All Posts

Where Your Data Lives Is Only Part of the Sovereignty Question

The easiest time to discuss leaving a technology platform is before the company depends on it. Once prompts, workflows, integrations, and operating knowledge are embedded in a service, the practical cost of change can exceed the price of the subscription. Identify what the business would need to take with it. That may include source information, conversation records, configuration, evaluation examples, workflow definitions, and evidence of approvals. An export feature is usef

On Tuesday, November 3, 360Views will be closed for Election Day. Our employees and contractors will have a paid day off to exercise their right to vote. I would like to see more businesses doing this. Our work helps businesses move forward. Our responsibility also extends to the communities where we live and work. Giving our team time to participate in our democracy is one way we put that responsibility into practice. We encourage other employers to make room for their teams

An AI initiative may begin as a small subscription purchase and grow into charges spread across departments and services. Licenses, consumption, integration, storage, evaluation, and employee review effort can all contribute to the real cost of delivering the work. Leadership needs to connect those costs to a business purpose. Identify the owner of each significant use, the budget supporting it, and the outcome expected in return. A central technology bill can hide which proc

An AI policy can be sensible on the day it is approved and gradually lose contact with the systems employees actually use. New features arrive, information sources change, agents gain actions, and a pilot becomes part of a critical process. Governance needs a way to follow those changes. Start with an inventory that connects each important use to a business owner, a technical owner, the information involved, and the decisions or actions the system supports. Define what trigge

A marketing image, proposal paragraph, or software component may pass through several people and tools before publication. By the time someone asks where it came from, the original inputs and decisions can be difficult to reconstruct. That history matters when the organization needs to answer questions about accuracy, confidentiality, licensing, or permitted use. An employee remembering that “AI helped with it” is not much of a record. For important content, keep proportionat

A customer does not necessarily distinguish between a statement from an employee and an answer delivered through the company’s AI assistant. If the assistant describes a delivery date, a refund, or a product capability, the customer may treat it as a commitment. Define the boundaries before launch. Which facts can the assistant provide? Which offers can it make? When should it collect information and hand the conversation to a person? Those decisions need input from the teams

Imagine a customer asking an agent to compare suppliers, check availability, and prepare a purchase. The customer still has a business need, but the first interaction with your company may come through software acting on their behalf. Preparation starts with dependable information. Product descriptions, service scope, pricing conditions, availability, and ordering rules need to be consistent across the systems that publish them. Conflicting information creates problems whethe

A prospective customer may ask an AI assistant to compare suppliers before visiting any of their websites. The answer can shape the shortlist, including how the customer understands your services, qualifications, and fit for the work. Start by testing ordinary questions a buyer would ask. What does the company do? Which problems does it solve? Where does it operate? What evidence supports its claims? Review the answers across representative services without assuming that one

An AI pilot can make one step in a process much faster. That is useful evidence, but it is not enough to determine how the entire department should be staffed. Consider a team that reviews complex submissions. AI may extract information and prepare a summary in minutes. Employees still resolve missing information, interpret exceptions, coordinate with other departments, and take responsibility for the final decision. Speeding up extraction can simply move more work into the r

In many organizations, the same tool can produce very different results depending on who is using it. One employee knows how to frame the task, provide useful context, and check the output. Another gets a weak answer and concludes that AI has little value. That difference becomes a business issue when useful techniques remain with a few individuals. Their work improves, while the rest of the organization continues with older practices or uses the tools without an effective re

If an applicant asks why they were screened out, the employer needs an answer more useful than “the system ranked you lower.” Introducing automation into recruiting changes the process, but it does not remove responsibility for the decision. Begin by identifying where the technology influences the outcome. It may organize applications, summarize experience, recommend a shortlist, or remove candidates before a person sees them. Each use requires a clear understanding of the in

A meeting assistant can produce a useful summary before the participants have returned to their desks. It can also create a durable record of a conversation that nobody expected to become searchable or shareable. Think about a discussion involving a potential acquisition, an employee matter, a customer negotiation, or an unfinished strategy. The recording, transcript, and summary may have different audiences and retention needs. The convenience of an automated recap does not

A folder can remain broadly accessible for years without attracting much attention. Employees may not know it exists, and finding the useful material inside it may take considerable effort. Introducing AI-assisted discovery can make those old access decisions much more consequential. The concern is not that every AI tool ignores permissions. It is that information already available to a user may become easier to find, combine, and interpret. Access that once seemed harmless c

The moment an employee submits a document to an AI service, a business information decision has been made. Understanding that decision requires more than knowing the name of the application on the screen. Where is the information processed? How long are prompts, attachments, and outputs retained? Who can access them? Which providers help deliver the service? What settings and contractual commitments apply to the account being used? These questions may have different answers a

A confident answer can move through a company faster than the evidence behind it. Once an AI-generated statement appears in a presentation, proposal, or management report, the next reader may assume someone has already checked it. The review process should reflect the consequence of being wrong. Drafting an internal announcement requires a different level of verification from interpreting a contract obligation, recommending a payment, or advising a customer about a technical

An assistant that drafts a recommendation and an agent that changes business records carry different responsibilities. Before connecting an agent to operational systems, leadership needs to define the job it is being trusted to perform. Consider an agent supporting supplier onboarding. It might collect documents, identify missing fields, and prepare a record for review. The authority to change bank details or approve a supplier introduces a different level of consequence. Tho

When employees use an unapproved AI tool, the first question is often who broke the rule. A more useful investigation also asks what work the employee was trying to complete and why the approved approach did not meet that need. Someone may paste a customer document into a personal account because summarizing it manually takes too long. A department may buy an AI service on a corporate card because the enterprise evaluation has stalled. These choices create information flows t

A company can have impressive AI usage figures and still struggle to explain what improved. Employees are opening the tools, generating content, and attending training. The business needs to know whether that activity produces better work. Start with a specific process. If AI helps prepare a customer proposal, measure the time from request to approved proposal, the effort required to review it, and the corrections needed before it can be sent. A faster first draft may be usef

An employee spends less time at the keyboard because she is resolving a difficult customer problem on the phone. Another produces a constant stream of activity but leaves colleagues correcting the work. A monitoring dashboard may tell a different story from the people who understand the results. That discrepancy becomes consequential when activity measures influence performance ratings, schedules, compensation, or promotion. Leadership needs to understand what the system meas