building with ai // agents and gemini api @google // ex-startup founder, @datadoghq // my own thoughts
Bay Area
Joined October 2016
- Tweets1.1K
- Following590
- Followers997
- Likes4K
Pinned Tweet
I’ve recently joined Google to work on the Gemini API, focused on helping enterprise customers build and deploy agents in production. Excited to spend time with customers and improve the product alongside @siddharthbhai @alihcevik @mcocirio @OfficialLoganK
OpenAI models are now available alongside Anthropic and Google models on GCP. Big win for customers to get true model optionality, making GCP the first hyperscaler to support all three leading model providers on a single platform.
Today, we're bringing OpenAI models to Google Cloud's model garden.
We're now the only hyperscale cloud provider offering access to all three leading closed source LLMs—alongside 200+ other models—on a unified enterprise platform → goo.gle/4zJ0VZQ
Enterprise customers can now offload longer knowledge works tasks to the unified Gemini agent, with industry specific skills and tools, persistent memory, and support for third party models
Today at Google Cloud’s Gemini at Work event, we announced Gemini, a new single universal agent for work that has all of your business context and can be used for everything from knowledge work to answering questions, and content creation to coding, all from a single prompt box.
It’s built around core architectural principles:
☑️ Unified Agent: Gemini can answer questions, do knowledge work and generate code from a single prompt box.
☑️Access: Gemini is web-based and can be accessed from any device and integrated into third-party applications. It can also operate without a dedicated user interface.
☑️Persistent Execution: It runs in the cloud, which means it maintains a single set of memories, context and one personalization graph no matter where you access it.
☑️Multi-Agent Orchestration: Gemini can create sub-agents to tackle multi-step tasks and can also act as a coworker agent with a defined role and its own dedicated identity
☑️Context: Gemini knows your tools, data, and work history and learns how you work the more you use it.
☑️Model Choice Flexibility: It orchestrates across multiple models to deliver optimal quality and lower your costs.
Can’t wait for customers to get a hold of Gemini 4 Argon! It’s massively improved my daily long running knowledge work
Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon!
It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback.
Here’s a look at the benchmarks:
Stop resending conversation history, enterprise customers can now use the Interactions API to build stateful agents and conversational apps with 7 days of built in context
You can now use the Interactions API on Gemini Enterprise Agent Platform (prev. Vertex AI) with Gemini models. Same `google-genai` SDK.
👉 docs.cloud.google.com/gemini…
Personal agents monitoring meeting notes are exposing how much of work is just following up on things people already said they’d do. Starting to see agents take on a lot of this follow through work companies have historically built entire roles and processes around.
Had fun chatting with Alex at The New Stack about how AI is changing observability workflows across proactive detection, incident investigation, and automated remediation
AI-powered observability is changing how teams investigate incidents, correlate signals, and resolve issues in production. Watch the on-demand session with @vigneshp_ of @datadoghq to see how these workflows operate in practice.
View now: bit.ly/43ln4PI
Bits AI is now powered with Datadog AI credits meaningfully lowering the price per investigation
Bits AI 周りの課金体系変わってる！
Bits AI SRE 含めて AI クレジット課金になり、Bits Chat (Bits Assistant) は利用するツールごとに単価が変わるけれど、比較的割安な単価が設定されてそうに見える👀
#Datadog
datadoghq.com/pricing/?produ…
Boy math is realizing one night out costs the same as Claude Max and deciding the subscription is basically free
Incredible milestone for PathAI to be acquired by Roche. Over the past decade, PathAI helped push digital pathology from an emerging idea into real workflows used by pathologists to improve patient outcomes at scale. Grateful I had the chance to help contribute to this vision.
Switzerland's Roche agrees to acquire US-based PathAI reut.rs/4whyxgB reut.rs/4whyxgB
Datadog’s Anomaly Reasoning Framework benchmark, built on real incident data, shows a pattern of humans and AI failing in different ways, and the best results come from combining domain expertise with model reasoning
The best AI models can only answer 63% of basic questions about production outages. Your on-call engineers get 72%.
But here's the twist: they fail on completely different questions. Combine them and you hit 87%.
We built a benchmark to measure this, and a new model (Toto-1.0-QA-Experimental) that matches GPT-5 at a fraction of the cost. Great work from @stephofx and team!
🔗 👇
Growing AI usage has made me realize an uncomfortable truth that most job functions never really had rigorous review processes, unlike software engineering where code review is built into the workflow. For PMs, data analysts, and content writers, work is often trusted by default, lightly spot checked, or reviewed through subjective stakeholder feedback. Now with AI, I’m realizing we don’t even have clear ways to evaluate what these roles produce in the first place.
As AI gets adopted inside companies:
2025: docs everywhere
2026: prototypes everywhere
2027: feature flags everywhere
Bits Assistant makes searching and analyzing over your observability data much faster than manually aggregating and correlating data across different datasets
