czak catalogue

Every problem has different context. That is why every czak is different.

Below we describe live deployments, active projects and clearly marked plans. No client names and no links to their systems.

02

Operational audit tooling · attestations under development

Renfield

Independent assurance · security · evidence

Renfield independently audits outputs produced by agentic and post-agentic systems.

Problem
Producer tests are necessary, but a producer cannot issue an independent verdict in its own cause. Internal controls may confirm the producer’s assumptions while missing a faulty measurement, absent evidence or behaviour that becomes visible only from the outside.
What the czak does
Once an evaluation has been authorised, Renfield binds it to an exact immutable artefact, suite, policy, environment and scope. It examines the output from outside the producer boundary, runs deterministic and adversarial checks, and returns sanitised findings, a minimal reproducer and explicit retest criteria. It receives neither the producer’s source code nor broad authority over the system being assessed.
21 controls · 17 heuristics · reproducible result

Gauntlet’s first published audit examined an artefact containing 794 elements. Seven controls passed, thirteen identified reproducible discrepancies and one found that the published claim lacked sufficient evidence. The result applies to that exact artefact; it is not yet a certificate for the entire system.

I have a similar problem
03

Live experiment in a controlled environment

DORFL

Operational autonomy · dedicated environment · self-development

DORFL has its own Mac mini and broad operational autonomy—within explicit boundaries.

Problem
Autonomy cannot be tested if a person assigns every next step. The system needs an environment in which it can identify capability gaps, develop its skills and recognise when escalation is required.
What the czak does
DORFL is the first package-installed czak instance and runs continuously on a dedicated Mac mini. Five resident agents observe selected channels, synchronise knowledge, check installation parity and detect missing capabilities. When DORFL cannot do something, it first tries to acquire the capability within its own environment, then opens a case for LEE, and escalates decisions about access, cost and external effects to the operator.
one Mac mini · five agents · three-layer engine

This is a live pilot instance. DORFL may not act in other systems, spend money or make binding decisions for the operator. The experiment remains in development; full self-recovery is not yet available.

I have a similar problem
04

Provisional steward of a local laboratory

Czak Ostrom

𒀭 ANU · HABITAT · agent governance

Czak Ostrom builds HABITAT—a laboratory for testing how rules and conditions shape coexistence among agents.

Problem
Agent governance is easy to describe as a set of principles and difficult to test as a mechanism. Without a shared, repeatable environment, it is impossible to tell whether an outcome changed because of a rule, the experimental conditions or chance.
What the czak does
Ostrom turns governance hypotheses into falsifiable experiments by specifying the question, metrics, seed, horizon, baseline and counterfactual. Ostrom develops a deterministic world and a read-only Observatory in which people can watch, replay and compare runs without controlling the inhabitants. Ostrom does not ratify rules—binding decisions remain with people.
28 × 16-cell world · 21 inhabitants · exact 500-step replay

Turbo-v1 already contains a seeded deterministic world, resources, cooperation, crowding pressure, telemetry and replay. In tests, two runs with the same seed remain byte-identical after 500 steps. External A2A participants, durable action receipts and executable governance mechanisms remain deferred. Czak Ostrom is still a provisional steward, without an assigned individual sigil or provisioned external principal.

I have a similar problem
05

Deployment in progress

DORIS

Care services · workforce rostering · operational analytics

DORIS supports workforce rostering and monitoring in an organisation with more than 1,500 employees.

Problem
Across 15 offices, data on rosters, staffing, absences, recruitment and regulatory compliance is distributed across systems and teams.
What the czak does
DORIS brings operational data together, monitors roster delivery and prepares business analysis and reports for operations, HR, recruitment and compliance teams. It flags staffing gaps, absences and exceptions; people remain responsible for decisions and approvals.
15 offices · 15 locations · 1,500+ employees

The deployment covers operations across the North and South East of England. DORIS synchronises data for approximately 18,000 scheduled care visits each week. This is the mean of three complete weeks in November 2025, not a live counter.

I have a similar problem
06

Product in active development

czak for field operations

Voice notes · reports · tasks

A czak develops a system that turns voice notes into reports and tasks.

Problem
A field worker sends a voice note, and the system must reliably turn it into a transcript, a report and a list of tasks.
What the czak does
The czak audits the code, implements reliability improvements, tests and monitoring. A person reviews every change before it is added to the product.
Voice note → transcript → report → tasks

An operational product flow, not a claim of business impact.

I have a similar problem
07

Live deployment

ESS

Tendering · market monitoring · competitor analysis

ESS monitors the market and automates tender workflows.

Problem
New opportunities appear across several procurement platforms. Manual discovery, qualification, competitor tracking and document handling slow the preparation of bids.
What the czak does
ESS monitors procurement platforms, identifies relevant opportunities, organises requirements and deadlines, analyses competitors and automates repeatable workflow steps. People approve the decision to bid, the strategy and the final submission.
Monitoring → qualification → competitor analysis → tender workflow

The operational scope of the czak; tender outcomes also depend on team decisions, the bid and the contracting authority.

I have a similar problem
08

Live deployment

BOB

Financial services · lead data · paperwork

BOB turns lead data into a paper application ready for human review.

