部落格
The Usability Layer Your Business Systems Are Missing
Why enterprise software answers questions so poorly, and what to do about it
The problem is not the data
Most organisations have more business data than they know what to do with. ERPs track every transaction. CRMs log every customer interaction. HRIS platforms hold every employee record, policy document, and org chart revision. Internal wikis accumulate years of institutional knowledge. The information is there. The trouble is that getting to it requires knowing where it lives, how the system organises it, and which sequence of clicks, filters, or query syntax will surface the right record.
This is a usability problem, not a data problem. The systems themselves work as designed. They were built for administrators and power users: people who enter data, configure workflows, and run reports on a regular basis. But the people who most often need information from these systems are not the ones who built them or maintain them. They are managers preparing for a quarterly review, sales reps looking up a customer's contract history before a call, new hires trying to understand how expenses get approved. These users encounter the system's interface as a barrier rather than a tool.
The result is a quiet productivity tax. People spend time searching, asking colleagues, or simply going without the information they need. A recent post on this blog explored how ERPs, in particular, become black boxes for anyone outside the finance or operations team. The same dynamic plays out across every major business system.
What a usability layer looks like
A usability layer is an interface that sits between people and their existing systems. It does not replace those systems. It does not require migrating data or changing workflows. It provides a way to ask questions in ordinary language and receive answers drawn from whatever systems hold the relevant information.
This is different from a dashboard or a search bar bolted onto an existing tool. Dashboards are preconfigured views of data. They answer the questions someone anticipated in advance. A search bar within a single tool can only search that tool's content, and it typically returns a list of matching records rather than an answer. A usability layer works across systems and understands intent. When someone asks "What's the renewal date for Acme Corp's contract?", it should be able to find that answer whether it lives in the CRM, a shared drive, or an email thread.
Fabric provides this through its connections to dozens of tools, its AI assistant, and its semantic search. The approach is to connect your existing systems, index their content, and let people query across all of it using natural language. The person asking the question does not need to know which system holds the answer.
Why search alone is not enough
Search is a necessary component, but it is not sufficient on its own. Traditional keyword search returns documents that contain the words you typed. Semantic search is better: it understands meaning, so a search for "parental leave policy" can return a document titled "Family and Medical Leave Guidelines" even if the exact phrase does not appear. Fabric's search works this way, using the principles behind retrieval-augmented generation to match queries with relevant content across all connected sources.
But even good search still returns results. It gives you a list of documents or records and leaves you to read through them. A usability layer goes further. It reads the relevant content, synthesises an answer, and cites the sources so you can verify. This is what Fabric's AI assistant does: it queries across connected sources, composes a response, and links back to the original documents or records.
The distinction matters because it changes who can use the system effectively. Search requires literacy in the domain. You need to know roughly what you are looking for and be able to evaluate which result is most relevant. An AI assistant that synthesises answers lowers that bar considerably. A new hire can ask "How do we handle returns for international orders?" and get a clear answer without knowing whether that information lives in the help centre, the operations wiki, or a Slack thread from six months ago.
The cost of not having one
The cost of poor usability in business systems is difficult to measure precisely, which is one reason it persists. It shows up as time spent searching, as decisions made with incomplete information, as questions asked repeatedly because the answer is hard to find, and as institutional knowledge that exists only in the heads of long-tenured employees.
Consider a few specific scenarios. A sales rep preparing for a renewal call needs to review the customer's history: past issues, feature requests, contract terms, and recent support tickets. That information might span the CRM, the support ticketing system, email, and a shared drive. Without a usability layer, the rep either spends thirty minutes pulling it together or goes into the call underprepared. With one, they ask a question and get a briefing in seconds. Fabric's sales knowledge solution is designed for exactly this scenario.
Or consider onboarding. A new employee's first weeks involve hundreds of small questions: where to find templates, how to request access to a tool, what the approval process is for a purchase order. Each question is simple to answer if you know where to look. But a new hire does not know where to look, and asking a colleague every time is inefficient for both parties. A shared knowledge base helps, but only if it is comprehensive and current. A usability layer that can pull answers from across all systems fills the gaps that static documentation inevitably leaves.
Customer service teams face a similar challenge. Agents need fast access to product information, policy details, and account history. The faster they can find the right answer, the better the experience for the customer and the more efficient the operation.
Legacy systems benefit most
The usability gap is widest with older systems. Modern SaaS tools tend to have better search, cleaner interfaces, and API access that makes integration easier. Legacy systems, including on-premise ERPs, older CRMs, and homegrown databases, are often the most hostile to casual users and the most valuable to query.
Fabric's approach to legacy system integration treats these systems as data sources rather than interfaces. By connecting to them and indexing their content, Fabric makes the information they hold accessible through the same natural language interface used for everything else. The legacy system continues to serve its purpose as a system of record. The usability layer makes its contents available to everyone who needs them.
