Agents
Altertable's agents are automation workers for recurring analytics tasks.
They help with source synchronization, modeling, querying, and insight generation while your team keeps control of goals and business context.
With your guidance, agents can synchronize sources, model datasets, write queries, and generate insights or dashboards. They can also monitor metrics, segment users, and run scheduled follow-up checks.
Agents operate asynchronously and can run continuously alongside human workflows.
At their core, agents orchestrate multiple large language models (OpenAI, Anthropic, Google, Mistral) through a unified, asynchronous job system.
How agents stay aligned
Agents work best when they have both explicit instructions and learned context.
- Organization context: describe your team, products, audience, and internal language in organization settings so assistants understand how your business talks.
- Knowledge base: save notes and files that assistants can search when they need durable reference material such as policies, terminology, or planning documents.
- Memories: agents accumulate observations and feedback over time, then retrieve the most relevant memories when they work on similar problems again.
This gives agents two kinds of grounding: knowledge you provide on purpose, and patterns they learn from actual usage.
Ongoing agent work
Agent work is not limited to one-off chats.
- Chats and recurring tasks can be titled automatically so longer-running work stays easy to scan.
- Task settings can include instructions, attached context, and triggers so an agent can run on a schedule or when related work changes.
- The result of that work often shows up as Notifications, which you can review, route to channels, and use as feedback.
Learn More
- Notifications: data findings revealed by agents
- Memories: knowledge that agents accumulate over time to improve their results
- Dashboards: combine insights and attach agents for monitoring
- Insights: visualizations and analyses that agents monitor