Solving Problems Using AI
You're under pressure to deliver AI that works in production, not just in a demo. Anthus builds and operates self-aligning AI systems — custom models, agent harnesses, and evaluation loops with a human in the loop, grounded in 14 years of production operations. Our Call Criteria work is the proof: two years of production RLHF, a data flywheel that keeps learning from expert feedback, serving real QA at scale.
What We Build
- Knowledge Bases — ontologies and taxonomies that learn from your data instead of going stale. See Biblicus.
- Self-Aligning Automation — systems that improve from production feedback, with a human in the loop. See the Call Criteria case study.
- Agent Systems — durable, governed agent procedures with sandboxed tools and rollback. See the approach.
- Machine Learning — custom models and fine-tuned classifiers aligned to your business. See the work.
Our Approach: Cybernetic Development
We don’t just build AI features—we build the governors that make them safe to operate: clear specifications, layered verification, staged releases, and feedback loops that incorporate production learnings.
- Specs first: define behavior before implementation.
- Defense in depth: sandboxed tools, CI gates, staged rollouts, and fast rollback.
- Operational feedback: telemetry and incident-driven regressions that tighten the loop over time.
- Simplify and delete: reduce degrees of freedom to eliminate entire classes of failure.
Learn more in our article on Cybernetic Development, or contact us to talk through your goals and constraints.
Ready to revolutionize your business? Let's talk about what we can do for you.
Featured Solutions
Our recent work showcases AI-driven solutions that demonstrate production-ready implementations of agentic AI, RLHF systems, and intelligent automation:
![B0rd]()
B0rd
2026 - ongoing
B0rd is a standalone LED-matrix desk display built by Anthus Microelectronics, a spinoff venture rooted in Anthus's ubiquitous use of AI agents and the constant all-day need to monitor their progress and steer them. It puts glanceable information — launch countdowns, agent status, notifications, an idle clock — on the wall or desk without a screen you have to sit in front of.
- Handbuilt microelectronics with a custom handbuilt (AI-assisted) OS
- Standalone appliance: setup over Wi‑Fi, no app store, no phone required
- Glanceable cues for long-running agent sessions — walk away from the desk and still know
- In sync by design: matching units stay in step with no pairing or hub
![Speaker Role Classifier]()
Speaker Role Classifier
November 2025
A real-world example of how the agentic programming paradigm enables rapid solution development. We faced a problem where ambiguity in speaker role names could have made a traditional programmatic solution complex and time-consuming. By giving an AI agent the right tools, we delivered a working solution quickly and moved on—demonstrating how agent-based programming enables much greater agility and speed. View the project on GitHub.
![Classification with Confidence]()
Classification with Confidence
September 2025
A practical exploration of LLM fine-tuning for text classification. By fine-tuning GPT-4o-mini, we significantly improved classification accuracy and alignment to specific business requirements. The project demonstrates a production-ready approach to model alignment with confidence scoring that determines when human review is needed—a practical application of human-in-the-loop principles that balances automation with expert oversight. View the project on GitHub.
![SQLBot: Your AI Database Analyst]()
SQLBot: Your AI Database Analyst
September 2025
SQLBot is a new kind of interface for your database. Instead of writing SQL queries yourself, you delegate high-level analytical tasks to an AI agent. It reasons through your request, executing a chain of queries and analyzing the results until it arrives at a complete answer—all while keeping your data safe with built-in safeguards. View the project on GitHub.
![Call Criteria]()
Call Criteria
October 2023 - ongoing
We transformed a call center QA company by converting human-led processes into cybernetic systems that blend AI with human expertise. This groundbreaking work led to the development of Plexus, our enterprise AI platform. The results: unprecedented scalability with costs reduced by orders of magnitude, AI call reviews that outperform human reviews in speed and consistency, and automated AI setup for new client scorecards that streamlined onboarding. Our solution solved immediate challenges while opening new avenues for continuous improvement in call center quality assurance, ultimately evolving into a comprehensive enterprise AI platform.
