Frequently Asked Questions

There are no call quotas; every call in your system is analysed. Pricing is per user: add users as the team grows, remove them when it shrinks. CALLAII supports the process in real time with agent performance and quality metrics.

The agent team add-on starts with at least 3 agents. We size it through a discovery phase and assign exactly the headcount you need. As volume grows, new agents come online quickly and your SLA holds.

We forecast the seasonal curves in your operation. For campaign peaks or sudden spikes, our standby agent pool comes online.

Yes, the platform has an agent team add-on. The team, shift planning, quality assurance, SLA tracking and reporting all run inside the platform, and you follow the outcome on your dashboard. The team is sized through a discovery phase and grows from a standby agent pool when volume rises. It can be set up for a single channel or just the evening shift. Licensing the software on its own and running the operation with your own team is equally possible.

On the phone, no: a human agent always speaks with the customer. On written channels a person answers as well, as long as it is working hours and an agent is at the console. When no agent is available CALLAII steps in, answers from your own knowledge base, and hands the conversation to the right agent the moment the topic exceeds its limits or the customer asks for a person, opening a task for the follow-up. CALLAII does its real work in the background: it analyzes calls, generates coaching and escalates risk to managers.

It does. There are prebuilt connectors for Salesforce, HubSpot, Zendesk, Freshdesk and Google Contacts on the CRM and support side; Shopify, WooCommerce, Ticimax, ikas, IdeaSoft and T-Soft in e-commerce; Asterisk (FreePBX, Issabel), NetGSM, Verimor, Bulutfon and Twilio on the telephony side. Anything not on the list connects through a generic webhook. The agent sees the customer's order, ticket and contact history together on the conversation screen. The current list and the status of every connector are published on the integrations page.

KPIs such as SLA, NPS, CSAT and handle time are visible in real time on a dashboard scoped to you. Quality scores come from every single call rather than a sample; the breakdown per agent, missed opportunities and risk alerts sit on the same screen. Real-time, end-to-end transparency.

5-10 working days for software only, about 20 working days for a fully managed operation. Call flows, reporting and CALLAII configuration are calibrated together.

Omnichannel: voice, email, live chat, WhatsApp and social - all in a single operations console.

No. CALLAII connects to the system you already use through a SIP trunk. Your PBX (3CX, Asterisk, an enterprise SBC), your numbers and your carrier stay the same, and your calls flow exactly as they do today. CALLAII never sits in the call path. It only listens to a copy of each call through a standard recording stream and analyzes it. You add the power of CALLAII to your customer operation without changing your habits.

A SIP trunk is the standard way phone calls travel over the internet, and most enterprise systems already use one. Your PBX forwards a copy of each call to CALLAII through SIPREC, the international call-recording standard. Setup is code-free: you enter the server address, port and transport in the panel, and the rest is automatic. The audio stays inside your organization and never leaves.

Yes. GDPR-compliant infrastructure (with KVKK for Türkiye operations). Data is held in isolated tenants per customer and is never used to train models for other organisations.

Your knowledge base - product docs, procedures, resolution records - is plugged into the workflow through CALLAII's RAG layer. Agents get real-time, in-call suggestions.

Yes. A dedicated prompt framework, business rules and terminology set is designed for each customer. Recommendations are on-brand, consistent and auditable.

Traditional QA teams can only sample a small fraction of calls. CALLAII scans every interaction with natural-language analysis and surfaces quality and risk metrics automatically.

No. For legal safety, dialling only starts after digital İYS and KVKK consent checks are completed on an XLS/CSV upload. In Türkiye, BTK fines for calling without consent start at 25,000 TL per contact.

A monthly subscription per user. There are four packages; each comes with a set number of users and you can add extra seats on top. No commitment, no setup fee, twenty percent off on annual billing. Managed operations and Edge deployments are prepared for you based on scope.

The pricing page shows the monthly and annual price of all four packages, the user counts, extra seat fees and sector add-on fees. Only Enterprise above 50 users and managed operations are prepared individually; the scope differs at every organisation, so we do not publish a list price for those.

No. There is no commitment; you pay monthly. Flexible contract structures are available. Your data is handed over or destroyed on request.

14 days free, no credit card. You open the account and use it with your own team. If you want the PBX and CRM connected, we set it up together: 5-10 working days for software only, around 20 working days for a fully managed operation. At the end of the trial the decision to continue is yours.

