In the world of AI, one particular company has seen explosive growth. From 2023 to 2024, the company grew 9x from $100 million to $900 million in revenue. By the middle of 2025, it had already generated $1.8 billion. Is it an open-source foundation model like Mistral or DeepSeek? Or is it the latest voice model, like ElevenLabs or Hume? Actually, it’s a consulting firm founded in 1989.
Since 2023, Accenture has been building up its practice around AI implementations. Already, the company had a Data & AI practice that focused on improving data infrastructure. Since then, it’s expanded to include practices around “Reinvention Services” and an “AI Refinery” for supporting “business user-facing agent building.” As of August 2025, the consulting firm had $5.9 billion in AI bookings. Outside of new-age AI hyperscaler-esque foundation models like OpenAI and Anthropic, which hit $8.6 billion and $5 billion in run-rate revenue, respectively, Accenture has been one of the most successful companies to monetize the AI wave.
That hasn’t gone unnoticed. This week, General Catalyst, a venture firm turned “transformation company”, announced the launch of Percepta. While the firm refers to it as yet another transformation company, it is effectively an AI consulting company. While this may seem out of left field for a venture firm, it’s indicative of two massive trends: (1) the changing face of venture capital, and (2) the immense difficulty of implementing AI effectively.
Publicly Traded Venture Funds
In February 2025, news broke that General Catalyst was making progress towards an IPO. Not for one of its portfolio companies, but for itself. There’s some precedent, with PE firms like Blackstone and KKR being publicly traded. But this would be a first for a venture firm. And GC is clearly building up a diversified revenue base in order to shore up its appeal in the public markets.
In September 2023, GC hired David Breslin from First Republic to launch a private wealth management practice.
In January 2024, GC acquired a health system.
In June 2024, GC launched a credit fund.
In December 2024, GC launched a buyout fund.
Now, with the launch of Percepta, GC is clearly taking aim at the AI transformation waves that are driving billions of dollars towards a firm like Accenture. As GC explains:
“Business transformation with AI relies on three critical components: advancing the intelligence and foundational model layer, building the underlying data infrastructure, and enabling the workforce to adopt AI so teams and agents can work together. Percepta helps synchronize all three of these pillars.”
GC frames its approach as enabling its team of “researchers, engineers, and product managers” with both the technology within GC’s own portfolio and a purpose-built software platform. The reality in most AI implementations is that the biggest obstacles are (1) data quality and access, (2) selecting the right AI product, and (3) change management within a given user base.
In our recent deep dive, The Vertical AI Playbook, we tackled the reasons behind a fundamental disconnect between AI value proposition and implementation. In May 2025, Anthropic’s CEO Dario Amodei delivered a warning that “AI could… spike unemployment to 10-20% in the next one to five years”. Yet at the same time, 42% of enterprise AI initiatives were discontinued in 2024, up from 17% in 2023. In our piece, we unpacked what the limitations are and how some verticals could better take advantage of AI. But the reality is that AI represents an incredibly difficult thing for most enterprises to get right.
High Friction AI
There are three common approaches to tackling the massive obstacle that is deploying AI:
Implementation Services: e.g. Accenture supporting enterprise customers in leveraging a broad swath of AI products
Forward-Deployed Engineers: Popularized by Palantir, this is a similar playbook but leveraged specifically to lower the friction of adopting a specific solution
AI Dev Shops: Organizations built around leveraging AI to spin up specific products, but not necessarily around institutional change management and implementation as much as opportunistic product build-outs
Offerings like Percepta outline a robust umbrella of services, including a research lab to develop new foundation models, partnerships with AWS and Anthropic, and other aspects that sound like flavors of forward-deployed engineers or AI development. But when you look at the use cases the team is addressing, it is very Accenture-esque: modernizing fraud detection, optimizing supply chains, helping governments streamline licensing and permitting processes, and automating clinical workflows.
The reason organizations like Percepta will likely be valuable, or why Accenture is able to generate $5 billion just to help people use AI, is because there is still a fundamental disconnect between the capabilities of AI and the ease of use. While we continue to see data points like 10% of all firms using AI, or 20% of Honeywell’s code written by AI coding tools, what we see less of is the failed projects, the delayed timelines, and the added cost or complexity of attempting to use these tools. The way that the massive build-up of AI infrastructure ultimately makes sense is if AI gets fundamentally easier to deploy effectively, or we get dramatically better at setting it up.
