FAQ

An AI voice bot is a system powered by artificial intelligence that conducts natural phone conversations with humans. It uses large language models (LLMs), text-to-speech (TTS), and speech-to-text (STT) to understand conversation context and respond in a human-like manner. Unlike traditional IVR, an AI bot doesn't rely on rigid scripts — it can handle unexpected questions and adapt its conversational tone.

It depends on your configuration. Coldbot by default introduces itself as an AI assistant at the start of a call, as required by GDPR and the AI Act. However, you can adjust the level of transparency — the bot can operate as an openly artificial assistant or as a fully natural consultant that only reveals its nature upon request.

Coldbot uses the latest ElevenLabs and OpenAI TTS models with Polish speech synthesis. The voice includes natural pauses, intonation, breaths, and filler words (e.g. 'hmm', 'well'). In A/B tests, listeners couldn't distinguish the bot from a human in 73% of cases. We also support regional accents and speaking rate adjustments.

Yes. Coldbot uses an advanced LLM pipeline with your knowledge base. When a caller asks something off-script, the bot references configured knowledge sources (documents, FAQ, website, pricing). If it doesn't know the answer, it politely offers a callback from a human agent or records the question for follow-up. The bot never hallucinates — it has strict safety guardrails.

Average response time (latency) is 600-900ms. Coldbot uses response streaming — the bot starts speaking before the full response is generated, creating a natural conversation feel. Latency depends primarily on the LLM model and internet connection, not on Coldbot itself.

Yes. The language models used in Coldbot understand colloquial language, industry jargon, and regional expressions. The bot also analyzes caller tone — it can detect irritation, enthusiasm, or hesitation and adjust its response. If negative emotions are detected, it can escalate the call to a human agent.

Yes, provided GDPR, data protection laws, and the AI Act are followed. Coldbot is fully compliant with these regulations. Key requirements: (1) lawful basis for processing (legitimate interest in B2B), (2) transparency (informing that the caller is AI), (3) opt-out and data deletion rights, (4) data minimization. Coldbot automatically records consent and handles GDPR requests.

A single bot can make 200-400 calls per day during 9 AM-8 PM hours. With 5 parallel lines (standard plan), that's 1,000-2,000 calls per day. In the Enterprise plan, we scale to 50+ parallel lines for 10,000+ daily calls. Actual numbers depend on call duration — more successful conversations mean fewer raw dials.

Average success rate (meetings booked / calls made) is 3-8% depending on industry and lead quality. For comparison, a human telemarketer achieves 2-5%. The bot never gets tired, always has the same energy, and can call at optimal times. Additionally, the bot learns from results and optimizes its approach.

Coldbot asks a series of qualification questions you define — budget, need, timeline, decision-making authority. Responses are analyzed in real time. The bot assigns lead scores (1-10) and only passes those meeting your criteria. You can define any qualification logic — from simple interest checks to complex decision trees.

The bot automatically checks availability in your calendar (Google Calendar, Outlook, Calendly), proposes time slots, books the meeting, and sends email/SMS invitations to both parties. Additionally, the bot sends SMS reminders 24h and 1h before the meeting. All lead data flows to your CRM via API integration.

Yes. Coldbot has a built-in objection handling system. When someone says 'I'm not interested', the bot doesn't end the conversation immediately — it gently probes for the reason and tries to overcome the objection. If the person firmly declines, the bot politely ends the call and marks the lead as 'rejected'. All objections are logged for analysis.

Basic deployment takes 2-5 business days. It includes: call script configuration, calendar and CRM integration, testing on 50-100 calls, and team training. Full deployment with a custom script, dedicated knowledge base, and advanced CRM integration takes up to 2 weeks.

Coldbot integrates with: Google Calendar, Outlook Calendar, Calendly, HubSpot, Salesforce, Pipedrive, Monday.com, Zoho CRM, ActiveCampaign, Zapier, Make (Integromat), Slack, Discord, and any system via REST API and webhooks. We also offer custom integrations on request.

No. Most integrations work out of the box — just log in via OAuth and select what should be synced. For custom integrations, we provide technical support and ready-made webhook templates. If you use a custom CRM, our engineers will help with the integration as part of the onboarding package.

You create scripts in our visual builder — drag and drop blocks representing conversation stages (greeting, presentation, qualification, objection handling, scheduling, closing). Each block is configured in natural language. No programming skills needed. For advanced users, there's an expert mode with a prompt editor and dynamic variables.

