Taking Our Own Medicine
We tell clients constantly: your CRM should fit your business, not the other way around. Most platforms are built for a generic company that doesn’t exist, and you end up paying for functionality you’ll never touch while the one thing you actually need is bolted on as an afterthought, if it’s there at all.
So when we found ourselves in exactly that position, the fix was obvious. Build the thing ourselves.
The problem
Acoustic is a small firm. Cold outreach has never been our strength, and we don’t want it to be. What we do have is a steady trickle of people arriving at our website, some ready to talk, most just looking. The gap wasn’t leads. It was what happened to them after they landed on the site.
Contact forms sat in an inbox. Newsletter signups went into a list and stayed there. Conversations through our own AI chat tool were interesting but disconnected from everything else. Three sources of genuine interest, none of them talking to each other, and no consistent way to follow up.
What we built
A CRM, but a small one, built specifically around how our own website generates interest.
It pulls data from three places: augmented conversations on the site, contact form submissions, and newsletter signups.
The chatbot piece deserves more detail, because it’s doing the most work of the three. It’s a RAG-based chat tool on the Acoustic website, but it’s not a generic assistant answering generic questions. When someone starts a conversation, it pulls in context about who they are and what they do, drawing on their own company website alongside whatever they’ve typed into the chat. It then matches that against everything we know: our services, relevant case studies, past client work, blog posts we’ve written on the exact problem they’re describing.
So if someone from a logistics company asks about automating customer support, the chatbot isn’t giving them a brochure answer. It’s pulling in the specific service that fits, pointing to a comparable project if one exists, and grounding the conversation in what we actually know about their industry and their situation.
That’s the difference between a chatbot that answers questions and one that’s actually useful for qualifying interest. By the time that conversation gets logged into the CRM, there’s already a rich picture of what the person cares about, not just a name and an email address.
Each entry, from any of the three sources, is then automatically enriched using information from the person’s own company website and from Apollo.ai, so by the time a name lands in front of us, there’s already useful context attached.
From there it suggests a path forward. Sometimes that’s an email, sometimes LinkedIn, always with something specific to open with. It doesn’t send anything automatically. Automated LinkedIn outreach is a fast way to burn a network, and we didn’t want any part of that. It suggests. We decide.
Everything then lives in a Kanban-style dashboard, so following up with each person is a matter of moving a card, not remembering who to chase and when.
Why it matters
None of this is groundbreaking technology. Enrichment tools, RAG systems, Kanban boards all exist elsewhere. What’s different is that everything here was built around one specific business, its specific channels, and its specific way of working, rather than adapted to fit a platform that assumed a different kind of company entirely.
That’s the argument we make to clients. This time we had to make it to ourselves.
Where this goes next
This isn’t a multi-tenant SaaS product, at least not yet. It’s a tool built for us, that happens to solve a problem a lot of small companies quietly have. If you’re looking at your own CRM and wondering why you’re paying for features you’ll never use while the integration you actually need doesn’t exist, get in touch. We’re happy to see whether something similar makes sense for you.