Head of Marketing Matt Fontein on how AI supports content, design, research and outreach across ClearPoint's marketing team, and what other marketing leaders might take from it.
ClearPoint built its name in AI-enabled software engineering, and has been extending AI into the non-technical parts of the business: finance, operations, talent, and marketing. The company's chief operating officer, chief financial officer and head of talent and culture have each written about how AI supports their function. This is the marketing team's turn, covering the current state and where to next.
Marketing teams generally have one thing in common. They want more budget and more people to achieve their objectives, whether that’s demand generation, brand awareness or something else. More money and people brings options and capability. In marketing, where pay-to-play is often the mantra, budget buys results. In the real world, budgets and people are in limited supply, and decisions need to be made around the opportunity cost of using budget and time on one thing instead of something else. AI can’t give us more budget, but it can give us the equivalent of more bandwidth.
Whilst it's still early days, the output increase is clear to see. In the past six months the team has produced 50% more content, ran four events vs one and tripled the number of LinkedIn-based campaigns, all without increasing team size.
Where it runs
ClearPoint began using Claude Cowork about five months ago, connecting it directly into the tools the team already works in: Slack, Gmail, CRM, and Google Drive, including Docs, Sheets and Slides. That means agents can draw on the business's actual context (such as briefs, threads, prior work) rather than whatever someone happens to paste in. It's also usable without any technical background, since the team sets up agents in plain English rather than needing engineering skill. An earlier attempt at building automations in N8N hadn't found the same traction, mostly because of the technical setup involved. Those two things are why AI now runs across content, design, research and outreach as a matter of course. Claude Cowork meets ClearPoint’s strict security requirements too.
Content
Case studies and blogs are the two formats ClearPoint relies on most, and both used to bottleneck in the same place. Case studies need technical detail out of colleagues who are busy and don't always find it easy to frame their work against a template. Blogs need an original idea and confident writing, and those rarely sit with the same person.
The way round it is to take the groundwork off the subject matter expert. Colleagues hand over the facts of what happened in whatever form suits them, technical or otherwise, written or spoken. A recording, a Slack thread and a set of half-finished notes become a structured first pass against ClearPoint's template, with the gaps and the questions still to be answered marked up. Marketing writes and edits from there, then checks the piece back with the original source before it goes out. The internal message is that nobody needs to be a strong writer. An original idea or a clear set of facts is enough.
What AI takes out is the slow part at the front, sitting with a transcript and working out what the piece is. The message, the argument, the edit and the final copy stay with the team, informed by ClearPoint's brand, tone and voice guidelines.
As a result the team has produced over 50% more content in the past 6 months compared with the previous 6 month period.
What limits the output is the quality of what goes in. AI is good at interpreting and expanding source material and poor at inventing it. Without someone who knows the work supplying the substance, the result is generic and reads that way.
Design
Often client-facing material at ClearPoint, such as RFPs, mid-project updates and candidate profiles are created directly by colleagues outside the marketing team, using existing templates and brand guidelines. Marketing has less control over how that turns out than people might assume: as soon as someone without a design background starts populating a template freehand, brand consistency is often the first thing that slips, and for a professional services firm, brand is one of the clearest signals of who a document actually came from.
ClearPoint's Claude-based design tool is built to close that gap. It ingested the company's brand guidelines and templates, takes in undesigned content or copy from anyone on the team, and produces on-brand slides or documents, fully editable in Google Slides rather than locked into some separate format. The tool is still new but the output already lands around 80–90% of the way there within one or two rounds of iteration, which lifts the quality of what goes out to clients and in some cases saves days of work
Research and analysis
Analysis and research are the parts of marketing work that a stretched team most often has to deprioritise. It's useful, but rarely urgent enough to get proper time. AI covers a good part of that gap at ClearPoint. The team runs daily competitive analysis that goes beyond surfacing what competitors are doing, into what it actually means for ClearPoint and what the team should do in response, alongside a daily scan of the macro trends shaping the industry.
A second daily view brings together sales activity, marketing engagement and anonymous website visits by organisation, tracked through reverse-IP identification. It gives a single picture of which target accounts are responding to outreach and which need more attention, in real time. A daily briefing from an AI agent acting as a senior marketing advisor adds go-to-market ideas and near and longer-term actions, and a separate creative director agent looks for ways to sharpen ClearPoint's brand and campaign thinking. Neither replaces judgement, but both add a useful second perspective to the day-to-day.
