This quickstart shows the core loop in TrueState: add data, ask the agent to analyse it, turn useful work into reusable assets, and create a job for work that should happen again.
1. Sign in and choose an organisation
Go to the Platform and sign in. If you belong to more than one organisation, choose the organisation you want to work in.
Use the avatar menu in the header to switch organisation, workspace, appearance, or environment.
2. Add or choose data
If your organisation already has datasets, open Datasets and choose one that is relevant to your question.
If you need to add data:
- Use Datasets for a one-off CSV or Excel upload.
- Use Connections or a pipeline integration step for external systems such as cloud storage, databases, business applications, Google Sheets, or REST APIs.
- Use Secrets when an integration needs credentials.
See Loading Data and Integrating Data for the detailed ingestion guidance.
3. Start a conversation
Open Conversations or use the main chat surface. Add the relevant dataset, dashboard, report, pipeline, document, or model to the conversation context from the message composer.
Then ask a focused question:
Analyse the sales dataset. Show the main drivers of revenue growth, call out anything unusual, and suggest what the team should investigate next.
The agent can query data, create charts, write commentary, inspect platform assets, and suggest follow-up work. If the first answer is broad, ask for a narrower follow-up rather than starting over.
4. Save the useful output
When the analysis becomes something other people should reuse:
- Create a Dashboard for interactive monitoring or exploration.
- Create a Report for a document-style narrative.
- Create or update a Pipeline when the work should be repeatable against refreshed data.
- Add durable business knowledge to Context when the agent should remember it for future work.
5. Create a job for recurring agent work
Open Agent → Jobs when you want an agent to perform a defined piece of work on demand or on a schedule.
Example job instruction:
Every Monday morning, review the executive revenue dashboard, identify the three most important changes since last week, and update the commentary box with concise bullets for the leadership team.
Jobs have:
- A reusable instruction
- An assignee agent
- Optional context assets
- Optional schedule
- Run history so you can inspect what happened
6. Improve shared context
Open Agent → Context to review or add durable knowledge:
- Business Model: how the organisation works
- Data Map: how business concepts connect to datasets
- Feature Catalog: metric and calculation definitions
- Use Cases: opportunities worth exploring
- Conventions: preferences and rules agents should follow
- Skills: repeatable methods for specialised work
Good context reduces repeated explanation and helps agents avoid redundant or silly recommendations.
Next steps
After this first loop, keep building in the direction of your work. Conversations are best for exploration, Jobs are best for repeatable agent work, and Context is what helps Dawn avoid starting from scratch each time.