Skip to content
Agent Canoe / AI edition
Agent Canoe AC markACagentcanoe
Databricks Genie Agent Now Handles Multi-Step Tasks With Extended Context
Depth
1 outlet · newest source 38h ago
Agent watch

Databricks Genie Agent Now Handles Multi-Step Tasks With Extended Context

Databricks updated its Genie AI agent to execute complex, unsupervised workflows using platform data, as demonstrated by the company’s chief executive.

Key points

  • The updated Genie agent can perform multi-step jobs that require extended time to complete.
  • Databricks built an ontology to provide the agent with necessary context from customer data.
  • CEO Ali Ghodsi used the tool internally to analyze survey data and prepare for interviews.

Databricks unveiled an updated version of its Genie AI agent on Wednesday, introducing capabilities that allow the system to execute multi-step jobs requiring extended periods to finish. This update enables the agent to operate unsupervised on complex tasks by leveraging a newly built ontology that provides context from information stored on the Databricks platform.Databricks updated its Genie AI tool to handle longer, more complex tasks automatically.Databricks updated Genie to support unsupervised, multi-step agent workflows via a proprietary context ontology.

How we got here

Databricks CEO Ali Ghodsi demonstrated the agent’s internal usage during a discussion with TechCrunch AI. He showed how he utilizes the tool to bypass the communication barriers inherent in large organizations, specifically referencing Dunbar’s number, which suggests humans can maintain stable relationships with only about 150 people.

Evidence: reportedDatabricks’ new AI agent tells CEO Ali Ghodsi the unvarnished truth

aifleet-techcrunch

What is new

The primary change is the agent’s ability to handle extended, multi-step workflows without constant supervision. Ghodsi illustrated this by showing chats where the agent analyzed parts of an employee satisfaction survey with low scores, helped rate and discard potential fixes, and summarized a failed sales pitch including specific feature requests from the prospect.

What to watch

The agent also prepared Ghodsi for a podcast interview by reviewing his previous appearances and developing likely questions. During the demonstration, the agent even provided advice on what Ghodsi should say to the reporter, showcasing its real-time analytical capabilities.

Availability and limits

Databricks is currently pitching Genie as an always-on agent for business leaders, but the source material does not confirm general availability for all customers. The demonstration relied on internal data, suggesting access may depend on existing platform data. Developers should note that the agent’s effectiveness depends on the quality of the underlying data and the ontology built by Databricks. There is no mention of pricing, API access, or specific limitations on task complexity beyond the “extended work” description.

Why it matters

  • Developers building on Databricks may need to consider how their data is structured for Genie’s ontology to function effectively.
  • The agent’s unsupervised nature raises questions about error handling and validation for complex, multi-step tasks.
  • The focus on “unvarnished truth” suggests a shift towards agents that can identify negative trends or failures without human prompting.
  • The reference to Dunbar’s number implies a specific use case for organizational intelligence in large teams.

Topics: AI agents, Databricks, enterprise AI, Genie

Updated

The strongest doubt

Counterpoint from the sources

The source material does not provide independent verification of the agent’s accuracy or reliability in unsupervised modes, relying solely on the CEO’s demonstration.

Ask the article

Local keyword search of this article; no generated answers.

Did this article give you the evidence you needed?

No vote saved on this device.
Your choice stays on this device. No public tally.

Questions readers ask

Can Genie agent perform tasks without human supervision?

Yes, the updated Genie agent can complete assigned tasks unsupervised, including multi-step jobs that take extended time.

What kind of data does Genie use for context?

Genie uses an ontology built by Databricks to access context from information customers store on the Databricks platform.

Is Genie available to all Databricks users now?

The source material states Databricks unveiled the update and is pitching it, but does not explicitly confirm general availability for all customers.

Connected stories

This storyOumi’s AI Engineer now builds agents and retrains their underlying modelsOumi’s AI Engineer now builds agLangChain hits $3B valuation after $300M CapitalG round for autonomous agentsLangChain hits $3B valuation aftMonday.com deploys internal expert unit to help clients realise AI agent returnsMonday.com deploys internal expeMeta Muse outbound calling opens to paid subscribers for US business contactsMeta Muse outbound calling opens

Reading receipt

  • The updated Genie agent can perform multi-step jobs that require extended time to complete.
  • Databricks built an ontology to provide the agent with necessary context from customer data.
  • CEO Ali Ghodsi used the tool internally to analyze survey data and prepare for interviews.
Time spent: 0:00 · Depth: Full

Sources

  1. TechCrunch AI

Related stories

Background reading