Airbyte

Radar Signals

14 linked signals.

• Keyword search misses "this feels steep" when you're looking for pricing concerns. It's not an issue with Airbyte Agents. 🔎 Semantic search in Airbyte Agents now covers @Gong_io + @Linear, consuming up to 80% fewer tokenAirbyte · 3d
• A custom @actblue connector built in ~1 week vs. an estimated month from scratch, thanks to Airbyte's Connector Builder. @ComTechAlliance runs durable data infra for political orgs where full pipeline transparency isn'tAirbyte · 12d
• Moving from Airbyte OSS to Cloud no longer means re-syncing from scratch. We built tooling that transfers your connections and sync state, so Cloud picks up exactly where OSS left off. https://airbyte.com/blog/airbyte-osAirbyte · 13d
• The new @HubSpot + @LinkedIn Ads connectors for Airbyte Agents let you build agents that execute actions, not just fetch. http://airbyte.com/blog/hubspot-linkedin-ads-agent-connectorsAirbyte · 4d
• Per-instance retry logic can't see what other agents already consumed. The quota belongs to the org, not the session. Airbyte CEO @MichelTricot wrote about how shared pools + backpressure fix it: https://bit.ly/4x3qbJg 🔍Airbyte · 5d
• There's nothing like summer in San Francisco.Airbyte · 17d
• Approval gates in agent pipelines aren't a UI feature. They're a distributed systems problem. Durable state, async queue, timeout, confidence threshold. Make sure to build them in the right order. New piece by Airbyte CEAirbyte · 18d
• Two years ago, RAG was the answer to everything. At its core, it was a token optimization. Our CEO @MichelTricot dug into what comes next at the @tokensandai engineering summit: shaping context before it hits the model.Airbyte · 19d
• "If an agent is running, it has to be looking at the most recent data available. No exception." Michel Tricot on @tfir_io on what production agents demand from their data layer. Latency is a business decision. FreshnessAirbyte · 6d
• Agents burn most tokens on discovery, not answers. Airbyte's Context Store cuts that: 40% fewer tool calls, up to 80% fewer tokens. Here's how to wire a @SlackHQ agent with @LangChain. https://airbyte.com/tutorials/reducAirbyte · 21d