ISAAC WOOTEN
← All work

Google

Twelve years designing Google Cloud, and today I lead the team behind BigQuery’s AI-native tools.

7
Products designed from zero
Sr. Staff
UX Design Manager
5 → 11
Team grown, at peak
14
Features to GA in three quarters
Overview

BigQuery is where companies keep their data, and getting answers out of it has always meant hand work: SQL for every question, a pipeline for every new source, time spent hunting for nulls and PII. I lead the design team building agentic BigQuery, so engineers and business users alike can ask their data a question in plain language and let an agent do the heavy lifting around it.

Result

Agentic BigQuery, BigQuery Studio and the Data Agent Kit are live on a top-3 revenue product in Google Cloud. They were built by a team I grew from five to eleven, and the patterns we set have been adopted across the org. Further back, Cloud Logging, which I designed from zero in 2014, is among the most used products in the console today.

Designing agentic BigQuery

Getting an answer out of a warehouse shouldn’t require writing SQL.

BigQuery’s promise is one place for all your data: connect your other sources to the warehouse and everything lives together. The cost has always been the hand work in between. Every question meant a SQL query written by hand; every new source meant a pipeline built by hand; every messy table meant hunting for nulls, PII and inconsistencies yourself.

Agentic BigQuery lets people talk to their data. Ask a question in plain language and get an answer, and let the agent do the heavy lifting around it: migrating data, setting up pipelines, cleaning tables and flagging PII, nulls and inconsistencies before they cause trouble. The goal is that technical and non-technical people alike can get from a question to an insight without writing a line of SQL.

Before

  1. Write the SQL by hand
  2. Build each pipeline yourself
  3. Hunt for nulls, PII and bad rows
  4. Get an answer, if you know SQL

With the agent

  1. Ask in plain language
  2. The agent plans the query or pipeline
  3. You review what it proposes
  4. It runs, and flags what looks wrong

You

  • Ask the question, in your own words
  • Decide what matters and what the answer is for
  • Review and approve changes to data and pipelines

The agent

  • Turn the question into SQL and run it
  • Migrate data and set up pipelines
  • Clean tables, and flag PII, nulls and inconsistencies
  • Bring other sources into the warehouse

The hard part: one product has to serve the engineer who lives in SQL and the business user who has never written a query. The same question has to come back at the right depth for each of them, with the agent’s work visible to the one who wants to check it and out of the way for the one who just wants the answer.

How I led it

I took the team over at a hard moment. The org was mid-reorg, the team had lost people, and its work was invisible to the leadership deciding the roadmap.

Three things fixed it, and the order mattered.

1 · Direction

I ran a week-long vision sprint for a platform-unification effort with one job: a north star, named personas and real customer journeys that UX, product, engineering and docs had all agreed to. Agreement on the problem is the expensive part of design; once you have it, most of what follows is cheap.

A five-day vision sprint: UX, product, engineering and docs lanes converge on one north star, with named personas and real journeys.

2 · Visibility

I put the team's work in front of senior engineering and product leadership on a fixed weekly cadence, and set up a standing cross-functional group so information could move without me in the room. Design that nobody senior has seen does not get built, however good it is.

A standing cross-functional ring where information moves without a relay, beside a weekly review cadence after which design starts landing on the roadmap.

3 · An operating layer

Then the things that make a team run: one project tracker that engineering and product treated as the source of truth, engagement models so every partner team knew what support it had and how to ask for more, a shared capacity planner, and an onboarding document that made a new designer useful within days.

Four layers under the team's work: project tracker, engagement models, capacity planner and onboarding document.

The team grew, its work landed in the roadmap, and I stopped being the bottleneck. A team in trouble usually does not need better taste. It needs a direction it believes in and an audience that can see it.

Beyond my own org, I led a six-month design collaboration across BigQuery, the streaming products and the Cloud Setup team that removed real friction from onboarding and unlocked new roadmap areas and additional funding for a partner team.

