Search
Getting Started
Feature guides
Integrations
Community support
For Pro+ users, you can toggle from the default chat mode to the advanced mode using the latest versions of frontier models like the Claude Sonnet series.
Dreambase runs on more than one model. Every dashboard, report, and Skill you build is powered by a full analytics harness, a suite of models from multiple providers that we continuously evaluate and route work across automatically. You never pick a model. Dreambase picks the right one for the job, and Pro and Team plans unlock a deeper bench of models as your data and questions get more demanding.
The analytics harness
We don't treat models as interchangeable. We classify every model we support, across every provider, on the dimensions that actually matter for analytics work: reasoning effort, creativity, analytical rigor, context window, and query accuracy. Then we assign classes of comparable models, regardless of vendor, to the specific Dreambase agents and subagents that do the work: the Data Engineer Agent (plus its data science and engineering subagents for schema optimization and the data dictionary), the Analyst Agent, and the Business Insights subagents.
This harness is evaluated continuously, not set once. As providers ship new models, we re-test them against our own analytics benchmarks and roll the best ones into the harness, so the quality of your dashboards and reports improves over time without you doing anything.
Data Engineer Agent
The Data Engineer Agent builds your actual data pipeline: connecting Postgres, APIs, and MCP connectors, defining KPI Skills, and optimizing schema and the data dictionary that every dashboard draws on.
All plans get a strong, comparable model class here, currently models like GPT 5.5 (medium reasoning) and Gemini Pro, which handle the majority of pipeline and schema work reliably.
Pro and Team add Anthropic's Sonnet and Opus into the mix for this agent, for the harder schema optimization and data modeling problems where that extra reasoning class pays off.
Analyst Agent
The Analyst Agent is the one building your charts, dashboards, and exploratory analysis, and creativity matters more here than raw rule-following.
Sonnet and Gemini Pro often out-create GPT for this kind of open-ended visual and analytical work, so Pro and Team get those models in the Analyst Agent's rotation.
Every plan still gets a capable Analyst Agent. Paying plans get the more creative end of the harness.
Business Insights
Business Insights answers the questions that can't afford to be wrong, like revenue. This is the one place where Anthropic's models can actually be too creative for the job, so we don't use them here. Instead we run a custom harness: a blend of Gemini Flash business-insight subagents feeding into a blend of GPT 5.2, 5.4, and 5.5, tuned specifically for accuracy over creativity.
What this means for you
Every Dreambase user, on every plan, gets a genuinely good data experience out of the box. As your data grows more complex and your questions get harder, Pro and Team plans open up the deeper, more capable end of the harness across the Data Engineer and Analyst agents, so the experience scales with you instead of hitting a ceiling.
Dreambase runs on more than one model. Every dashboard, report, and Skill you build is powered by a full analytics harness, a suite of models from multiple providers that we continuously evaluate and route work across automatically. You never pick a model. Dreambase picks the right one for the job, and Pro and Team plans unlock a deeper bench of models as your data and questions get more demanding.
The analytics harness
We don't treat models as interchangeable. We classify every model we support, across every provider, on the dimensions that actually matter for analytics work: reasoning effort, creativity, analytical rigor, context window, and query accuracy. Then we assign classes of comparable models, regardless of vendor, to the specific Dreambase agents and subagents that do the work: the Data Engineer Agent (plus its data science and engineering subagents for schema optimization and the data dictionary), the Analyst Agent, and the Business Insights subagents.
This harness is evaluated continuously, not set once. As providers ship new models, we re-test them against our own analytics benchmarks and roll the best ones into the harness, so the quality of your dashboards and reports improves over time without you doing anything.
Data Engineer Agent
The Data Engineer Agent builds your actual data pipeline: connecting Postgres, APIs, and MCP connectors, defining KPI Skills, and optimizing schema and the data dictionary that every dashboard draws on.
All plans get a strong, comparable model class here, currently models like GPT 5.5 (medium reasoning) and Gemini Pro, which handle the majority of pipeline and schema work reliably.
Pro and Team add Anthropic's Sonnet and Opus into the mix for this agent, for the harder schema optimization and data modeling problems where that extra reasoning class pays off.
Analyst Agent
The Analyst Agent is the one building your charts, dashboards, and exploratory analysis, and creativity matters more here than raw rule-following.
Sonnet and Gemini Pro often out-create GPT for this kind of open-ended visual and analytical work, so Pro and Team get those models in the Analyst Agent's rotation.
Every plan still gets a capable Analyst Agent. Paying plans get the more creative end of the harness.
Business Insights
Business Insights answers the questions that can't afford to be wrong, like revenue. This is the one place where Anthropic's models can actually be too creative for the job, so we don't use them here. Instead we run a custom harness: a blend of Gemini Flash business-insight subagents feeding into a blend of GPT 5.2, 5.4, and 5.5, tuned specifically for accuracy over creativity.
What this means for you
Every Dreambase user, on every plan, gets a genuinely good data experience out of the box. As your data grows more complex and your questions get harder, Pro and Team plans open up the deeper, more capable end of the harness across the Data Engineer and Analyst agents, so the experience scales with you instead of hitting a ceiling.