This is also where Fabric's ability to build connectors to new data sources becomes relevant. Most integration platforms are limited to their catalogue of pre-built connectors. If your system is not on the list, you are on your own. Fabric takes a different approach: through the MCP protocol, it can connect to any tool with an API, including niche, proprietary, or internally built systems. These connections sync continuously, so the usability layer always reflects the current state of the source system rather than a snapshot from the last manual export. That continuous sync is what makes the difference between a useful knowledge layer and an outdated one.
Documentation that stays current
One of the quieter benefits of a usability layer is what it does for documentation. Most organisations recognise that their internal documentation is incomplete, outdated, or both. The effort required to keep it current is substantial, and it rarely gets prioritised.
When an AI assistant can answer questions by pulling from live systems, the pressure on static documentation eases. The documentation does not need to cover every edge case if people can ask the AI and get an answer drawn from the source system. Fabric takes this a step further with self-writing documentation: the system can generate and update documentation based on what it finds across connected sources.
This does not eliminate the need for well-written guides and policies. It does mean that the gap between what is documented and what people need to know becomes less costly.
Governance and trust
A reasonable concern with any AI layer is accuracy and governance. If people are getting answers from an AI assistant rather than looking at source systems directly, how do you ensure the answers are correct? And how do you control who has access to what?
The answer is that the usability layer should respect the same access controls as the underlying systems. Fabric's approach to AI governance ensures that users can only query content they already have permission to access. The AI does not bypass access controls; it operates within them. Answers are cited with sources, so users can verify claims against the original record. And because the system connects to authoritative sources rather than maintaining its own copy of the truth, the information stays current.
For teams that need an auditable trail of decisions and the reasoning behind them, Fabric also supports a decision log that captures the context and evidence behind key choices.
Who benefits
The short answer is anyone who needs information from business systems but is not a daily power user of those systems. Product managers pulling together context from customer feedback, engineering tickets, and market research. Sales reps preparing for calls. Executives reviewing cross-functional metrics. New hires trying to learn the ropes. Support agents looking up policy details mid-conversation.
The usability layer does not make the underlying systems less important. It makes them more useful to a wider set of people. And for AI agents that automate multi-step workflows, the same connected infrastructure means the agent can pull data from one system, reason about it, and take action in another, all without manual data gathering.
Organisations that have invested heavily in their business systems deserve to get full value from them. A usability layer is how you close the gap between what your systems know and what your people can access.
Frequently asked questions
What is a usability layer for business systems?
A usability layer is an intelligent interface that sits on top of your existing business tools. It connects to systems like ERPs, CRMs, HRIS platforms, and wikis, then lets people ask questions in natural language and receive synthesised answers rather than raw search results or database views.
Does a usability layer replace our existing systems?
No. It works alongside them. Your existing systems remain the source of record. The usability layer connects to them, indexes their content, and provides a more accessible way to retrieve information. No data migration or workflow changes are required.
How does Fabric connect to our existing tools?
Fabric offers prebuilt connections to dozens of common business tools, including CRMs, project management platforms, cloud storage, and communication tools. For systems without a prebuilt connector, Fabric can build new connectors through the MCP protocol, connecting any tool with an API. These connections sync continuously, so the usability layer always reflects live data.
How is this different from the search bar in our existing tools?
A search bar within a single tool can only search that tool's content and typically returns a list of matching records. Fabric's semantic search works across all connected systems and understands the meaning of your query, not just the keywords. The AI assistant goes further by synthesising answers from multiple sources rather than returning a list of results.
Can new hires use Fabric to get up to speed faster?
Yes. New employees typically have hundreds of small questions during their first weeks, and they rarely know which system holds the answer. Fabric lets them ask questions in plain language and receive answers drawn from across all connected systems, reducing their reliance on colleagues and making onboarding more efficient.
How does Fabric handle access controls and data governance?
Fabric respects the access controls of the underlying systems. Users can only query content they already have permission to access. Answers include citations to source documents, so users can verify information. The system does not bypass or override existing security policies.
What types of questions can the AI assistant answer?
The AI assistant can answer questions that draw on information stored across your connected systems. This includes factual lookups (contract dates, policy details, product specifications), contextual summaries (customer history, project status), and procedural questions (how to submit an expense report, what the approval process is for a purchase order).
Does Fabric work with legacy or on-premise systems?
Yes. Fabric can connect to legacy and on-premise systems through its connector framework and the MCP protocol. This makes the information in older systems accessible through the same natural language interface used for modern cloud tools, without requiring changes to the legacy system itself.
How does Fabric keep documentation up to date?
Fabric can generate and update documentation based on what it finds across connected sources. Because the AI assistant pulls answers from live systems, the gap between what is documented and what people need to know becomes less of an issue, even when static documentation falls behind.
Can Fabric automate tasks across multiple systems?
Yes. Fabric's AI agents can perform multi-step tasks that span multiple connected systems. For example, an agent could gather data from a CRM, cross-reference it with information in a project management tool, and compile a summary, all without manual data gathering.