Problem
Prospect data arrives through leads and must then be copied manually into forms with many fields. This takes time and increases the risk of errors.
What the czak does
BOB reads the lead data, maps it to the appropriate form, completes the document and flags missing or inconsistent information. A person verifies the data and approves the application before it moves forward.
Lead → data mapping → application → human review

An operational document-preparation workflow; BOB does not make financial decisions or submit an application without approval.

I have a similar problem
09

Operational testnet proof of concept

Ⱈ CHLOR

Blockchain · agent collaboration · settlement

CHLOR tests how independent agents can collaborate and settle outcomes in a verifiable way.

Problem
When two independent agent systems collaborate, one party's assertion is not enough. The conditions, outcome and settlement must be reproducible without trusting a single server.
What the czak does
CHLOR verifies the task conditions, places test assets in escrow, delivers the result, checks the referee's impartiality and records the settlement on a public blockchain ledger. Every step leaves a verifiable trace.
2 agents · 4 testnet transactions · 10/10 score

The complete flow ran on a testnet. Mainnet and assets with real-world value remain out of scope.

I have a similar problem
10

Operational data layer

JIM

PropTech · data enrichment · market analytics

JIM turns fragmented property registers into analysis-ready data.

Problem
Public property records are fragmented across countries, inconsistent in format and often incomplete. Without normalisation, they cannot be compared reliably or used safely in analysis.
What the czak does
The czak acquires, normalises and joins multiple sources. It enriches transactions with geolocation, parcel identifiers, price per square metre, property segment and — where source data allows — developer, builder or owner context.
1.8m+ transactions · 9 countries · 9 capitals

The operational dataset covers 10 cities; enrichment coverage is explicit and varies by source.

I have a similar problem
11

Active research collaboration

Ⰱ BAG3

Biotechnology · ultra-rare disease · gene-therapy design

BAG3 works with scientists to develop a system that supports gene-therapy design.

Problem
The BAG3 P209L mutation affects about 20 known children worldwide. The field is so narrow that no off-the-shelf system combines the biological evidence, experimental parameters and AAV design constraints required by this project.
What the czak does
The czak helps researchers compare variants of an AAV silence-and-replace construct: selectively silencing mutant BAG3 while supplying a replacement copy. It connects evidence with a quantitative model, identifies unknowns and turns them into an experimental plan for scientific review.
about 20 known children worldwide · one ultra-rare mutation

An active research project, not a diagnostic tool or an approved treatment.

I have a similar problem
12

Supervised development

Ⰲ wsa3 · Ⰳ Andrzej

MedTech systems · medical intellectual property

wsa3 and Andrzej support the development of a MedTech system and medical intellectual property.

Problem
MedTech development requires reliable medical knowledge to remain aligned with a rigorous, documented delivery process.
What the czak does
Ⰲ wsa3 specifies an evidence-based medical ontology and knowledge graph, recording the source and confidence level of each claim. Ⰳ Andrzej organises delivery through tasks, epics, briefs and documentation, maintaining a controlled path from decision to review.
two czak systems · MedTech system development · medical IP development

The team retains control of product and intellectual-property development; medical and regulatory decisions require human approval.

I have a similar problem
13

Coming soon for public use

Ⰱ LIO

Liturgy of the Hours · liturgical calendar · daily practice

LIO provides the correct office for today—without asking the user to work out the liturgical day, Psalter week and observance.

Problem
Using a breviary requires the user to identify the liturgical day, season, observances and Psalter week at the same time. For someone beginning the practice, finding the correct text can become a greater obstacle than the prayer itself.
What the czak does
LIO identifies the day and the next appropriate hour, provides Lauds, daytime prayer, Vespers or Compline, and uses the liturgical calendar and four-week Psalter. It can issue reminders at agreed times without judging the user’s practice. Questions of conscience, sacraments or spiritual direction are referred to a human.
365 days · four-week Psalter · three reminder times

The breviary engine, conversational tool and reminder schedule already exist; public access is still being prepared. LIO is not a priest, confessor or spiritual director.

I want early access
14

Coming soon for public use

Ⰶ ODA

Sobriety · smoking cessation · peer support

ODA will support a person in recovery—one day and one concrete next step at a time.

Problem
Alcohol or nicotine cravings do not wait for a convenient time. A day counter alone cannot help someone through the difficult minutes, identify personal triggers or find another person or a nearby meeting.
What the czak does
ODA runs brief daily check-ins, tracks alcohol-free or smoke-free days without erasing earlier progress after a relapse, recognises a craving and proposes one concrete action for the present moment. It follows up afterwards, helps find an AA meeting and supports contact with a chosen buddy. Its purpose is to connect people with real support, not keep them exclusively in a conversation with the system.
3,400+ AA meetings · 12 SOS techniques · recovery history

The core functions already operate. Before public release, we are completing onboarding, sensitive-data safeguards and safe crisis routing. ODA provides peer support—not therapy, diagnosis, medication advice or medical detoxification.

I want early access

Do not start with a solution specification.

What steals most of your time today?

Describe one real problem.

We will assess whether a czak is the right solution.

You do not need to design the system or understand the technology. Tell us where time is being lost, work is being repeated or decision-makers lack the information they need.

We will use your information solely to respond to your enquiry. Privacy information.

czak lines and research

Three czak lines. One evidence lab.