There are four packages. Go: 3 users, for a small team working over messaging; customer records, website live chat, messaging and appointments. Connect: 5 users, for a team working on the phone; everything in Go plus the call centre, calling from the browser, call transcripts and unmetered conversation analysis. Suite: 15 users; on top of Connect, shift planning, field teams, projects and quotes. Enterprise: above 50 users, prepared for your company with sector add-ons and on-premise deployment. CALLAII is the AI layer inside every package and is included in the subscription price. The health, automotive and collections sector add-ons sit on top of your package.

It runs on WhatsApp Business (Meta Cloud API). You connect your own WhatsApp Business number and Meta account; incoming and outgoing messages are handled on the same screen as live chat in the agent panel. CALLAII suggests draft replies to incoming messages, supported by your company-specific RAG library and Playbook rules. Instagram DMs and Facebook Messenger are handled from the same inbox.

Your subscription covers the Callmenta software and the CALLAII AI (AI processing runs on our own GPU infrastructure at no extra charge). Channel usage fees are paid directly to the provider, at cost: to your operator (NetGSM/SIP) for call minutes, and to Meta for WhatsApp/Instagram/Facebook conversation fees. Callmenta adds no commission on top of these amounts; you pay from your own account on your own tariff.

Meta defines pricing on the WhatsApp Business API itself, per conversation and by country and category (marketing, service, authentication). That is why the fee is charged directly to your Meta account, with no intermediary. This keeps it transparent and adds no markup on top of the real cost.

Yes — that is exactly the model (BYO, bring your own account). You register your own NetGSM/SIP line and your own Meta WhatsApp Business account in the panel. The numbers, history and billing stay with you; Callmenta provides only the software and the AI layer.

If you are a small team working over messaging channels (site chat, WhatsApp, Instagram), Go. If you run a call centre over the phone, Connect. If you also want shifts, field teams, projects and quotes in one place, Suite. If you are above 50 users or need sector add-ons or an on-premise deployment, Enterprise. If you are unsure, try it free for 14 days and your need will make itself clear.

CALLAII Edge is the on-premises version of the entire Callmenta platform (panel, call center, CRM, industry modules, and CALLAII AI: speech recognition, language model, RAG, scoring) deployed on your own servers. The license determines which modules are enabled. Call recordings, transcripts, and evaluations are never transferred to Callmenta servers under any circumstance. Designed for data-sovereignty-critical sectors such as banking, insurance, healthcare, legal, and defense.

Suitable. A single NVIDIA workstation is sufficient for small businesses: a card with 16 GB memory (e.g., RTX 4080 or RTX A4000), 32 GB RAM, 1 TB NVMe SSD. CALLAII 12B and speech recognition run together on this card. CALLAII 26B requires a single 48 GB card or two 24 GB cards. Measurement: on two 24 GB NVIDIA L4 cards (one running CALLAII 26B, the other running speech recognition), approximately 700 calls per day are analyzed under continuous load. Exact capacity is measured and determined before deployment based on your call volume.

For institutions requiring high availability, such as banks and insurers: 2+1 nodes, one NVIDIA card with 48 GB memory per node (e.g., L40S or RTX 6000 Ada), ECC RAM, and NVMe RAID. Automatic failover and a fully air-gapped option are included.

Yes. Fully offline installation is supported; the license is validated via an offline file. Model and software updates are applied as signed packages, either manually or over the internal network.

A monthly licence at 1.6 times the cloud list price. Installation is quoted after a discovery call, hardware not included. The licence portal lets you manage nodes, push version updates, monitor health metrics and handle renewals - all from one place.

Because data never leaves your premises, there is no personal-data transfer. Every evaluation stays in your isolated tenant. Regulatory reports for BDDK, SPK and the Ministry of Health are ready to ship.

In the service model we manage the connection: a copy of the call audio reaches Callmenta’s KVKK-compliant infrastructure in Europe through a standard recording stream, and you start fast. In the Edge model the SIP trunk connection, recording and analysis all run on your own server and devices. The audio never leaves your organization and can even run fully offline (air-gapped). In both cases you keep your existing phone system. The only difference is where the processing happens.

Standard CALLAII scores the human agents in your contact center. CALLAII Pro scores human agents AND the AI agents in your operation against the same competency framework. Every approved evaluation becomes training data in one click.

ChatML, Alpaca and JSONL. Each row is one approved call evaluation - customer turn, agent reply, rubric scores and agent type (human or AI) - ready to drop into a fine-tune for your own language model with your real operational data.

Upload your AI agent's call transcripts to the console. CALLAII Pro scores them against the same rubric as human agents. Human vs. AI is broken out on a dedicated comparison panel.

No. Only records your QA lead marks as "approved" are included in the export; "rejected" and "in review" stay out of the dataset.