Almost every conversation I have with a distributor these days turns to AI. Most already have a pretty good sense of where AI should be making a difference in their business. Increasingly, I'm seeing staff and execs take matters into their own hands. People who've never written a line of code are vibe coding their own apps to automate this work. I love this trend so much. When it works, it’s magical. 

But the painful reality is that most distributors still aren’t seeing the impact to their business. Building a useful AI app is one thing. Relying on it to run critical parts of your business every day is another thing entirely.

What it takes for agents to perform critical daily operations

It turns out that if you’re going to ask AI to perform critical daily operations, you need more than just a capable AI model.

1. A connected workspace

The foundation is a workspace that’s connected to your data and the tools you use to do the job. Agents need access to your ERP and other systems, memory for the countless preferences and exceptions that make your business unique, and clear rules for what they're authorized to do. 

2. An agent builder

A workspace that restricts you to a few pre-configured tools doesn’t do much. You need to be able to create custom workflows that are tailored to the way your business actually runs, and then be able to actually train your agents on the nuances of your internal operations.

3. Traceability

When you have agents involved in critical operations, you need visibility into what they’re doing. Your workspace should let you audit the actions your agents have taken, including each task they completed and the impact from what they did. 

4. Hosted run-time environment

If you’ve ever vibe coded an app, then hit a wall once you tried to deploy it in a real-world workflow, you’ll have realized—even if you didn’t know the jargon—the importance of a hosted run-time environment. It’s what enables you to have agents that are always on and listening, as well as security infrastructure and action validation. You can’t graduate from a simple app that does a task when you hit a button to agents that autonomously complete important tasks in the background without it.

5. Deployment support

Finally, you need a team that’s going to partner with you as you deploy agents in the field. If you’ve been following the AI space closely, you’ve heard the term forward-deployed engineer or FDE. FDEs can help you leapfrog technical blockers and build reliable agents faster with expertise from standing up real-world deployments.

Agents doing work that matters

When we put all of the above together, we’ve seen amazing results. We’ve spent a lot of time with distributors in the field building agents for different use cases. Here's a sample of the deployments we’ve been working on:

• Order processing agents that turn voicemails, texts, and emails into draft orders in your ERP, freeing reps to focus on selling instead of order entry. We have over 200 customers using this agent, saving 20+ hours of manual order entry time per week. 
• Pricing agents that review cost changes and markups to calculate adjustments that reflect your pricing strategy. At one distributor, our pricing tools have generated over $200K in annualized incremental gross profit on items with optimized pricing, without losing sales volume.
• Billback agents that identify rebate opportunities that otherwise get missed. At one distributor, our billback agent recovered over $100K in annual billbacks on products they were already selling but not claiming through their GPO program.
• Purchasing agents that analyze inventory, sales data, and historical buying patterns, including large holiday buys, to help distributors avoid costly shortages, overbuying, and last-minute emergency purchases. After just two weeks working with our team, one distributor's owner described the tool as "more accurate and faster than a human" and expects it to eventually handle 95% of his purchasing.

But every new agent can’t mean another application for your team to learn and manage. As agents take on more work across sales, purchasing, and finance, you need one place to give instructions, review decisions, and keep track of what’s been done. More agents should mean less work for your team, not more software to manage.

We had to build the infrastructure ourselves to make these deployments work. Now we’re making it available to you.

A workspace to build and work with agents

Dispatch brings together your data, tools, and AI agents in one place. It's where your team can give agents instructions, review their work, and decide what they should handle independently.

Some agents work continuously in the background, while others get to work when you ask. They don't just analyze data and make recommendations. They can also take action in Pepper and other tools connected to your workspace.

Dispatch also makes it dramatically easier to build and deploy your own agents, tailored to your business. If there's a workflow we haven't tackled yet, our forward-deployed engineers can work directly with your team to build an agent and get it working in your business.

We've spent the last six years building technology to help independent distributors grow faster and run more profitably. AI is giving us an opportunity to take that mission much further, and we're already seeing the results. It's hard to think of a more exciting time to be building for our customers.

If there's a workflow in your business that you think AI should be doing, give us a call. We're working with distributors to build and deploy these solutions every day, and we'd love to build the next one with you.

You can sign up for early access to Dispatch here: usepepper.com/dispatch