Enterprise AI can be held back not only by what a model can do, but by whether it can reach and correctly interpret information inside a company’s existing systems. That was the central argument in a HumanX Amsterdam conversation between Omri Hurwitz and Idan Shuster, co-founder and chief product officer of Unfold, as reported by TechBullion.
Why access to company data is only part of the problem
Many organizations have years of valuable information in business software they already rely on. But those systems may be proprietary, old, difficult to integrate, or inconsistent in the ways they expose data. Shuster’s and Unfold’s view is that this can make information difficult for AI tools to use, even when the information already exists.
There is a second hurdle: records alone do not necessarily explain what they mean. An application’s screens, workflows, custom components, and underlying structure help people interpret information. As The San Francisco Tribune reported, Shuster cautioned that “If you let the agents interact directly with the data, sometimes it doesn’t make sense.” The point is not simply to connect an agent to a database; it is to preserve enough of the system’s context for the data to be useful.
What Unfold says it does
Unfold presents itself as an integration layer for enterprise systems, including ones without convenient APIs or export options. Its official product page describes a process in which the platform examines a system layer by layer, then delivers normalized, governed outputs into existing tools and AI workflows. The page names destinations and related technologies including Splunk, Cortex, Microsoft OneLake, Snowflake, Databricks, and AI agents.
Recommended Free Tools
#1 Best Overall
This is Unfold’s description of its product, not independent evidence that it works with every system or delivers a particular level of performance. The distinction matters: “AI-ready” output depends on the source system, how its context is mapped, the quality of the resulting data, and the controls applied before that data is used.
What “context” means in this conversation
Context is the information needed to interpret a record as it functions inside a real application. A field may have a particular meaning because of a workflow, a custom configuration, or its relationship to other parts of a system. An agent that can retrieve the field but misses those details may return an answer that is incomplete or misleading.
That is why Shuster’s argument goes beyond connectivity. Mapping how an application works and normalizing its data are intended to make information more consistent for downstream tools. The interview does not establish that this approach eliminates errors or replaces the need for human oversight.
Examples from the interview
The coverage uses two examples to show why companies might want to connect hard-to-access systems to AI workflows. They are examples Shuster gave in the interview, not independently audited customer outcomes.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match- Healthcare acquisitions: Shuster described an organization acquiring roughly 50 clinics a year, with integration of each clinic’s existing technology taking months. The example illustrates how varied systems can complicate combining operations after an acquisition.
- Retail fraud analysis: He also described a large retailer needing data held on a mainframe for fraud analysis. The example shows how important information can remain in a legacy system rather than a newer analytics environment.
The reporting does not name these organizations or quantify savings, fraud reductions, or other results. The examples therefore explain the problem Unfold is targeting; they do not prove a particular business outcome.
What the reported onboarding timeline does—and does not—mean
Unfold says initial onboarding for one system typically takes around seven days, and its website uses the phrase “Any system. Live in 7 days.” The timeline is a company-reported claim for initial work on a system, not an independently measured or guaranteed service level. The interview coverage also says onboarding involves human verification, a relevant qualification for a product that aims to interpret system structure rather than merely copy data.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Who may care about this problem
The interview coverage says Unfold initially focused on security and fraud-related data access, then broadened its focus after a healthcare CISO connected the team with a CIO seeking data from proprietary healthcare systems for AI workflows. It reports that Shuster worked in cybersecurity, including penetration testing and offensive security, before product management at Varonis, and that he served in Israel’s Unit 8200. These biographical and company-history details are reported by the interview coverage.
Unfold says it initially targets large enterprises. For those organizations, the question is not only whether a model can generate useful answers, but whether it can access the systems employees depend on and retain the context needed to interpret their information. Shuster’s broader point, as quoted by TechBullion, is that companies with a mix of modern and legacy software may eventually encounter this problem.
Best Value
What an enterprise should evaluate
The interview is not a product comparison or an independent technical assessment, so it cannot establish that Unfold is better than alternatives or appropriate for a particular organization. A company considering this kind of integration should investigate the practical details behind the promise:
- System coverage: Which specific applications, versions, and custom configurations can be connected, especially when APIs or exports are absent?
- Context mapping: How does the provider capture workflows and application behavior, and how can the organization check that the resulting interpretation is accurate?
- Governance and review: What access controls apply, how are outputs validated, and where is human verification required?
- Destinations: Can normalized data reach the organization’s actual analytics tools and AI workflows?
- Evidence of outcomes: Can the provider share customer references or independently verifiable results for comparable systems and use cases?
These questions follow from the challenge and product description in the interview; the coverage does not report answers to them or compare competing products.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
