, and adjusts tone to match the visitor. The shape of the work is closer to a junior support rep than a Q&A FAQ widget.

Billing

Q1How long does it actually take to set up?

For most businesses, under 10 minutes from signup to a live agent on your site. Upload docs or point us at your website, pick a template, paste the embed snippet. Training is automatic.

Q2Is there a free trial?

Every plan ships with a 7-day free trial. A card is required at signup so the trial can convert seamlessly into your chosen plan — no second payment step at the end.

Q3What’s the cheapest plan?

Starter — £39/month billed annually (£49 monthly). Includes 1 AI agent, 2,000 message credits per month, 20MB storage, and the same intelligent foundation (mood, intent, grounded chat AIs) as every higher plan.

Q4What’s the difference between plans?

Capacity, chat-model power, branding, and SLA. Every plan runs the same three AIs in parallel — mood detection, intent classification, and grounded RAG chat. Higher plans bring more agents, more credits, a sharper chat model (Cutting-Edge LLMs from Growth and up), advanced analytics, and team-member seats.

Q5What’s a "credit" and how do I use them?

One credit = one AI message reply, with the chat-model class fixed by your plan. Starter messages cost 1 credit each (Midrange LLMs); Growth/Scale/Enterprise messages cost 2 credits each (Cutting-Edge LLMs). Your plan includes a monthly allowance; top-up Credit Packs are available as add-ons that don’t expire.

Q6Can I add Custom Branding on Starter?

Yes — Custom Branding is an addon on Starter and Growth (£49/month) and is included on Scale and Enterprise. It removes AI Smarty branding from the widget and lets you train the agent on your brand name.

Q7What if I cancel?

Cancel any time from your dashboard. Your data is removed within 30 days; sooner on request. No retention games, no export fees.

Security & Compliance

Q1Where is my data stored?

In the UK and EU only. Embeddings, conversation history, and uploaded documents stay in EU-region storage. LLM providers (OpenAI, Google, Anthropic, Microsoft) do not retain your prompts or completions.

Q2Are you GDPR compliant?

Yes. We provide a signed DPA, maintain a public sub-processor list, support data subject access requests, and ship right-to-be-forgotten flows by default. See the Data Processing Agreement page for the full text.

Q3What about LLM providers seeing my data?

Prompts and completions sent to OpenAI, Google, Anthropic, and Microsoft are not retained by them — confirmed by their commercial-tier no-retention agreements. We don’t fine-tune third-party models on your data.

Q4Is the data encrypted?

AES-256 at rest, TLS 1.3 in transit.

Q5How does multi-tenant isolation work?

Per-organisation Row Level Security at the database layer, plus service-context scoping at the application layer. Zero cross-tenant data exposure.

Q6Are there audit logs?

Yes. Every mutation is tracked. Audit log export is included on Enterprise.

Q7Where’s the DPA?

See the Data Processing Agreement page — countersigned via DocuSign on request.

Integrations & Setup

Q1How do I embed the agent on my site?

One line of HTML — paste the script snippet from your dashboard. The widget streams responses, captures leads, and reports back to your dashboard live.

Q2Can I match it to my brand?

Yes — logo, colour scheme, agent name, tone of voice. Full white-label (no AI Smarty footer) is available on Growth plans and above as the Custom Branding add-on.

Q3What can I train it on?

PDFs, Word docs, plain text, your website (we crawl it), Notion or Confluence exports. Storage limits scale with plan tier.

Q4Does it integrate with my CRM?

Lead-capture data is exportable via webhook today; native CRM connectors (HubSpot, Salesforce, Pipedrive) are in active development.

Q5Can I use my own model?

On Enterprise we support customer-supplied LLM endpoints; on lower tiers the model class is fixed by plan tier (Midrange / Cutting-Edge).

still have questions?

The fastest answer is a five-minute trial.

7-day free trial. UK/EU hosted. Cancel any time.

FAQ — AI Customer Support Agent Questions | AI Smarty