Google CEO Sundar Pichai acknowledged that the company is “a little bit behind” on the agent-coding frontier.
Pichai called coding “very fundamental” to Google’s AI work. He made the comments on the New York Times Hard Fork Podcast. The interview took place a few days after Google’s I/O developer conference.
Where Google sees the gap
Asked about Google’s position in the AI race, Pichai highlighted its strengths and where Google falls behind.
Google’s models perform “very well” in text, multimodality, voice, audio and reasoning, he said. But when it comes to agent coding, tool usage, following instructions, and long-term tasks, Google is “a little behind at the moment.”
On what that looks like for developers, Pichai drew a clearer line. Google has succeeded in creating unique web front-ends. The gap lies in longer-running tasks, where developers work on complex code bases.
Pichai said:
“There is a gap from the border where others are, but we are working, you know, we are well aware of that.”
Why Pichai says Google lacked coding data
Pichai pointed out a gap in the developers’ products. Google didn’t have the same external coding product surface generating data feeds for developers, he said.
He cited Anthropic’s relationship with Cursor as an example. Google “maybe didn’t have the space” that its competitors had, he added.
This is changing. At I/O, Google announced Antigravity 2.0 as a standalone desktop application for agent-based coding workflows. Internal usage at Google is growing rapidly, according to Pichai.
He added:
“We’re doubling every week and people are really putting the models to good use. It’s helping us move up the ranks quite a bit.”
During the I/O keynote, Pichai had shared the internal token usage numbers. He called the growth unlike anything he had seen at the company.
What Pichai said about Gemini 3.5 Flash
The interview took place a day after Google launched Gemini 3.5 Flash and made it the default model for AI mode globally. Pichai acknowledged early complaints about pricing, model quality and usage limits.
Google had tightened usage limits at launch to prevent outages, he explained, calling the restrictions “rightly a source of frustration.” The company is said to be making progress on the boundaries “very soon”.
Regarding the quality of the model, he acknowledged that the new model could have regressions in certain areas. Some issues are “easy to fix” with post-training, and Google will fix them quickly, he added.
SAY covered the Gemini 3.5 Flash launch and other I/O announcements earlier this week.
Why it matters
Pichai’s comments go further than Google’s keynote address about the company’s delays. During the event, the focus was on Gemini 3.5 Flash and Antigravity. This interview offered a more candid reading of the competitive situation.
Pichai described this discrepancy as a feedback loop problem. The coding products that developers use every day generate interaction data that improves the next model. By its own account, Google is currently building this loop through Antigravity after lacking a similar developer-facing surface.
Looking to the future
Pichai said Google is making strides in coding and called the space “very dynamic.” Google says Gemini 3.5 Pro is in use internally and is expected to roll out next month. It was not said whether the model would reduce the coding gap described by Pichai.
Featured Image: YouTube / Hard Fork Podcast