Set a high bar and a sharp direction, then give designers the context and cover to do the best work of their careers.

How I lead

From question to answer

Agentic BigQuery, an illustrative render of the workflow. The shipped UI stays under NDA.

Products

Seven of these I designed from zero: Cloud Logging, Cloud Trace, Cloud Profiler, Cloud Debugger and Security Command Center in the Stackdriver years, then Managed Service for Apache Kafka and BigQuery Engine for Apache Flink.

Most of what my team is building now is unreleased: the agentic Data Cloud, and the future of BigQuery Studio, Conversational Analytics, BigQuery Graph, the Data Agent Kit and the Gemini Enterprise integrations. Below is every product I have led or shaped that Google documents publicly, linked to the real thing.

Now · Data Cloud, 2022 to present

BigQuery and BigQuery Studio

The console surface my team designs: Explorer, SQL, notebooks, data canvas, pipelines and Gemini assistance in one workspace over the same data.

BigQuery ↗BigQuery Studio docs ↗
BigQuery Studio today: the Explorer on the left, a What's new in Studio panel introducing chat with data, and Create new actions for SQL, conversation, notebooks, AI and knowledge, and data engineering.
BigQuery Studio today.
Animation: code composition, code completion and the chat interface with Gemini in BigQuery.
Code composition, completion and chat with Gemini in BigQuery. Google Cloud blog.
Animation: proactive agentic workflows in BigQuery, the agent suggesting and carrying out the next steps.
Proactive agentic workflows in BigQuery. Google Cloud blog.

Gemini Enterprise

Google's agentic platform for the workplace, and my team owns the part where it meets the Data Cloud, so answers and actions are grounded in BigQuery.

Gemini Enterprise ↗

Data Agent Kit for VS Code

Data agents inside the editor: plain-language questions, SQL or Python over BigQuery, and pipelines built and deployed without leaving VS Code.

Data Agent Kit docs ↗

Data Apps for Notebooks

A notebook analysis becomes a shareable, managed app: pick the cells, publish, and hand the business team something they can use instead of a screenshot.

Data Apps docs ↗

Managed Service for Apache Kafka, and BigQuery Engine for Apache Flink

The two streaming products I led design on from zero: Kafka end to end from MVP and now generally available, Flink serverless stream processing in preview. Both were built inside a large org with an existing roadmap, which is a different skill from founding something.

Managed Kafka ↗Flink announcement ↗

Dataflow and Pub/Sub

I designed more than twenty features across the two, fourteen of which reached general availability inside three quarters, plus the Pub/Sub and Managed Kafka integrations into BigQuery.

Dataflow ↗Pub/Sub ↗

Then · Cloud Diagnostics and platform, 2014 to 2021

Cloud Logging

Designed from zero in the Stackdriver years: a query, a histogram of volume over time, and the entries underneath. The 2019 and 2020 shots are as I shipped it; the first is the surface today. It rolled into Cloud Operations and is among the most used products in the console.

Cloud Logging ↗
Logs Explorer today: a query bar, a histogram timeline of log volume, and a results table of audit log entries.
Logs Explorer today. Google Cloud documentation.
Logs Explorer in 2019: the query builder, histogram and log entries in the earlier console.
Logs Explorer, 2019, as shipped.
Logs Explorer in 2020 with the histogram and expanded log entries.
Logs Explorer, 2020, as shipped.

Cloud Trace

Distributed tracing, designed from zero alongside Cloud Profiler: the trace list, the waterfall and the daily analysis reports, with the heatmap as the explorer today.

Cloud Trace docs ↗
Trace Explorer today: a latency heatmap over time above a table of spans.
Trace Explorer today. Google Cloud documentation.
A trace waterfall: nested Spanner spans with timings, from a 2016 build.
The waterfall, 2016.
Trace overview with insights, recent traces and a daily analysis report chart.
Overview and daily analysis reports, 2016.
The trace list: a scatter of request latency over time with a selected trace below.
The trace list, 2016.
Trace analysis reports: a table of system-generated daily reports and a custom reports section.
Analysis reports, 2016.

