The Phocas approach to AI is grounded in trusted data

Artificial Intelligence is moving faster than most businesses can keep up with. Our role as a technology partner to distributors is to stay on top of these developments, make the right decisions about what belongs in Phocas, and apply AI where it delivers real value. Generative AI, machine learning and agentic AI get talked about like they're one thing, when in practice each solves a different problem. 

Behind every successful AI use case sits a decision about what data it's built on, how much a business can trust the answer and whether it matches the way distributors work. Phocas is getting the data right first and then deciding what to build based on the needs of distribution businesses.

Phocas is setting out to create artificial intelligence that suits distributors by linking its use cases to sales efficiency, workforce planning, rebates, inventory management, demand forecasting and managing customers. Phocas co-founder and CEO Myles Glashier has watched plenty of technology trends promise the world and disappear. His view on AI is different, as it is a technology with significant investment and real teeth, but the potential only turns into something useful when it's connected to trusted data.

The Phocas approach to AI is built on a data foundation. For 25 years, Phocas has worked with distributors, pulling their ERP data into sales, financial and operational databases so every team in the company can work from the same source of truth for data analysis. This integrated data can then be drilled into by any variable, used to measure KPIs or to compare one month to the next and see a clearer picture of financial performance across the business. Every AI feature sits on top of that same data foundation, so an answer from Phocas AI traces back to that powerful data platform. Garbage in still means garbage out, no matter how advanced the model, and poor-quality data creates its own risk management problem long before AI enters the picture. Phocas puts as much engineering effort into cleaning, standardizing and keeping that data current as it does into the AI itself.

What happens behind the AI?

A lot of the AI use cases showing up in business software today rely on large language models or LLMs. LLMs are pattern-matching systems trained on enormous volumes of text, capable of understanding a plain-language question and generating a clear answer in return. Phocas uses this natural language processing (NLP) capability in Phocas AI, letting anyone ask a business question like "which customers grew fastest this month?" and get an answer instantly, with the reasoning behind it shown step by step. Unlike a general-purpose chatbot pulling from the open internet, Phocas AI answers come from a distributor's own connected data. We think this transparency matters and helps people learn, because it shows how the system got there and builds confidence in the tool which is what responsible AI should look like in practice.

This natural language chat has been live in the Phocas BI platform for more than two years, helping to make analysis available to more people. Security runs alongside all of it and transactional data stays inside the Phocas environment and isn't shared externally. Customer data processed through AI features is not used to train public or shared foundation models.

A different kind of AI in Sales Insights?

Where the data approach comes together most clearly is in Phocas Sales Insights, which is purpose built for sales teams in distribution. The inspiration was a clear observation that most reps are short on time to make sense of the data before the next call. Sales Insights takes the sales, product and customer data already sitting in Phocas and turns it into a clear picture of where a rep's effort should be directed. AI customer summaries update themselves as a distributor's business changes, based on the RFM criteria; recency of purchase, frequency of purchase and monetary value of purchase. Built on the same predictive analytics used elsewhere in Phocas, these meaningful plans help sales reps plan, understand and action the insights in their data quickly.

The efficiency gain is real, but the real value for distributors is a sales team that operates more strategically. Sales reps spend time on the accounts and conversations most likely to grow revenue, rather than spreading attention evenly across a territory regardless of where the opportunity or risk sits. That's a shift to a more considered approach where account priority is set by buying patterns, order trends and early signs of a customer pulling back, surfaced automatically much like a recommendation engine would in other industries.

Sales activity has traditionally been the hardest part of a distribution business to tie back to overall performance. Insights runs on the same connected data as Phocas Analytics and CRM, meaning a sales leader and a rep can work from the same big picture.

None of this is Phocas guessing at where AI might be useful, because 25 years of working inside distribution means the use cases are relatively straightforward to work out. As Phocas' own AI development continues, new capabilities are already underway, including a Phocas MCP Server to connect trusted data to other AI tools. The consistent approach at Phocas is to start with connected data, apply transparent context and use AI where it delivers real value for distributors.