Parloa

What it is

Parloa is an AI agent management platform for contact centers — voice first, built to take the phone call itself rather than steer the caller into chat. The product is organized around the agent lifecycle Parloa names on its own platform pages: design, test, scale, optimize. Founded in Berlin by Malte Kosub and Stefan Ostwald, it raised a $350M Series D announced 15 January 2026, led by General Catalyst with EQT Ventures, Altimeter Capital, Durable Capital Partners, and Mosaic Ventures, at a $3B valuation and more than $560M raised in under four years.

In its May 2026 update the company reported crossing $50M in annual recurring revenue with 150% net revenue retention and 430 staff across New York, Berlin, London, and Munich. Named customers include Allianz, Booking.com, HealthEquity, SAP, Sedgwick, Swiss Life, and TeamViewer, alongside European references BarmeniaGothaer, TUI, and ATU.

Why it shows up in CX and support stacks

• Voice is the product, not a channel it bolted on. The chat-first platforms added telephony; Parloa started there. Its named integrations are the CCaaS layer a phone-heavy org already runs — Genesys, Five9, NICE, Avaya, and Verint — plus direct SIP/VoIP connection when you would rather not go through the CCaaS at all. If the decision on your desk is “who answers the 40,000 calls a month,” this is the shortlist it belongs on.
• It bills per minute, and argues the case in public. Parloa’s stated model is per-minute consumption pricing, not outcome-based revenue sharing, and it published the argument in January 2026: a minute and an API call are measurable, whereas whether a five-minute call counts as a “resolved interaction” is a negotiation with your vendor every quarter. That puts it on the opposite side of the meter from Intercom Fin at $0.99 per outcome and Zendesk at $1.50 per automated resolution.
• The compliance set is built for regulated phone work. ISO 27001:2022, SOC 2 Type 1 and 2, PCI DSS, HIPAA, and DORA. DORA is the one that matters if you are an EU financial-services buyer, and it is not table stakes among the US-founded competitors. The Epic partnership covers the HIPAA path into healthcare.
• It sells through partners, not only direct. SAP took a strategic investment and integrated Parloa with SAP Service Cloud; Microsoft Azure is the hosting foundation; the BPO route runs through Teleperformance, Concentrix, and Foundever. If your contact center is outsourced, your BPO can likely already deploy it.

Pricing reality

Nothing published. There is no pricing page, and OMR Reviews records zero reviews carrying price information. The billing model is per-minute consumption plus API calls, which is Parloa’s own stated position rather than a third-party inference. Secondary buyer analyses put the practical entry point near $300K per year with average contract values above $350K — treat those as estimates, not quotes; Parloa confirms no number publicly.

The operational consequence matters more than the entry price: your bill tracks average handle time. An agent that resolves an intent in 90 seconds costs 40% less than one taking 150 seconds on identical call volume. Model it as minutes multiplied by containment rate, and get three things in writing before you compare it against an outcome meter — the per-minute rate, the minimum annual commitment, and the overage rate above that commitment.

Best for

Enterprise CX and contact-center leaders in regulated, phone-dominant industries — insurance, financial services, utilities, healthcare, travel — already running Genesys, Five9, NICE, or Avaya, who need the AI to hold the call rather than deflect it, and whose compliance team raises DORA or HIPAA in the first meeting. The ROI band is best above roughly 500,000 annual voice minutes, where a few points of containment pay for the commitment.

Skip it if your contact mix is chat-dominant, if your volume cannot repay a six-figure first-year commitment, or if you need published pricing you can approve without a procurement cycle.

Versus the alternatives

The installed-base incumbents are Zendesk and Intercom — pick them when the AI already attached to your help desk clears your containment bar and adding a vendor is not worth the deployment, which is usually the answer when chat carries most of your volume. The fastest-growing entrant is Sierra, at roughly $200M ARR by May 2026; take Sierra when the agent must execute transactions inside PCI-scope flows across both chat and voice and you will accept outcome-based billing to get it. Decagon wins on aggressive chat deflection. Ada is the pick when you want the invoice decoupled from agent performance entirely and you need channel breadth — WhatsApp, Instagram, in-app — that Parloa does not compete on. Parloa’s case against all four is narrow and real: it is the one that treats the telephony stack as the primary surface instead of an integration.

Watch-outs

• The valuation runs far ahead of the revenue. $3B against the $50M+ ARR reported in May 2026 is roughly a 60x multiple. That says nothing about the product and quite a lot about the negotiation: a company priced for growth pushes multi-year terms and volume floors. Guard: cap the first contract at one year, and put shortfall rollover in writing rather than accepting a use-it-or-lose-it minimum — a floor you miss is the single most common way a voice-AI contract turns into shelfware you keep paying for.
• Per-minute billing charges you for latency and for a talkative agent. The meter that protects you from a vendor gaming “resolutions” exposes you to slow model responses and verbose prompts. Guard: instrument average handle time per intent from week one against your human baseline, fix the per-minute rate for the contract term, and add a review trigger if AHT rises above a named threshold rather than trusting that tuning will happen.
• Channel breadth is narrower than the chat-first platforms. The integration list is CCaaS and CRM — Genesys, Five9, NICE, Avaya, Verint, Salesforce, ServiceNow, Microsoft Dynamics, SAP, Twilio, Zendesk. Social and in-app messaging are not where it competes. Guard: count your contacts by channel before shortlisting; if voice is under 40% of volume, run the evaluation against Ada or Decagon and treat Parloa as a specialist you add later, not the platform decision.
• There is no MCP server, and API access is gated. The documented APIs at docs.parloa.com require a client-specific key issued by Parloa support, so conversational querying of your own CX metrics from Claude or ChatGPT is not available out of the box. Guard: request the API key, the endpoint list, and the rate limits during the trial — not after signature, when you have nothing left to trade.