Story of Labelbox: Founder and CEO - Manu Sharma
Story of Labelbox:
Early Vision and Founding
In 2018, Manu Sharma, Dan Rasmuson, and Brian Reiger founded Labelbox in
San Francisco, aiming to simplify and improve data labeling for AI. The idea
was to create a collaborative platform where teams could efficiently annotate
and manage large datasets, including images, videos, text, and audio. The
founders foresaw the rise of AI and believed that high-quality training data
would become a cornerstone of the industry.
Recognizing the Market Gap
Sharma had observed that internal teams in companies like Planet were tinkering
with data labeling infrastructure, but no scalable solution existed. To
validate their idea, the team needed to understand whether AI experts outside
their circle would pay for a general-purpose labeling tool. They knew that
creating a product that could serve multiple use cases would be challenging but
potentially transformative.
Early Customer Discovery
The founders started by pitching their concept to a small group of AI experts
in San Francisco. Sharma designed mock-ups and sketches of the Labelbox
platform to demonstrate its potential. These early conversations provided
insights into the problems teams were facing, revealing both skepticism and
enthusiasm. Some experts doubted a general solution could work, while others
were eager to use any prototype available.
Expanding Feedback Channels
Once initial reactions were collected, the team broadened their conversations
to a wider audience. They reached out to technical leaders such as CTOs, VPs,
machine learning engineers, and product managers across organizations working
on computer vision applications. The goal was to gather structured feedback to
validate both the problem and the potential solution.
Problem and Solution Validation
Two key indicators helped the founders decide to move forward. First, problem
validation: multiple teams confirmed that existing tools were insufficient, and
internal solutions were inadequate. Second, solution validation: some teams
shared their internal tools and workflows, allowing the Labelbox founders to
map industry approaches and see common challenges they could address with a
general platform.
Crafting the Right Questions
Sharma emphasizes the importance of asking open-ended, guided questions during
early discovery. Instead of framing questions with yes/no answers, Labelbox
sought insights by letting users describe their challenges and workflows. This
approach helped the team uncover real pain points without imposing their
assumptions, ensuring the product addressed authentic needs.
Story of Labelbox: Founder & CEO - Manu Sharma#startup#age#net worth# investor#Labelbox story#AI experts and validates problem & solution#entrepreneur
The Moment to Build
The team received a pivotal signal during an in-person meeting with a CTO from
a major medical imaging company. The CTO offered to hire co-founder Dan
Rasmuson to enhance their internal labeling system, confirming a real demand
for Labelbox’s vision. This was the decisive moment for the team to begin
building version one of the platform.
Early Product Development
With feedback in hand, the founders developed Labelbox V1, focusing on
collaborative data labeling with real-time quality control and annotation
tools. Their goal was to accelerate machine learning workflows and improve the
accuracy of models by providing high-quality, structured training data. The
platform also incorporated algorithms to guide the annotation process and
reduce errors.
Scaling with AI and Talent
Labelbox has grown into a full-featured data platform, connecting global AI
experts through its Alignerrs network. These experts provide human evaluation
and bespoke datasets to support AI labs, allowing organizations to scale data
operations efficiently. The company has raised $188.9 million in funding from
investors like Andreessen Horowitz, SoftBank Vision Fund, and Gradient Ventures
to further expand its solutions.
Looking Ahead
Today, Labelbox continues to innovate in frontier AI, offering data labeling,
model evaluation, and enhancement solutions for research and development. The
company emphasizes high agency, technical excellence, and rapid execution as
core principles, creating an environment where teams can tackle the most
challenging problems in AI. With a growing client base and strong funding,
Labelbox is positioned to remain a leader in AI training data infrastructure.
Timeline of Labelbox’s Journey
- 2008–2011:
Manu Sharma earns a B.Sc. in Aerospace Engineering from Embry-Riddle
Aeronautical University.
- 2012–2014:
Completes Master’s in Aerospace, Aeronautical and Astronautical
Engineering at Stanford University.
- 2017–2018:
Sharma works as Product Manager at Planet, gaining insights into data
labeling workflows.
- 2018:
Labelbox founded by Manu Sharma, Dan Rasmuson, and Brian Reiger in San
Francisco.
- 2018–2019:
Conducts early customer discovery with AI experts and validates problem
and solution.
- 2019:
Key validation moment with medical imaging CTO prompts the team to start
building V1.
- 2020–2022:
Labelbox expands platform capabilities for collaborative labeling, quality
control, and AI-assisted workflows.
- 2022–2023:
Builds global Alignerrs network to provide expert human evaluation and
bespoke datasets.
- 2023–2025:
Raises $188.9M in Series D funding from top investors, scaling product and
global operations.
- 2025
and beyond: Continues advancing AI training infrastructure, enabling
teams worldwide to build better AI models efficiently.
