Cookbooks / Departmental cookbook
Data, AI and Decision
Accelerator for operating in the EDAVEN hub
Source, data product, metric and decision context with a defined contract and semantics, with declared agents and clear limits.
A ready starting point for your industry
The cookbook is the offer version of the knowledge on this topic: it comes from the BoK, is implemented by Services with the CoE and starts running in the EDAVEN fabric, with people and agents working together.
- The knowledge
- BoK. Method, decisions, recipes and data for the topic, ready to adapt to your case.
- The implementation
- Services. The Arcogi team and the CoE apply the recipes to your case, as contracted.
- Where the operation runs
- The fabric. Data Ready, Organization and Agents and Decision Intelligence, as the case requires, with people and agents.
What the market is facing
Data and AI deliver value when decisions use data that is ready and agents have a clear place in the work.
Pains the cookbook solves
Each pain has an answer in recipes applicable to your case.
What it costs today: Nobody trusts the numbers.
What the cookbook changes: Data products with an owner, a contract and a single definition.
What it costs today: Agents without reliable context.
What the cookbook changes: Decision context assembled from ready, current data.
What it costs today: AI in use with no view of the whole.
What the cookbook changes: An inventory of agents and tracking of their behavior.
Executive view
Decisions gain method, indicators defined beforehand and a reading of the effect afterwards.
Recipes in the cookbook
- Discover source and qualify origin
- Publish a data product with a contract
- Certify metric and semantics
- Compose temporal decision context
- Declare agent, capability and limits
- Admit, block or degrade context
- Observe drift, quality and behavior
- Preserve evidence, lineage and learning
What your leadership receives
- Prioritized decisions, with an owner and a baseline.
- Recipes applied to your case, first in shadow operation.
- Indicators defined beforehand and measured afterwards.
- The proposal for the next step, with an exit criterion.
Indicators tracked
Quality, freshness, coverage, lineage, confidence, exception rate and decision value
Technical view
Which data goes in, how the recipe is organized and where it connects.
Data and source systems
- Data sources and catalog
- Metrics and definitions
- Quality and freshness
- Use of agents and models
Connections
Open to adapters and connectors with the ecosystems you already use, such as Snowflake, Databricks, AWS, GCP, Microsoft and others. Connections are defined in each project.
What comes with each recipe
- Problem and decisionGoal, context, alternatives and expected result.
- Data and indicatorsSources, definitions, quality and starting reference.
- People and agentsOwners, activities and approvals.
- Application and evidenceAdaptation roadmap, tests, exceptions and measurement.
How to start
From the first case to continuous operation, a ladder with an exit criterion at each step.
Organize and govern what already runs
Organization & Agents
Organize and govern the agents that already exist, each with an owner, an accepted limit and a record of every act.
Structure the first case
VÉRTICE Executivo
A critical decision is read end to end: data, owners and baseline. It ends with the chosen recipes and the proposal for the next step.
Apply a capability
Decision Sprint
One to three recipes run in shadow operation, beside the human decision, on the baseline.
Operate and follow
MDC · Managed Decision Cell
The recipes start operating under supervision, and the effect is tracked against the baseline.
Recipes guide adaptation and supervised implementation. Scope, integrations and acceptance criteria are defined in the proposal; examples and screens use synthetic data.
Which part of your operation needs to move forward?
Start with a recurring decision or an outcome that matters. Together we identify the data, people and first capability required.