Parloa

The best test-before-deploy discipline in voice AI — simulation, LLM-judge evals and red-teaming are the actual product — priced for enterprises that spend $300K a year to replace an IVR.

The verdict, read by Jan · AI clone of Jan’s voice, generated locally with his consent.

Best for: Enterprises with high-volume voice contact centers replacing legacy IVRs, especially in regulated industries where the simulation and red-team lifecycle justifies a six-figure annual contract.

Scope15/20

Quality8/10

Where the quality sits

8Reactive

6Retrieval & Memory

8Orchestration

9Validation

8Models

Enterprise platformVoice AgentsAutomation & AgentsVoice & SpeechChatPaid

Vendor

Parloa · www.parloa.com

Origin

DE — Berlin

Pricing

Not published — custom enterprise quotes only; independent reviews report typical deployments from ~$300K (~€275K) per year

Users (official only)

Not disclosed; company-stated ARR above $50M, 380 employees, customers include Allianz, Booking.com, SAP, Swiss Life (source, 2026-01-15)

Element scores

Reactive

Retrieval & Memory

Orchestration

Validation

Models

Primitives

Compositions

Deployment

Emerging

Tap or hover any element to see why it got that score.

Strengths

Testing is the moat. Thousands of synthetic conversations per launch, LLM-judge plus deterministic evals, the Dojo benchmarking system that turns model comparisons into overnight evidence, and red-teaming with 1,000-1,500 adversarial conversations per engagement. Agent Skills on MCP (06/2026) cut integration build time from 4-8 weeks to hours. The voice stack is component-tested: STT on word error rate, TTS on blind listening tests. $560M raised, $3B valuation (01/2026), with Allianz, Booking.com and SAP in production.

Honest dings

No customer fine-tuning — adaptation is instruction-based only, where Cresta tunes on your data. Entry is reported around $300K/year with no published pricing. Independent reviews flag 700-900ms voice latency and a lengthy, IT-heavy implementation. Public review volume is near zero (one G2 review), so independent user signal is thin. And at company-stated ARR above $50M against a $3B valuation, you are betting the vendor grows into its price.

Prices and details change — this passport is re-verified at least quarterly.

Sources (10) — every claim traceable

Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).

parloa.com/platform — Official AMP lifecycle: Studio, skills, simulations, evaluations, versioning, model orchestration; GDPR, ISO 27001, SOC 2 compliance claims (accessed 2026-08-01)
openai.com/index/parloa — OpenAI case study (07/05/2026): GPT-5.4 in production, simulation/eval architecture, RAG grounding, per-component voice stack testing (accessed 2026-08-01)
techcrunch.com/2026/01/15/parloa-triples-its-valuat… — $350M Series D at $3B valuation, company-stated ARR >$50M, competitive context vs PolyAI/Decagon, customer list (accessed 2026-08-01)
parloa.com/parloa-in-the-press/parloa-valued-at-3-b… — Official Series D release: $560M total raised, 380 employees, Berlin/Munich/New York offices, named Fortune 200 customers (accessed 2026-08-01)
parloa.com/blog/agent-skills-accelerate-compliant-a… — Agent Skills on MCP (08/06/2026): deterministic tool calls, Success Conditions, 4-8 weeks to hours integration claim, 20% routing reliability gain (accessed 2026-08-01)
parloa.com/labs/insights/how-parloa-stress-tests-pr… — Red-teaming methodology (18/06/2026): 30-40 scenarios, 1,000-1,500 adversarial conversations, LLM-as-judge plus deterministic checks, full tool-call trace logging (accessed 2026-08-01)
parloa.com/labs/insights/parloa-s-model-evaluation-… — Dojo model evaluation system (02/07/2026): reproducible benchmarks, harness/component separation, 8-12 hour eval runs (accessed 2026-08-01)
marketplace.five9.com/s/product/parloa-ai-agent-man… — Five9 marketplace listing: 130+ languages, real-time language switching, PCI DSS/HIPAA/SOC 2 Type II, SIP-based CCaaS integration (accessed 2026-08-01)
synthflow.ai/blog/parloa-review — Independent (competitor) review: no published pricing, ~$300K/year typical deployment minimum, enterprise-only positioning (accessed 2026-08-01)
eesel.ai/blog/parloa — Independent (competitor-disclosed) overview: 700-900ms latency reports, lengthy implementation, Salesforce/Genesys/Verint integrations, voice-first trade-offs (accessed 2026-08-01)