Outreach
Outreach at ClearPoint sits primarily with the sales team, most of whose time goes to servicing existing clients rather than prospecting new ones. Doing it properly means understanding an organisation and the right person inside it well enough to send something worth reading, rather than a blanket message. This all takes time.
ClearPoint built a Claude-based process that pulls an ideal customer profile at the organisation level from Apollo, in line with local privacy and data laws. It then researches publicly-available information on each contact individually: current and past roles, and anything public about who they are as a person, alongside company-specific announcements and the wider trends facing that business. It's close to the depth of research that would once have taken a dedicated analyst days, applied at a scale that would otherwise take considerably more time. ClearPoint is now moving this workflow from a Claude-built tool to HubSpot's own Prospecting Agent feature, for the guardrails it gives over how each sales rep uses it, its native integration with the rest of HubSpot, and the autonomy it offers with less manual input required. It's an example of buying an existing tool rather than building the equivalent internally.
It's close to the shift ClearPoint's Chief Operating Officer, Paul Scott, has written about in his own piece on personal versus managed agents: agents that depend on one person prompting them, moving toward agents that run with less day-to-day input from any individual. Marketing is part of that same shift.
Mentorship
One of the more unusual applications is a personal one: a daily career coach agent that reviews Slack messages, meeting transcripts, email, calendar and Google Drive activity, and reports back each day on what went well and what could improve. One recent recommendation flagged that approvals were sitting longer than they should, and suggested setting aside 30 minutes each morning specifically to clear them. It's a small example, but a useful one: the same tool that helps the team can just as easily hold its own manager to account.
What this looks like in the numbers
Output is hard to measure precisely, when other factors mean correlating AI and any single metric requires careful criteria and analysis. What it can point to: brand awareness, engagement and lead volume have all held up well over a period with no material change in team size or budget, and against various macroeconomic events. In-person events remain one of ClearPoint's strongest demand drivers, and the team has run four of them in the past three months. Running an event well still relies on very little AI, but AI has freed up time elsewhere,, so more of it can go toward events and other work that doesn't lend itself to automation.
Looking ahead
ClearPoint's engineering business assesses AI maturity on an eight-stage model, running from manual work with the occasional chatbot question at one end, to building and customising an organisation's own orchestration tooling at the other. The team has adapted a marketing version of the same scale and places his team at stage two to three today, with parts of it reaching stage four. The next six months are about moving further along it: using agents more autonomously, and shifting from agents that rely on one person prompting them toward managed agents that take input from several people and run more independently in the background.
Alongside that, ClearPoint intends to build a marketing equivalent of the F25 framework it already uses to assess engineering capability, plus a formal assessment of the team's own AI capability. The approach so far has largely followed the tools and use cases as they've come up; the next phase is to put a more consistent foundation under that, with the understanding that there's no fixed end point and some of what reshapes the function in six or twelve months hasn't been built yet.
What other marketing leaders should take from this
- Make it a habit, not a pilot. A one-off proof of concept proves little. AI needs to become one of the first things opened each day, the same way a CRM or an inbox is, before it changes how a team actually works.
- Start from your real pain points. The tools that stick are the ones built around a genuine constraint, like limited budget or headcount. The ones built to be technically interesting rarely earn their keep.
- The output is only as good as the source. AI is very good at interpreting and expanding on original material, but it still needs that material to come from somewhere. Without strong source thinking from the people who actually know the business, the result is generic.
- Keep a human checking the work. AI can be extremely convincing while being wrong. Treat that as a given, not an edge case, and check the data and the interpretation before it goes anywhere near a client.
- Watch for the muscle that stops getting used. Writing, analysis and critical thinking are skills, and handing them to an agent means they aren't being exercised as often. That isn't necessarily a problem, but it is worth going into with eyes open.
The bigger point underneath all five is that this is a capability question, not a headcount one. What a small marketing function can deliver has more to do with how systematically it puts these tools to work than with how many people are on the team, and that isn't unique to marketing, or to ClearPoint.