Portfolio
Our journey spans decades of solving complex business challenges, from serverless architectures to AI-enabled systems:
![Plexus]()
Plexus
March 2024 - ongoing
Plexus is our custom MLOps platform built as the core infrastructure for the AI/ML lab at Call Criteria. It manages the complete machine learning lifecycle for creating AI agents and ML models for classification at production scale:
- Custom-tailored MLOps lifecycle management for manual lab work
- Full RLHF data flywheel at scale with continuous online learning
- Constantly aligns more closely with human feedback over time
- Manages training, evaluation, deployment, and monitoring of self-evolving AI agents
- Enterprise-grade platform supporting production-scale classification systems
Ticketing Data Analysis
December 31, 2023
Engineered a serverless data lake for Tao Group Hospitality, streamlining complex Las Vegas ticket sales reporting. Our solution automates compliance with resort partners, auditors, and the Nevada Gaming Control Board, while reducing operational overhead.
![Vault]()
Vault
July 2014 - January 2024
Implemented a cloud-based enterprise data warehouse (EDW), aligning with the growing trend in 2014 of cloud adoption for data-intensive operations. Vault served as the cornerstone of data operations for over a decade, enabling seamless integrations with major platforms like Marketo, Salesforce, and Salesforce Marketing Cloud. This approach provided scalability and flexibility while reducing infrastructure costs. By centralizing data from multiple sources, Vault empowered businesses to derive actionable insights, streamline marketing operations, and enhance customer relationship management. It also gave us a platform for exploring ML use cases, like customer volume forecasting. Its longevity and adaptability underscore our commitment to building future-proof solutions that deliver long-term value.
Ticket Driver Copilot
July 04, 2023
An AI agent capable of displaying live charts of CloudWatch metrics from the Ticket Driver event ticketing system, with the ability to report on the system's performance. Used internally to monitor and manage the system. Proactively notifies us in a Slack channel when there are CloudWatch alerts from the system.
Store Driver
November 01, 2022
Serverless Angular app replacement for Venue Driver's legacy event ticket store, which previously ran on a Ruby on Rails app running on live servers that generated HTML on the fly.
Ticket Driver
November 2022
Modernized Venue Driver with a serverless architecture, improving reliability and efficiency by eliminating server management. Implemented a React frontend and GraphQL backend, dramatically improving system reliability and performance while reducing operational costs.
Menu Driver
July 04, 2021
An emergency project aimed at elevating web restaurant menus to a high-availability service during the post-pandemic reopening phase, when touchless QR code menus became a mission-critical requirement for large restaurant groups. Menu Driver uses AWS Comprehend to make decisions about menu formatting based on analysis of the menu content.
![Venue Driver]()
Venue Driver
2007-09-15
A pioneering nightclub management system that became the backbone of Las Vegas hospitality operations for 16 years, processing over a quarter of a billion dollars in total revenue:
- Handled peak volumes of tens of millions of dollars per month
- Began as an on-premises solution for reservations and guest lists, tailored for Las Vegas nightlife
- Underwent successful cloud transformation to AWS, enhancing scalability and reliability
- Transitioned to serverless architecture, dramatically reducing operational costs while improving performance
- Adapted continuously to support new business models and regulatory requirements
- Integrated seamlessly with other systems, creating a comprehensive hospitality management ecosystem
Checkout Driver
April 30, 2021
Serverless upgrade for the mission-critical event ticket store checkout, making a transition from a traditional payment gateway to a tokenized system that eliminated all processing and storage of customer credit card information.
Integrations
Paytronix
March 15, 2022
Delivered a certified integration between Tao Group Hospitality's event ticket store with Paytronix to support new Tao Rewards loyalty program. Paytronix also certified another integration for gift card features for the Tao Group Rewards iOS and Andriod apps.
Salesforce Marketing Cloud
October 17, 2018
Moved Hakkasan Group marketing automation from Marketo to Salesforce in an attempt at deeper integration. Honestly, it sucked. We should have stayed with Marketo.
Salesforce
April 30, 2018
Streamlined VIP table reservations for Hakkasan Group's global nightclubs by integrating Salesforce with their data warehouse and reservation system. This solution optimized the sales process, enhancing revenue management for high-value bookings in Las Vegas and international venues. The integration enabled real-time data flow, improving decision-making and customer service for premium clientele.
Marketo
July 08, 2014
Implemented Marketo integration for Hakkasan Group, activating meticulously designed customer journeys. This project enabled highly personalized marketing campaigns, improved targeting accuracy, and enhanced marketing efficiency. The result: streamlined operations, higher customer engagement, and increased revenue through data-driven marketing automation.
The Process
