AI SOLUTIONS
Solving problems using AI
You’re under pressure to deliver AI that works in production, not just in a demo. We build and operate self-aligning AI systems, and we run our own solutions on the same stack we use for clients.
- Reviewer says no
- Rule
- Passed by rule
A reviewer says no once, and we write down why. After that, the rule handles the same case without them.
01 / Our solutions
What we built from the work
Four in-house solutions, in use on our own work first.
- Reviewer says no
- Rule
- Passed by rule
A reviewer says no once, and we write down why. After that, the rule handles the same case without them.
Cyclotron
One AI decision that learns from your team’s corrections.
It starts on day one with a decision model’s general training, refits a small model on each correction, and lets an LLM propose better questions that code tests before they go live. The loop is open source under the MIT license.
- Approved
- Declined
- Needs approval
People and bots share one room. A bot asks before it acts: a person approves or declines each request.
Chatticus
Shared rooms where people and bots work together, with approvals.
A collaborative workplace with named teammates, one shared computer, skills, routines, and human approval boundaries.
- Approved
- Held
- Agent
- Editor
Agents watch the beat, assign, report and draft. A person, the executive editor, approves each item before it publishes.
Papyrus
An automated newsroom: agents watch your beat, research and draft, and your editors decide what publishes.
Stories move through Kanbus-governed stages from pitch to draft to published, with a human editor approving each move. Anth.us runs its own newsroom on it, and the board is public.
- Question
- Yes
- No
A question goes in. A verdict comes out, with a confidence we can check.
Primus
The classifier lab: evaluate, deploy and improve AI with reviewer feedback.
The Anthus MLOps platform and agent incubator: it evaluates, deploys, and continuously improves AI agents with human feedback in the loop.
02 / Supporting technology
What the solutions run on
Kanbus
Issue tracking in your repository, where agents can participate.
Tactus
A programming language for reliable, tool-using AI agents.
Biblicus
Corpus analysis: extraction, retrieval and evidence-first knowledge workflows.
Auritus
Listen to any page: a drop-in embed that turns your content into speech with open models you choose.
Limatus
Finds where AI-written prose sounds generic, measured against your own house style, without rewriting it.
Pudicus
A commit gate that stops agents and people from committing secrets, with a receipt the deploy can check.
03 / What we build for clients
Four kinds of system
Knowledge Bases
Ontologies and taxonomies that learn from your data instead of going stale.
Self-Aligning Automation
Systems that improve from production feedback, with a human in the loop.
Agent Systems
Durable, governed agent procedures with sandboxed tools and rollback.
Machine Learning and Decision Models
Custom models, fine-tuned classifiers, and hosted decision models like Jev, aligned to your business.
04 / Engagements
Client work with a business outcome
The systems we built, ran, and kept improving in production. The proof is years of production RLHF for a call-center QA operation: a data flywheel learning from expert feedback across hundreds of scorecards and millions of interactions.
DataParade
DataParade is a privacy and security product built on a semantic knowledge graph, with a custom ontology and taxonomy that ties into the OCSF standard. Anthus is working with DataParade to develop it. It belongs to the new class of tooling that lets people see what AI agents are doing across a project once they've stepped back from writing the code themselves. • Semantic knowledge graph as the system of record for what happened, to what, and by whom • Custom ontology and taxonomy mapped to the Open Cybersecurity Schema Framework • Built for the operator who has delegated the work and still has to answer for it Visit dataparade.io.
- Pass
- Miss
Each call is one document and gets one scored row. We have scored calls this way across hundreds of scorecards and millions of calls.
Call Criteria
Call Criteria scored call-center QA with custom fine-tuned LLM classifiers run by an outside service. It worked, but the vendor was too expensive and too slow to scale to every new scorecard, and they wanted control. We built them an in-house classifier lab on Primus in their own AWS account and operated it for years: hundreds of scorecards, millions of calls, the cheapest model that clears each bar, and reviewers whose corrections became policy.
- Event
- Function
- Delivered
Ticketing, VIP tables and guest lists on serverless, event-driven infrastructure, wired into loyalty, ServiceNow and a data warehouse.
Venue Driver
A nightclub management system that has been the backbone of Las Vegas hospitality operations since 2007, processing over a quarter of a billion dollars in total revenue: • Processed as much as $64 million in revenue a year • 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 with other systems, creating a comprehensive hospitality management ecosystem Venue Driver's longevity, financial impact, and successful technological evolution demonstrate our ability to create and maintain mission-critical solutions that grow with our clients' needs, from on-premises to cloud to serverless architectures.
05 / How we work
From use case to long-term operation
We work under specs, tests, staged rollout, and a person who can say no: cybernetic development. Every AI application we build follows the same lifecycle. You own and keep everything we build for you. Our solutions are open source and stay open. Integration code specific to your project is yours and private, and anything it depends on in our public projects stays public.
- 01
Scope
- Define the use case
- 02
Deploy
- Deploy a basic proof-of-concept application scaffold
- 03
Select model(s)
- Choose an existing model
- Or train a new model
- 04
Adapt and align model(s)
- Prompt engineering
- Fine-tuning
- Human feedback
- Evaluation
- 05
Application integration
- Augment the model(s)
- Optimize deployment for inference and long-term management
PRESS START
Bring us the problem.
Even if you can’t yet say what you need. Working that out is part of the first month, and it’s often the most useful part.