Building AI alongside the people who sell

Alma Intel is the project that has shaped me most. AlmaHub was the product we built inside it. The company began with an observation, then changed through many conversations with the people we wanted to help.

While working in financial markets, I kept noticing inefficiencies in how people used technology. There were salespeople in my family, so I also had a view into another world: successful people selling valuable products with tools that seemed far behind the work they were doing. I thought there might be something I could fix.

Start with a conversation

I began arranging meetings with sales representatives, managers, and directors. I wanted to understand how technology had changed their working lives over the previous five years.

Many described difficult experiences with established tools, including Salesforce and HubSpot. Some had left large providers for smaller businesses that could adapt more closely to their operations. These were the experiences of the teams I spoke to. What interested me was the gap between buying software and getting useful work out of it.

After understanding a team’s needs, I would propose a second meeting. I came back with a business plan: where AI could help, what we could build first, and how we could work together. I was already presenting myself as someone working on sales technology, even before the company formally existed.

Some opportunities were immediate, such as filling a CRM from conversations. Others were more ambitious: applying econometrics to commercial data and using the context extracted by AI to identify skill gaps and support personal training. With frequent turnover in sales teams, helping new representatives learn faster was a problem people recognized.

Help the person doing the work

The idea that connected most strongly was simple: I wanted to make the sales team more capable.

Managers needed a clearer view of what was happening. Representatives needed an easier working day. Both had to benefit. A system that only collected more information for management would not earn the same trust as one that also helped a salesperson decide what to do next.

We went into each company and worked beside its teams. We learned which tools they already used, where information accumulated, and which tasks took effort without helping a sale move forward. That work revealed opportunities I had not seen in the first meeting.

The most rewarding part was discovering how much we had not known before we started.

From conversations to useful next steps

Two features made the idea concrete.

Smart summaries turned long conversation histories into useful context. The information was already there, but reading it all could consume a large part of the day. A representative needed to recover the relationship and understand the next step without starting from the beginning every time.

Smart follow-up helped order active conversations by which ones deserved attention that day. This was particularly useful for junior representatives facing a large number of leads. A manageable starting point helped them act.

The working flow was:

1. Understand the team’s current operation and the tools it already uses.
2. Collect and structure the context in its sales conversations.
3. Turn that context into summaries, CRM information, and follow-up priorities.
4. Give representatives a practical next action and managers a clearer view of the operation.
5. Use what happens in the real workflow to improve the product and identify training opportunities.

The point of collecting data was to return something useful to the people producing it.

A service becomes a platform

After we renewed our first large contract for another twelve months, we began turning the highly tailored service into a platform.

That became AlmaHub. We focused new client onboarding on operations that sold high-value products through WhatsApp, from the first contact through to the sale. This gave us a clearer product boundary while keeping us close to the work that had made the service valuable.

During the company’s operation, we analyzed more than 100,000 conversations and helped clients drive more than US$80 million in revenue. That figure refers to client revenue, not Alma Intel’s own revenue.

The product logic behind AlmaHub

The central technical problem was turning unstructured conversations into information people could act on. A conversation contains context across many messages. A useful product has to preserve that context while making it easier to retrieve and use.

Layer	What it did
Conversation context	Gathered the information already present in the sales operation.
Structured information	Turned conversations into CRM data and useful descriptions of customer relationships.
Summaries	Reduced the time required to understand a long interaction history.
Follow-up priorities	Helped representatives decide which active relationships to revisit.
Management and learning	Made team activity clearer and informed opportunities for coaching.

The training and econometric ideas belonged to that same product direction: use the accumulated context to understand where a team needed help. The practical work began with the parts people could use immediately.

The most important feedback came from watching a representative use the output. A summary could look good in isolation and still miss the detail needed for the next conversation. Working inside the company made those gaps visible.

Taking the experience to Enter

Enter approached me about joining its AI Deployment team. I decided to bring what I had learned to that work. My partner and I then chose to close Alma Intel and helped our clients integrate our technology into their own workflows.

The lesson I carry with me is to stay willing to revise my ideas. A plan is useful, but the people and operations in front of you keep teaching you things the plan did not contain.

Hold your convictions lightly enough to keep learning from the company you are actually working with.

Working beside sales representatives made me a better builder. It gave me a much more concrete understanding of what it means for technology to help someone.

Alma Intel on LinkedIn