Case studies

Customer names stay under NDA. Tech, problems, and outcomes are open.

We respect the contracts. Customer names, deal sizes, and specifics that could identify the organization stay out of public case studies. What stays in: the technology we used, the problem we solved, and what was achieved. If you're comparing vendors, the codename pages below should tell you enough.

Tell us your hardest deployment problem About Agrus

Healthcare

Healthcare app messaging + AI integration

Compliant in-app patient messaging with AI triage layer for a North-American healthcare platform.

Problem

Patient–clinician messaging needed both real-time chat and an AI triage layer, all under HIPAA, with provincial Canadian data-residency requirements. SaaS AI tools failed the InfoSec review.

What we achieved

Production deployment with zero PHI flowing to public AI vendors, signed HIPAA + provincial compliance map, audit log delivery into the customer's existing SIEM, sub-1-second triage response latency.

Technology

Ethora chat SDKDedicated cloud tenancyOpen-weights LLMHIPAA BAA chain

Full write-up coming

Insurance

Workers-comp insurance document processing pipeline

AI-driven claims-document extraction and adjuster copilot for a workers-compensation carrier.

Problem

Manual document review on first-notice-of-loss was the bottleneck; existing OCR was inaccurate; adjuster-side analysis required policy-specific RAG with adverse-action audit trail.

What we achieved

60%+ reduction in document-handling time, NAIC-aligned governance package signed by Chief Underwriting Officer, bias-testing program covering protected classes, fair-discrimination defensibility documented end-to-end.

Technology

Multi-agent orchestrationRAG over policy libraryBias-tested adverse-action loggingNAIC-aligned governance

Full write-up coming

Fintech / Blockchain

Regulated blockchain platform — self-hosted chat + agent SDK

Compliant chat and AI-assistant layer embedded into a regulated digital-assets platform's SDK.

Problem

Platform needed messaging and an embedded AI co-pilot inside its developer SDK without giving up auditability or violating regulator-mandated record-keeping requirements.

What we achieved

SDK shipped to customers with a compliant chat + AI layer; AML hooks active from day one; SOC 2 evidence collection automated and audit-ready; zero customer data crossing into third-party model logs.

Technology

Ethora SDKSelf-hosted XMPP + agent runtimeAML hook integrationSOC 2 evidence pipeline

Full write-up coming

Agrus itself

The Agrus website (this site) — built on Ethora

An AI-first agency website running on the same stack we ship to customers.

Problem

The marketing site needed to demonstrate, not just describe, our AI-first delivery model. Visitors needed real personas, AI agents needed a programmatic engagement path, and the whole experience had to be a credible Ethora deployment.

What we achieved

Live persona-led site, the public MCP server at mcp.agrus.ai (in development), every visitor's experience is a working Ethora deployment, every conversation a working agent build. The dogfood case is the strongest demo we have.

Technology

Next.js 15Ethora chat backend (in development)Anthropic Claude personasPublic MCP server

Full write-up coming

Why no names

The compliance customers we serve prefer it this way.

Most of our customers operate under data, contract, and regulatory constraints that make public case-study attribution awkward. We don't ask them to break that for our marketing.

What we offer instead: codenamed case studies that describe the technology, the problem, and the outcome openly — and, on a call, references to customers in similar situations who have consented to a private conversation with serious prospects.

For reference calls, ask on a scoping call. We'll match you with a customer whose situation looks like yours and who's agreed to talk privately with prospective Agrus engagements.

Become the next codename

Bring us a hard deployment problem.

Book a scoping call Talk to a persona