Yes. Import your database via CSV, API, or CRM integration. Coldbot automatically validates numbers, removes duplicates, checks against DNC (Do Not Call) lists, and optimizes dialing order based on historical performance (e.g., calling industries with higher conversion rates first).

Yes. Coldbot supports numbers in over 50 countries. The bot automatically matches the caller ID to the recipient's country (local presence) and accounts for time zones. For international markets, voices are available in 29 languages.

All data is stored on EU servers (Frankfurt, Germany) in compliance with GDPR requirements. Coldbot does not transfer data outside the European Economic Area. Infrastructure is hosted on certified AWS (ISO 27001, SOC 2). All data is encrypted at rest (AES-256) and in transit (TLS 1.3).

Yes. Recordings are encrypted and stored in an isolated S3 bucket with IAM-based access control. You can configure automatic deletion of recordings after X days. Only authorized users on your account can access recordings. Coldbot NEVER uses recordings to train models without your explicit consent.

Yes. We are fully prepared for the AI Act (effective from 2026). Coldbot qualifies as a limited-risk system. We meet all requirements: transparency (AI conversation labeling), human oversight (agent escalation), technical documentation, decision auditability, and training data management. We provide a complete compliance documentation package for your DPO.

Coldbot automates GDPR request handling. When someone requests data deletion during a call, the bot confirms the request, deletes data from the system in real time, and sends email confirmation. The dashboard provides a full request log with dates and statuses. Access to personal data is logged and monitored.

Coldbot uses a multi-layered security approach: OAuth 2.0 + MFA authorization, end-to-end encryption for API calls, isolated prompt environments (preventing prompt injection), real-time anomaly monitoring, and automatic blocking of suspicious activity. Our security team monitors the system 24/7.

Coldbot offers 3 plans: Starter (from €119/month + call costs), Professional (from €239/month, up to 5 parallel lines), and Enterprise (custom pricing, 10+ lines, dedicated support). Outbound call costs are additional (approx. €0.02-0.03/min). The first 50 test calls are free.

You pay a monthly subscription (platform access, bot configuration, integrations) + usage-based call minutes. There are no hidden fees. You don't pay for: number of bots, number of dashboard users, integrations, updates, or support (Professional+ plans). Minutes are billed per second.

Our clients report ROI of 6-15x over 3 months. Example: a solar company paying €350/month for Coldbot books 40-60 meetings monthly. With meetings valued at €500-1,200 and 15% conversion, that's €3,000-10,800 in revenue — 8-30x ROI. Compared to a sales rep cost (€1,500-2,500/month + commission), the savings are significant.

Yes, we offer a 14-day trial on the Professional plan. You get 50 free test calls, full access to all features, and onboarding support. No credit card required. If you're not satisfied — you pay nothing.

Yes. You can change plans anytime from the dashboard. Upgrading charges a prorated difference for the remaining billing period. Downgrading takes effect from the next period. There are no penalties or fees for changing plans.

Coldbot uses a combination of models: GPT-4o and Claude 3.5 Sonnet as primary conversational engines, ElevenLabs and OpenAI TTS for speech synthesis, Deepgram and Whisper for speech recognition. Models are automatically selected based on the task — e.g., GPT-4o for sales conversations, Claude for sentiment analysis. We regularly update models to the latest versions.

Coldbot has a built-in failover system. If the primary AI provider (e.g., OpenAI) is unavailable, traffic automatically switches to a backup (e.g., Claude/Anthropic). If both fail, the bot uses a local fallback with limited functionality and queues the call for a callback. Historical Coldbot uptime is 99.95%.

Coldbot's architecture is horizontally scalable. We currently serve clients making 100,000+ calls per month. The system auto-scales resources based on load. There's no hard limit — the only constraint is the allocated phone lines in your plan.

Coldbot monitors quality for every call: latency, jitter, packet loss. With poor quality, it automatically switches to a lower-bitrate codec (Opus 8-24kbps) to maintain the conversation. If quality drops below the threshold, the bot offers a callback and ends the call. All quality metrics are available in the dashboard.

Yes, on the Enterprise plan. We offer on-premise deployment or in your VPC (AWS, GCP, Azure). We provide full data isolation and dedicated AI models if required by industry regulations (e.g., banking, insurance). On-premise deployment typically takes 2-4 weeks and includes a security audit.