Zillow Offers, the company’s AI‑powered home‑buying, was shut down after purchasing thousands of homes at inflated prices. The model was designed to identify undervalued properties, but reportedly overestimated home values and made offers the market couldn’t justify, and that owners couldn’t refuse! What was meant to improve decision‑making instead created volatility, inconsistency, and operational risk.
The issue wasn’t the AI. It was the data the AI was built on.
When the foundations were historically biased, incomplete, or overly optimistic, the model amplified those weaknesses. Rather than correcting the problem, it surfaced the underlying complexity of the system it depended on.
The same dynamic is playing out across reconciliation and financial controls.
Companies must focus on data consistency and AI execution at scale
AutoRek’s 2026 Investment Capital Markets research found that 98% of capital markets firms are using AI, yet only 14% have fully integrated it into operational processes.
Many firms are introducing AI into environments where core operational processes are still heavily manual. According to the research:
- 53% still rely on spreadsheets for reconciliation
- 82% of reconciliation activity remains manual
- 41% cite data integration and compatibility as their biggest operational challenge
- 40% cite transaction volumes
- 38% cite regulatory compliance (rising to 42% among UK firms)
When foundations are inconsistent and data is fragmented, AI outputs become inconsistent too, just as Zillow Offers demonstrated.
Trust is where AI succeeds or fails
Zillow’s pricing teams quickly lost confidence in the model as inconsistencies piled up. The same pattern has emerged across financial services.
When an AI tool:
- mis‑prioritizes a break,
- flags exceptions it cannot explain, or
- produces outputs that feel unpredictable,
teams instinctively revert to manual review. AI becomes an additional verification step rather than a removed one. Instead of accelerating work, it slows it down.
AutoRek’s research reflects this erosion of trust:
- 36% cite data privacy concerns
- 31% cite dependence on data quality
- 29% cite regulatory compliance challenges
These are not ‘AI problems.’ They are foundation problems that AI makes impossible to ignore.
As Jim Sadler, AutoRek’s Chief Transformation Officer, notes:
“AI doesn’t fix poor‑quality data, it amplifies it. The organizations seeing the greatest returns are those that invested in strong data foundations and automation before expanding AI across operations.”
What getting the sequencing right looks like
The organizations that avoid Zillow‑style failures follow a different sequence. They stabilize the foundations first, data, controls, workflows and only then introduce intelligence.
This is the shift described in the whitepaper: from deterministic automation to intelligent financial control.
Firms that modernize their data, automate core controls, and strengthen governance create an environment where AI can operate reliably. Those that skip these steps experience the same pattern Zillow Offers did, AI amplifying the weaknesses of the system beneath it.
Preparation comes before adoption
The lesson from Zillow Offers is simple: build the right foundation before scaling AI.
That means:
- standardizing and normalizing data
- automating core controls
- ensuring AI decisions are transparent and explainable
- increasing autonomy gradually
- letting trust build alongside capability
AI is an accelerator. It amplifies whatever operating model already exists.
For firms with fragmented data, inconsistent controls, and manual processes, that means scaling inefficiency faster. For firms with clean data, automated controls, and well‑governed workflows, it means unlocking the productivity gains AI promises.
Zillow Offers is a real‑estate example, but the lesson carries across financial services: When people stop trusting AI tools, they stop using them. The way to avoid that is to get the foundations right first, then scale.
To explore how intelligent financial control strengthens operational foundations, download our whitepaper When the System Outgrows the System: The New Era of Financial Control.