Error Reporting and Cloud Debugger

Error Reporting grouped, counted and trended errors so the one that mattered surfaced first; the Debugger, which I designed from zero, took snapshots in production without stopping anything.

Error Reporting docs ↗
Error Reporting in 2016: a table of error groups with occurrence sparklines, first and last seen, and resolution status.
Error Reporting, 2016, as shipped.
Cloud Debugger: source code with a snapshot taken in production and the captured variables on the right.
Cloud Debugger, 2016, as shipped.

Security Command Center

The security posture of a whole organization on one page, designed from zero while the product was still finding its shape. That v1 has since grown into a large suite of security products.

Security Command Center ↗
Security Command Center findings: a summary by severity with per-source panels below.
Findings, 2018.
Security Command Center assets: a filterable table of every resource in the organization.
Assets, 2018.

Network Topology

A live graph of the network and the traffic moving through it, with insights on the edges and a detail panel that turns a node into something you can act on. The same design later carried into Anthos Service Mesh.

Network Intelligence Center ↗Cloud Service Mesh ↗
Network Topology in Network Intelligence Center: a graph of regions, VPC networks and workloads with insights on the traffic between them.
Network Topology, the network visualised with insights. Google Cloud blog.
Network Topology egress traffic breakdown: a selected workload with its egress traffic split by destination.
Egress traffic breakdown. Google Cloud blog.
Network Topology with an auditing filter applied, narrowing the graph to the paths under review.
Auditing with a filter. Google Cloud documentation.

Onboarding and the setup checklist

Growth UX for the console: an organization's first hour, written as a checklist a new administrator could actually finish. A later six-month collaboration on this journey, across BigQuery, the streaming products and Cloud Setup, unlocked new roadmap areas and additional funding for a partner team.

Setup checklist ↗
The Google Cloud setup overview: a personalised checklist for setting up an organization, with steps for identity, resource hierarchy, access, billing and networking.
The setup overview, 2020.
Setup checklist step for Cloud Identity and organization, with guidance and a sign-in call to action.
Identity and organization, 2020.
Setup checklist step for planning and creating the resource hierarchy.
Resource hierarchy, 2020.

Product icons are Google Cloud's official icons. Screenshots are Google Cloud's, from its public documentation and from the product as it shipped at the time, framed here to show work I led. For the unreleased work I can talk through process, decisions and tradeoffs in a conversation, within what I am allowed to say. Book thirty minutes ↗.

Role

I have been at Google for twelve years, across three offices and seven teams, and went from individual contributor to design manager along the way. Today I lead the developer-experience design team in Data Cloud: BigQuery, Gemini Enterprise, BigQuery Studio, Databases Studio, the Data Agent Kit for VS Code and Data Apps for Notebooks.

My depth is data and big-data tooling, and at the center of it sits BigQuery, a Gartner Magic Quadrant leader and a top-3 revenue product in Google Cloud. Data is the substrate of modern AI, and the analysts and engineers who work with it are exactly who the agentic era has to serve well.

Still technical

I still read the code and the Figma before I weigh in. It makes my calls faster and more specific, and engineers can tell the difference.

Growing designers

I grew the team from five to eleven, and hiring was the easy half. The harder half is the mentoring and calibration that mean people leave more senior than they arrived.

Direction over decisions

I give the team sharp intent and a high bar rather than my sign-off on every screen. A clear direction scales to eleven designers; a queue outside my door does not.

Influence at scale

Design lead in an org of several thousand people, where nobody is obliged to take your advice. Landing work at that scale means building the case until the right thing is also the obvious thing.

