Anecdote

OnClarity — Anecdote until the September 2026 rebrand — is worth understanding as two products sharing one data layer. The first is a Voice of Customer analytics engine: it ingests calls and IVR transcripts, chat, email, WhatsApp, social, app reviews, tickets, CRM records and surveys into one stream with PII removed at ingestion, then joins public signal to contact-centre signal on the same layer. That join is the genuinely hard part, and it is why the product's customer list skews to organisations with hundreds of millions of users and national contact-centre operations. The second is an agentic support stack. The AI Support Agent resolves chat and voice inside your existing contact centre; Agent Assist drafts replies for human agents with the source and confidence score attached; anything under your threshold is handed over with context. Then Verify does the work nobody else does well: Agent QA scores each handled conversation against your rubric with reason codes, and Brain Guard separates "the answer was wrong" from "no source existed to check it against." Verified findings are written back to the knowledge base as new articles and rules — the loop's Learn stage. Strengths. The verification taxonomy is specific and defensible; most competitors stop at deflection rate. Deployment flexibility is real: same topology in-region cloud, sovereign private cloud or fully on-premise, which is often the deciding factor in regulated procurement. The release cadence is fast and documented — Azure OpenAI endpoint support, feature-flag editor, canvas verification, publishable mini-apps with version history and rollback, live figure widgets and AI-expanded search all shipped in September 2026 alone. And there is a published body of buyer-facing research: an AI voice agent evaluation guide covering 30 recorded calls, five Arabic dialects, latency percentiles and PDPL residency checks, plus a Brandwatch alternatives comparison that scores itself against Sprinklr, Qualtrics, Medallia and Chattermill, and an AI ROI business-case guide with a 90-day measurement plan for CFO approval. Weaknesses and honest caveats. This is not a product you switch on. Brain Guard can only ground answers in knowledge base content that exists and is maintained, and the seed material is blunt that organisations without anyone maintaining the KB should look elsewhere. Breadth — five agent stages, two support surfaces, a VoC analytics suite and a dashboard builder — implies a real learning curve and ongoing configuration work. The product reads as web-based enterprise software rather than something you install locally and forget. And because this is an enterprise sale, expect a procurement cycle rather than a credit card checkout. Where it fits. Banks, telcos, insurers, healthcare providers and large marketplaces with data-residency requirements, a contact centre already running (or about to run) an AI agent, and someone who owns the knowledge base and can act on the fixes the