The race to enterprise context: Enterprise change management in an AI-native world

A huge part of being a startup founder is an optimism that technology can shape the future for the better. In that respect, it has never been a better time to build a company. A few years after the ‘ChatGPT moment’, LLMs are everywhere, driving real value for workflows across industries, geographies, and company stages. 

Most of my time is spent with startup founders and other investors who share this perspective. However, technology startups represent a fraction of the economy in comparison. In my view, the next hurdle for AI is widespread adoption by legacy companies: massive, mid-market, and emerging. 

I meet many founders trying to tackle this value pool: selling generative solutions to old-school businesses. From these conversations, I’ve heard something curious. Founders tell me that technology’s sophistication doesn’t always determine success when selling to this category. 

I wanted to dig deeper, so I spent time with operators from a previous generation of AI: the automation leaders at major technology outsourcing firms and legacy robot process automation platforms. Learning from the past is an important input in understanding the future.

They consistently told me that the human element can make or break deployment success, regardless of how brilliant a piece of technology may be.

In this blog, I outline some core learnings from these conversations. Specifically, I focussed on the legacy enterprise as a buyer, aiming to understand how they procure and adopt automation solutions. I also touch on potential budget pools in the hopes it informs builders rethinking these problems with AI.

Half know what they want; Half need your help

Automation engagements inside legacy organisations typically follow one of two paths: clearly defined or exploratory. This fundamental split shapes everything that follows.

"About 30-35% of our business comes from clients who know exactly what platform they want," explained a senior director at a multi-billion dollar technology outsourcing firm. These "boxed conversations" typically conclude within nine weeks—a clean, straightforward sales process.

The other half begins with a problem rather than a solution.  A director at a leading global consulting company observed: "A general problem statement common among clients is: 'I want to reduce X in this department, and I don't know where to start.'"

I believe this divide is an important insight for go-to-market strategies. The "solution-ready" clients might be easier sells, but the "problem-aware" segment often yields deeper, more valuable relationships as you help them discover where automation creates genuine value.

My colleague Lucy Tan uses the phrase ‘the race to enterprise context’ to explain this paradigm. Becoming a technology solution of choice doesn’t just require products to automate reliably, affordably, and with a high degree of quality. It will require vendors to understand enterprises’ challenges in a nuanced and human-led way.

The automation value curve: Easy wins vs. real transformation

Simple processes automate easily, while transformational workflows demand expertise and patience. A former VP of Sales at a major RPA platform told me, "Every customer went into it thinking they had processes ripe for automation... However, these processes weren't always the most valuable. Organisations hit a wall post-initial automation phase."

I’ve seen this first-hand from my work with Steve Hind and Jamie Hall, the co-founders of Lorikeet, the leading customer support automation platform for complex workflows. Steve and Jamie’s view is that customer support costs aren’t distributed by the frequency of an issue type but by the complexity of issue types. Companies stand to save a lot more money by automating the 20% of most complex tickets rather than the 80% of tickets fairly simple.

Leading platforms address this through industry specialisation and pre-built templates, but even with these resources, complex automation requires substantial discovery work and, as another director at a global consulting firm put it, "a heavy degree of handholding."

The four horsemen of successful implementation

Four consistent barriers determine whether automation thrives or withers in enterprise environments: pace, politics, access and integrations, and ongoing maintenance. 

1. Pace

Business processes evolve faster than traditional IT projects can be delivered. When automation requires a three-month sprint, business users won't wait—they'll just change the process or deploy solutions themselves using SaaS products and their corporate cards.

An experienced RPA sales leader explained to me: "In some cases, applying traditional IT project management methodology means a 3-month sprint, which is too long in the automation world. When business users need help automating a function and have to wait three months, they'll change the process themselves."

By the time a process is automated, it’s been changed by the operational teams driving it. 

2. Politics

People I spoke to mused that many early RPA conversations focused on teams’ fears about job displacement. But today, resistance comes from protecting departmental territories. "Resistance typically comes from managers responsible for the process," a consulting director told me, "not from frontline workers."

This same consultant elaborated: "There is political resistance we get from process owners - not the junior or mid-level employees, but the manager or senior manager who is responsible for the process... Sometimes, they intentionally delay because they know if an idea gets automated, they may lose team members."

Organisational change management becomes especially critical in distributed organisations. Vendors cannot be solely responsible for change management, particularly with new clients where developing institutional knowledge takes time.

3. Access and integrations

Even experienced teams inevitably hit integration hurdles, especially in industries with specialised, closed systems. Companies in insurance and financial services, for example, use specific tools and platforms. They're not as accessible as major enterprise products, so integrating requires meaningful lift.

My sense is there will be real value in solutions with robust, pre-built connectors for industry-specific platforms. You can imagine a world where a horizontal automation product excels for having deep, industry experience context from day one as part of their go-to-market strategy. 

4. Ongoing maintenance

Traditional UI-based automations break when interfaces change. A consulting director explained to me: "Changes in policy systems, upgrades, regulatory shifts, and operational aspects can disrupt automated workflows’ performance”. For example, he shared how altering buttons in a purchase order workflow in a legacy enterprise system can completely break a workflow. 

Hopefully, generative solutions can understand these changes, smoothing out the frequency of error rates. For example, one of our portfolio companies, Mindverse, is building a large action model to help workflows automatically adapt as this ‘enterprise context’ shifts. 

And more generally, processes change all the time. New managers join, alter standard operating procedures with new policies or steps, and suddenly old workflow automation isn’t fit for purpose.

Navigating enterprise budgets

Success requires understanding who controls budgets and how they're allocated for larger automation transformation projects. My conversations revealed these key value pools:

• CIOs, CISOs, and IT departments typically hold the purse strings
• Finance functions often provide an effective entry point, with CFOs "more comfortable moving ahead without IT's blessing"
• Budget maturity varies widely, from enterprises with $10M+ dedicated automation budgets to those funding from general IT resources

Perhaps most tellingly, customers frequently underestimate the services required for successful implementation. As a former RPA platform executive observed: "A lot of customers in the early days were like, “Yeah, sounds great, we'll take 100 workflows.' Then, a year or two into their journey, they had rolled out five."

Where do we go from here?

The race to enterprise context isn't just about having superior technology—it's about understanding the human organisations that technology serves. In the end, AI founders who master this will be the ones who transform industries, not just processes. My sense is these founders will share a mix of similar qualities:

1. They embrace a consultative sale. The most valuable enterprise relationships begin with problems, not solutions. Founders must develop the patience and expertise to guide the "problem-aware" segment through discovery rather than just pursuing quick wins with "solution-ready" clients.
2. They can target budget holders with precision. Successful founders identify and speak directly to the economic buyers—often CIOs, CISOs, and increasingly CFOs—with clear ROI frameworks that address both immediate value and long-term transformation. They’ll be able to articulate both top-line growth opportunities from automation and bottom-line cost savings. 
3. They will explore channel sales strategy. The reality is that enterprises trust their existing partners. AI founders who build alliance networks with systems integrators and consultancies gain trusted pathways into organisations that would otherwise take years to penetrate. Channels sales was once an ‘icky’ word in the startup ecosystem, but I wonder whether it gains saliency in this new era.
4. Build for maintenance from day one. The "set it and forget it" mentality doesn't work in enterprise environments. Solutions must adapt to changing UIs, evolving business processes, and shifting regulations. This capability should be core to product design, not an afterthought.

I’d love to speak to founders who strongly disagree with this or founders building solutions capable of driving meaningful change inside legacy enterprises. Get in touch: jethro@squarepeg.vc.