Most of what ships is designed by the team, not by me: the surfaces have owners, and those owners make the calls. I keep the first pass at whatever does not exist yet. The team is six senior-and-above designers today, most with more than five years at Google, and in a discipline where two years is a normal tenure that retention is the number I am proudest of. Their names are theirs to publish; ask me and I will tell you who did what.

AI-native, every surface

Agentic tools have to follow the developer wherever the work happens, so we design for every surface they use.

Five surfaces on one agentic core. The developer moves across Web, IDE, Notebooks, CLI and API, and Mobile; the same agent follows.
Five surfaces, one agentic core. The developer moves; the agent follows.

Brand world

Google Cloud's brand DNA, locked and rendered across surfaces: product UI, out of home, data visualisation and the data story itself. Studies I made with my own tool, not Google's work, and the marks are Google's.

  • Google Cloud brand world. Hero: translucent data plates under the cloud markBrand world · rendered from Google Cloud's brand DNA
  • Google Cloud brand world. Data visualisation sheet
  • Google Cloud brand world. Mobile app screens
  • Google Cloud brand world. Presentation template
  • Google Cloud brand world. Billboard: All your data. More possibility
  • Google Cloud brand world. Website hero
  • Google Cloud brand world. Social kit
  • Google Cloud brand world. Icon system

Career arc

Pixels first, then patterns, then people. Twelve years, from IC to manager:

  1. Sr. Staff UX Design ManagerData Cloud UX from Jul 2026 · team of 11, 6 today
    Nov 2024–Present
  2. UX Team Lead / ManagerBigQuery UX & BigQuery Experience, lead · team grown to 11
    Mar 2024–Nov 2024
  3. Staff UX Design LeadData Analytics UX, lead
    Oct 2023–Mar 2024
  4. Staff UX DesignerData Analytics · Streams & Lakes
    Feb 2022–Oct 2023
  5. Staff Interaction Designer / UX EngineerGoogle Cloud
    Jul 2018–Jun 2021
  6. Staff UX ArchitectGoogle Cloud Platform
    Jan 2016–Jul 2018
  7. Senior Interaction DesignerStackdriver · Cloud Diagnostics
    Mar 2014–2016

The team ran at eleven from 2024 until November 2025 and is six senior-and-above designers today, after company-wide layoffs. The months between 2021 and 2022 are LunarCrush, where I went full-time as Chief Design Officer before returning to Google in February 2022 on the streaming products. Lead of the Data Analytics UX team from October 2023, of BigQuery UX from March 2024, and of Data Cloud UX from July 2026.

Scope

Products I led or oversaw design for, with research, engineering and product.

Data & analytics

  • BigQuery
  • Gemini Enterprise
  • BigQuery Studio
  • Agentic BigQuery
  • Databases Studio
  • Data Agent Kit (VS Code)
  • Data Apps for Notebooks
  • Dataflow
  • Dataproc
  • Dataform
  • Dataplex
  • Composer
  • Pub/Sub
  • Kafka
  • Flink
  • Spark

Observability, security & platform

  • Cloud Trace
  • Cloud Profiler
  • Cloud Debug
  • Cloud Logging
  • Error Reporting
  • Cloud Security Command Center
  • Network Topology → Anthos Service Mesh
  • Terraform Blueprints & onboarding
  • Critical User Journeys (GCP)
  • Perceived-performance patterns
  • 3rd-party auth & data regionalization

Capabilities

  • Design leadership & management
  • UX architecture
  • Developer experience (DX)
  • Data & AI product design
  • Design patterns & systems
  • Cross-org strategy
  • Cloud & enterprise UX
  • Mentorship

What I’d tell you over coffee

A team in trouble rarely needs better taste.

It needs a direction it believes in and an audience that can see it. Get everyone to agree on the problem first, because that is the expensive part; after that, most of design is cheap. Then put the work in front of the people who decide the roadmap, every week. The process comes last. Direction, visibility, process, in that order.

Next

The full role history lives on LinkedIn.

Connect on LinkedIn ↗