United StatesTechnology and telecom2026 – ongoingBlend Labs
Background agents that take the first pass on engineering tickets
Blend Labs said internal AI agents lifted engineering output by more than 1.5x measured in pull requests
Problem
Blend's engineers spent time on routine tickets such as bug reports arriving in the support queue. The company wanted more engineering output without a matching rise in headcount.
Approach
Blend built internal AI agents, which it called Blend background agents, over a few months. An agent picks up a ticket, for example a bug report from the support queue, finds the bug and prepares the fix for a human to approve. The agents can reach internal tools and the codebase. The CEO said the company also wanted accounting work prepared by agents before the accounting team picks it up.
Results
Takeaway
Blend measured the gain by counting pull requests, which shows volume of work and not its quality. Use of the agents outside engineering was still an aim at the time of the call.
Categories
- Industry
- Technology and telecom
- Business function
- IT and R&D
- Innovation type
- Process innovation
- AI technology
- AI agentsGenerative AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
Transcript of Blend's first-quarter 2026 earnings call. CEO Nima Ghamsari described the background agents built over the previous few months and an example of an agent handling a bug report from the support queue. He gave the productivity figure of more than 1.5x based on pull request counts and spoke of extending the approach to accounting work.
Full details are in the original sources.