In partnership with

People are building AI copilots for everything these days, and sales is a compelling example… if you can get the software and the user experience right.

The Frontier AI team think they’re on the right path, and we meet them today.

But first:

• Liverpool-based Baltic Ventures has announced its latest accelerator cohort. Check them out here. PreSeed Now has previously featured Future Greens from the list.

– Martin

Training cutting edge AI? Unlock the data advantage today.

If you’re building or fine-tuning generative AI models, this guide is your shortcut to smarter AI model training. Learn how Shutterstock’s multimodal datasets—grounded in measurable user behavior—can help you reduce legal risk, boost creative diversity, and improve model reliability.

Inside, you’ll uncover why scraped data and aesthetic proxies often fall short—and how to use clustering methods and semantic evaluation to refine your dataset and your outputs. Designed for AI leaders, product teams, and ML engineers, this guide walks through how to identify refinement-worthy data, align with generative preferences, and validate progress with confidence.

Whether you're optimizing alignment, output quality, or time-to-value, this playbook gives you a data advantage. Download the guide and train your models with data built for performance.

Download the guide

AI for sales: The final Frontier?

AI co-pilots are everywhere these days. They’re in your office suite, they might be in your browser (I currently use newly Atlassian-owned Dia), and if you’re a developer they might be helping you code.

A general-purpose co-pilot probably isn’t much use if you’re in a high-pressure job like sales. That’s exactly where Frontier AI wants to step in and make a real difference, helping sales teams hit their targets, in a world where the majority don’t.

A commonly bandied-around statistic is that 60% of reps don’t hit their sales targets, leading to a high turnover of staff.

“We're creating a world where your sales team hit target every time,” pitches CEO Howard Kingston.

The idea for Frontier AI stems from Kingston’s own experience, running companies with large sales teams. Training them up could take months, and even then, when a lead came to nothing it was too late to do anything about it.

Analysing a call recording to see the mistakes a sales rep made was fine, but could only happen after the fact when the lead was a lost cause.

Hence Frontier AI has been built to integrate with video call software and listen into conversations, a bit like a meeting assistant AI does. As the call progresses, it draws on relevant, contextual information to help make sure the sales rep has the most important information on hand at the right time.

“We're building an AI knowledge graph that pulls in all of your company's sales material, like sales playbooks, scripts, and product information, and combines it with all the information that is contextually relevant about the customer, such as past conversations. It combines this and feeds the sales rep in real time with perfect thing to say when the prospect asks a question,” Kingston explains.

“So instead of a sales person crossing their fingers and the sales manager just telling their sales reps to ‘always be closing’, Frontier AI gives support to reps when they need it most, live on the call.”

When a new customer signs up with Frontier AI, the software will ingest things sales teams have, such as sales scripts, product information, and guides to handling objections that prospects might have. This is used to train a dedicated instance of an LLM for that customer.

The customer then installs Frontier AI software on their team’s computers, which can be brought into action when a call begins.

“The technical secret sauce that we have is very low latency. If a customer asks a question, we need to get an answer to a sales rep in half a second. If it takes 20 seconds, it's not useful,” Kingston says.

After the call, Frontier AI generates a report, scoring the rep on their performance. Did they stick to the script? Did they answer all the questions effectively? What was the sentiment of the call? Did they move to the next stage in the sales pipeline?

The software also offers tips for doing better next time, where needed. Usually, a sales manager will give feedback to reps from just a tiny sample of recordings, whereas Frontier AI can do that for every call.

The sales manager, meanwhile, can use Frontier AI to get a dashboard of their team members’ performance.

“This is a way for managers to see where they should be spending their time. Currently, sales managers really only have lag metrics, such as revenue, and sometimes that can be two or three months after calls. We’re giving them guidance on which sales reps are performing well on calls in real time,” Kingston says.

The story so far

Kingston’s experience in the startup world stretches back to one of the UK’s earliest accelerator programmes, Ignite 100, later known as Ignite. There he built Future Ad Labs, which then rebranded as Adludio.

Jaded by adtech, he eventually left the business at Series A. He then set up an edtech company called Velocity Academy, inspired by teaching he had done for General Assembly. It was through training sales teams that he spotted the potential for Frontier AI, which he is now focused on full-time.

Kingston is building Frontier AI with CTO Stef Lewandowski, who he met years ago at White Bear Yard in London, where they were working on separate projects.

The next steps for the startup are building out and refining the AI knowledge graph that underpins the product.

“The US and UK will be our main markets. We're really focused on going to the US quite quickly, because that's where the big go-to market teams are. We already have two paying customers in the US,” Kingston says.

Subscribe to Premium to read the rest.