April 25, 2025
When McKinsey released its 2024 projections on how generative AI would reshape software, it was worthy of taking notice. It could also have been taken with a grain of salt as AI was still relatively new. The thinking? “Sounds big, but we’ve got more urgent problems: rates, volume, margins.”
But here’s the thing: the projections made in mid-2024 were too conservative. Gen AI didn’t trickle into enterprise workflows—it exploded. Software categories have since started collapsing. Vendors are being swapped like outdated plugins. And some lenders are already outpacing others by baking AI into their operations—not as a gimmick, but as a growth lever.
If you’re still evaluating AI like we’re in a discovery phase, it’s time to recalibrate. The future is moving faster than the mortgage industry is used to. And if you wait for perfect clarity, you’ll likely miss the window to lead.
🔁 Looking Back: The Forecast vs. Reality
McKinsey’s view in mid-2024 was already bold. But the speed of real-world change has outpaced even their most aggressive estimates. Here’s a quick summary of the projection versus reality:
1. Vendor Switching Hits Critical Mass
- Original Projection: 5–10 percentage point increase in switching due to gen AI.
- Accelerated Outlook: Expect 15–20+ point churn, particularly in commoditized software categories (CRM, customer support, BI tools).
- Implication: Vendors without embedded, differentiated AI features—or who fail to evolve UI/UX with gen AI—will lose relevance rapidly. Expect a flood of RFPs focused on “AI maturity” by Q2 2025.
2. Internal Builds Become the Default, Not the Exception
- Original Projection: 2–4 point shift from buying to building.
- Accelerated Outlook: Closer to 8–10 points, especially among mid-size to large enterprises building gen AI apps tailored to their unique data and workflows.
- Implication: ISVs need to position as platform enablers, not just product vendors. Tooling, interoperability, and support for enterprise AI builders become more valuable than finished apps.
3. Gen AI-Native Categories Emerge (and Replace Legacy Ones)
- Original Projection: Reimagination of software categories in progress.
- Accelerated Outlook: By 2026, legacy category boundaries (e.g., BI vs. CRM vs. ERP) blur. AI agents and embedded assistants will redefine value through workflows, not modules.
- Implication: Expect vertical-specific copilots, agents, and hyperautomation layers to eclipse monolithic platforms. Buyers will want modularity and orchestration—not suites.
4. AI Differentiation Shifts from Features to Models and Data
- Original Projection: Novel use cases and UI improvements via gen AI.
- Accelerated Outlook: Buyers will demand AI that’s fine-tuned on proprietary or vertical-specific data—not just shared foundation models.
- Implication: Value shifts from features like “chat with your data” to performance benchmarks on nuanced tasks (e.g., interpreting mortgage docs, forecasting retention, segmenting churn risk). Vendors that can’t demonstrate measurable, vertical-relevant AI intelligence will fade.
5. Economic Pressure Forces Consolidation and AI ROI Discipline
- Original Projection: Investment and R&D reallocation underway.
- Accelerated Outlook: By 2026, ROI from AI adoption will be non-negotiable. Budgets tighten, and buyers drop vendors who can’t quantify gains in throughput, time savings, or decision accuracy.
- Implication: Expect a consolidation wave. Many AI-first startups will fold or be acquired. Procurement checklists will prioritize proven benchmarks, not pitch decks.
🔍 What Companies Have Learned
- The “wait and see” crowd fell behind.
While some companies ran pilots and watched panels, others deployed AI to triage leads, classify docs, resolve disclosures, and slash time-to-close. Early adopters now have playbooks—and momentum.
Some parts of AI may not be ready for prime time, but that does not mean that all of AI is not ready. - AI isn’t a sidecar—it’s the new engine.
LOS bolt-ons aren’t cutting it. The winners are building with AI at the core. More and more, we are seeing new POSs, CRMs, and automation engines that are built from the ground-up to use AI.
Slapping a chatbot in a side window just doesn’t provide the uplift. - Control and trust matter more than hype.
The few companies who implemented off-the-shelf AI without governance learned the hard way: hallucinated disclosures, misclassified docs, and “black box” automation don’t cut it when regulators come calling.
Back-testing before going live is now mandatory to ensure AI’s guardrails really are as reliable as your vendor is promising.
🚦 Looking Forward: Strategic Moves for Lenders in 2025–2026
So, if we look back at McKinsey’s predictions and realize they were too conservative and that AI is displacing software vendors faster than expected and it is being adopted faster than expected, what conclusions might we draw about the near-term future?
✅ Do: Assume Every Mortgage Tech Stack Is Up for Reinvention
Your LOS, CRM, or doc processing platform isn’t a sacred cow. It’s a workflow enabler—and AI will rewrite how that workflow works. Be ready to rethink, not just “enhance.” If you are in the position to renew your license on any of your key platforms, you should not get sucked into a long-term agreement.. Ai will not replace your key systems in 2025, but it will not be shocking if the AI-native versions are in-market by Fall 2025.
✅ Do: Focus on ROI-Proven Use Cases
Lead engagement, disclosure validation, doc classification, and income analysis are low-risk, high-yield AI starting points. Tools like Vallia Lead Expeditor and DocFlow are delivering measurable ROI—not AI theater. Look beyond chatbots to more impactful solutions.
🧠 Conclusion: Lead Like It’s Already 2026
If you’re still shopping AI tools like you did CRM plug-ins in 2014, you’re not just behind—you’re at risk. Gen AI is shifting who wins and loses. Fast.
By the end of 2025, we’ll see:
- Major lender workflows rearchitected around intelligent agents
- Document tasks and borrower interactions fully automated
- Internal build vs. buy strategies reshaped by vertical-specific copilots
Those who wait will be stuck retrofitting yesterday’s tech to solve tomorrow’s problems. Those who move now will define what the new mortgage stack looks like.
📚 References & Supporting Sources
- Original McKinsey Article – “Navigating the Generative AI Disruption in Software”
- Published mid-2024
- McKinsey & Company
- Provides baseline projections on AI adoption, vendor switching, category disruption, and internal builds.
- Update on Gen AI Adoption Acceleration – McKinsey Global Survey 2024
- By 2024, 71% of organizations reported using generative AI in at least one function, more than double the 2023 rate.
- McKinsey: The State of AI 2024
- Generative AI Reshaping Competitive Moats in Software
- Business Insider: “Software companies are getting squeezed by AI’s rapid rise, warns AlixPartners”
- Describes margin pressure, increased switching, and the collapse of traditional software category advantages.
- Business Insider
- Internal Builds and Citizen Developer Momentum
- Rise of enterprise-grade gen AI tools enabling non-technical teams to build in-house solutions.
- Also emphasized in McKinsey’s 2024 outlook and recent commentary on in-house copilots and workflows.
- Microsoft Azure AI & Brimma’s Mortgage-Focused AI Stack
- Vallia solutions by Brimma (DocFlow, Lead Expeditor, Data Connect) are layered on Microsoft Azure AI and designed specifically for lending workflows.
- Internal documents: “Brimma AI Insights & Strategy,” “Vallia DocFlow from Brimma,” and “VallIA – Intelligent Assistant.”
- Limitations of AI Governance & Platform Moderation
- MIT Technology Review: “Why AI still can’t moderate content” (2023)
- Brookings Institution: “The Challenges of Regulating Emerging Technologies”
