GLOBAL CONSULTING LEADER
Building a keystone product
for a global consulting leader
We worked with a multi-billion dollar consulting firm to ship their first AI product and open the door to a new revenue stream.
THE CHALLENGE
Unfamiliar territory
The SVP of AI at our client, a multi-billion dollar global consulting leader, saw a massive
market opportunity. If they could encode their deep expertise in an AI-powered application, it
would dramatically speed up project delivery and potentially unlock a new licensing revenue
stream.
But building AI applications was new territory for them, and their small-but-mighty team was having
challenges turning their compelling prototype into something ready for production.
With our thirteen years of experience productionizing AI systems, we stepped in to help.
MAXIMIZING ROI
Compounding
interest
Generative AI apps are often unreliable since the same inputs don't always return the same
outputs. This was a major pain point for the team, who had (understandably!) traded off
reliability to ship faster. The stakes were high: if these reliability issues couldn't be
resolved, the product would end up a failed experiment.
Our solution: create a “pit of success” where the most expedient way of writing code was also
the most resilient.
To that end, we decided to focus on DX investments that would compound across the entire life
of the project, mitigating the risk to the project while providing an outsized return on investment.
SPEED AND QUALITY
Pit of success
During our first week, we interviewed the team and audited the codebase. Our goal? Find
low-hanging fruit that would prevent bugs, speed up development, and improve UX.
Over three months, we incrementally migrated the app to a small core of sturdy UI components
specifically built for the challenges of generative AI (long latency, unreliable outputs).
The bottom line? We cut 6,000 lines of boilerplate while dropping page load times from 14 seconds
to two seconds. After release, we estimate preventable errors dropped by half.
REAL USER FEEDBACK
A virtuous cycle
Supported by our DX investments, the team's hard work and ingenuity propelled the system
from “exciting prototype” to “ready for business-critical applications.” This led to a
tipping point where real users started using the product and kicked off a virtuous cycle:
1. Direct feedback from internal users led to a more useful product.
2. A more useful product helped the client's internal teams work better and faster.
3. Which drove production adoption as more and more teams asked for access.
4. Which resulted in more feedback and more improvements, starting the cycle over.
EXCEPTIONAL RESULTS
Delivering a
dream outcome
By the time we wrapped up, the product had been deployed across twenty of our client's
engagements with dozens of consultants. As part of our work with those teams, we heard from
multiple individuals that it had saved them a week or more of work.
And the company was well-positioned to license their new software across their network of partners
and clients, unlocking a major new revenue stream with significantly improved margins.
SAME OR DIFFERENT
Staffing
for resilience
Consultancies have an incentive to paint a sanitized picture of their projects, but we
believe frank reflection is better at driving business value.
Our biggest takeaway from this project is how a project with a single team member has a single
point of failure. We learned this the hard way when our sole engineer had a housing emergency.
Our leadership team stepped in to fill the communication gap, but our ability to deliver for
that week really suffered. A second team member could've kept output up to our usual high standards
while the emergency got resolved.
