The interface may be becoming optional. The value underneath it isn’t.
This week’s big idea: Your interface is a front door. Your knowledge is the asset.
BEHIND THE IDEA
Did they stop using your product, or did the interface simply stop being the product?
For the last twenty years, software companies have been trained to think that getting customers into the product is the goal. More logins, more dashboard usage, more time spent inside the application. AI is starting to challenge that assumption.
Imagine that one of your customers can ask for the answer they need from Slack, an AI assistant, or another tool they already have open. They never visit your dashboard or navigate the menus your team spent years improving. But they are still using your data, relying on your rules, triggering your workflows, and depending on everything your company knows about their business.
That distinction matters for established software companies. A mature application has accumulated something more valuable than its screens: years of customer history, workflow logic, permissions, integrations, domain rules, and thousands of decisions about how an industry actually works. AI creates a new way to expose those assets without throwing away the product that created them.
The better measure is the work your product gets done for customers, not the time they spend inside it.
So instead of asking “How do we add AI to our software?”, put a more useful question on the whiteboard: What valuable thing inside our software should customers be able to use without opening it?
IN THE NEWS
Salesforce just made its interface optional.
On September 15 at Dreamforce, Salesforce announced AIforce and made a striking case: people should not have to log into Salesforce to use it. The new layer brings Salesforce data, workflows, business logic, and permissions into the tools people already have open, like Slack and Claude.
The interesting part for an established software CEO is not that Salesforce launched another AI product. It is what the move says about where Salesforce believes the durable value of enterprise software sits.
For years, software companies competed to become the application customers opened every morning. If AI becomes another front door, the interface matters less. The assets that are harder to replace sit behind it: proprietary data, business rules, workflows, permissions, customer history, and deep knowledge of how a particular industry operates.
The risk is real: if the assistant becomes the front door, you need to own what it cannot copy.
What this means for your company: A company that spent two decades getting people to log in now says the login is optional. If you price per seat, or your renewal health score counts logins, you are measuring the front door. The companies that win will charge for the work done, not the seats filled. Which data, rules, or workflows does your product uniquely own?
THE FRAMEWORK
The Agentic Layer Pattern
The Agentic Layer Pattern puts an AI workflow layer on top of an existing product so one person can get more done, without rewriting the system underneath.
The starting point is not “Where can we put a chatbot?” If the interface is only the front door, the work is finding what sits behind yours. Start with one workflow customers repeat every week, and map it before you decide where AI belongs.
1. Information. What does the customer need to know?
2. Decision. What judgment do they make?
3. Action. What happens in the system next?
Example: a supervisor asks which work center lost the most time last week, decides where to send maintenance, and the work order gets created.
Once those steps are visible, ask which part your product could begin doing for them.
This moves the conversation away from adding AI because the market expects an AI feature. The team starts looking for valuable work the product already understands well enough to help perform.
Put This Into Practice: Pick one workflow. Map the information, decision, and action. Then check whether your existing product has enough trusted data, rules, and context to help perform it. If the answer is yes, that workflow is your first candidate. If it is no, the gap is your roadmap.
FROM THE FIELD
The old product wasn’t the problem.
At Halo, we worked with an established manufacturing software company that planned to rebuild its product screen by screen. More than 200 features. A year or more of work. A newer front door on the same house.
Then the team put a different idea in front of customers: ask the system a question in plain English. Which work center had the most downtime last week? Asked when they would want it, one customer answered with a single word: now.
The plan changed. Build on the data and the rules first, and let the screens follow. One question could replace four or five screens and a phone call.
An industry analyst on the company’s advisory board put it this way: “I would have been disappointed if you didn’t come to us with a concept like this.”
That is one of the advantages established software companies have over new AI startups. An older product may hold years of customer behavior, industry rules, integrations, and operational knowledge that would take a new entrant a long time to recreate. The opportunity is rarely to discard that history. It is to find a new way in that makes it more valuable.
It is early. The proof of concept starts with the data layer.
The old product may look old. The knowledge inside it isn’t.
ONE QUESTION TO CONSIDER
If customers could get the outcome they need without opening your software, what would they still need your company for?
Jason Weaver is the founder of Halo, where we help software companies find and build their next product. If you map a workflow and get stuck, email jason@halo.ceo and I will tell you where I would look first.
The Whiteboard arrives by email each week. Subscribe here.