Automating Calls and Email: What It Looks Like in an Irish Business

Ask most people what AI automation means and two examples come up before anything else: the AI that takes notes on a call, and the AI that sorts an inbox. They are the examples people reach for because every business has both problems. Calls that end with nobody quite sure who is doing what. An inbox that eats the first hours of the day.

This piece shows what both look like when they are built properly. Not a demo and not a feature list, but two systems already running in production: one in an Irish solicitor practice, one in a founder's inbox.

Calls: from transcript to action list

An Irish solicitor practice was losing fee-earner time to the work that happens after a client call. Extracting the action items. Logging next steps. Assigning follow-ups. Going back through call notes to check what was actually agreed. All of it is necessary, and every minute of it comes out of a fee-earner's day.

Every client call is now captured, transcribed and run through an action-extraction model. The output is a structured action list: what needs to happen, who should do it, and by when. The fee-earner reviews the suggested actions, edits anything that needs adjusting, and assigns with one click. By the time they walk out of the call, the list is already waiting.

The detail that matters most is the audit trail. Every action carries a transcript reference. If a client asks "did we agree X on the call," the answer is a click away, not a re-listen of the whole recording.

The full build is documented in the Legal Call-Actions case study.

Email: from 200 messages to the ten that matter

In a 30-person business, the founder's inbox is a part-time job. Two hundred messages a day. Most of them are routine: suppliers, scheduling, simple internal requests. A handful genuinely need the founder, and missing one of those is expensive.

The system reads incoming mail and does three things with it.

• Classifies by priority Every message is sorted by importance and urgency, so the founder sees the ten things that matter first. Newsletters, automated notifications and internal CCs go into a separate stream, which keeps the inbox a working tool rather than a reading list.
• Drafts replies to routine items Scheduling requests, FYIs and simple asks get a drafted response ready for one-click send, in the founder's own voice and style rather than a generic template.
• Escalates only what needs direct attention Specific senders, keywords and customers are flagged. The system knows which clients are mid-deal and which suppliers are critical, so big customer escalations, board matters and deal flow reach the founder directly.

The founder now works from a triaged inbox, not a 200-message one. Eight hours back every week, without losing visibility on anything that matters. The full build is in the Email Triage case study.

Why these two are usually the right place to start

Most automation is sector-specific. A construction business needs its project pipeline handled. A waste operator needs collections and routes coordinated. Those systems are valuable, but they only make sense inside that kind of business.

Calls and email are different. Whatever a business sells, it has meetings and it has an inbox. That makes them the easiest entry point for a business that is unsure where to start with automation, for three practical reasons.

• The cost is already visible Everyone knows how long the post-call write-up takes and how much of the morning goes on email. You do not need a process-mapping exercise to find the time being lost.
• The work happens every day A saving that applies to every call and every morning's inbox adds up quickly, in a way that a once-a-quarter task never will.
• A person stays in control In both builds the system proposes and a person decides. The fee-earner approves the action list. The founder sends the draft or edits it. That makes it a sensible first step for a team that has never worked alongside AI before.

If you are still working out whether your business is ready for any of this, Is Your Business Ready for AI? walks through the questions worth answering first.

What this is not

It is worth being clear about how this differs from the tools most people already know.

Workflow tools like Zapier and Make connect apps together, so that something happening in one system triggers an action in another. Note-taking tools like Read AI and Otter transcribe meetings and summarise them. Both are general-purpose products, designed to work the same way for every business that uses them. They capture and summarise.

The distinction matters because a summary is not the job. The job is the follow-up after the call, and the decision about which ten emails deserve the founder's morning. What Keystone builds is specific to the business's own process: structured action lists routed to the right person, an inbox triaged in the founder's own voice, working alongside the systems the team already uses. The output is something to act on, not something else to read.

The bottom line

Automating calls and email is not about recording more or summarising more. It is about removing the admin that sits between a conversation and the work that follows it. Twenty-five minutes into three after a client call. Eight hours back every week from one inbox.

If you want to know what that would look like for your calls or your inbox, that is what the Discovery Call covers. It is free and takes twenty minutes. You can also see the wider range of what we automate on the AI Automation page.