Slack AI reporting assistants
People used to wait for a Tableau export just to check their hours or leave. Now they ask in Slack and get the answer back in seconds. I built three of these assistants with Claude and Workato.
Most teams have plenty of data but not enough answers to what happened, why it happened and what to do next. I clean and connect their data into trusted gold layer models, then build dashboards that tell the story. AI assistants handle the quick questions in between. Leaders see the insight at first sight and spend their time on decisions, not reports.
Every organisation has a story in its data. The hard part is finding the right answer, the one leaders actually need to make a decision. It’s about asking the right questions, fixing what’s broken underneath, and knowing which step comes next. Here’s how I do it.
Exports from HR, finance and CRM systems, profiled for gaps, duplicates and broken formats.
SQL and Python scripts clean, test and model the data into one trusted source that refreshes itself.
The clean output becomes KPIs, trends and drill downs leaders can explore on their own.
The numbers become a clear narrative and a recommendation, so the meeting ends with a decision.
People used to wait for a Tableau export just to check their hours or leave. Now they ask in Slack and get the answer back in seconds. I built three of these assistants with Claude and Workato.
Senior leaders wanted answers about headcount without having to learn a dashboard. So I built an agent with Claude and Workato that lets them just ask.
Talent Acquisition was running on spreadsheets. I replaced them with Tableau dashboards for headcount, attrition and the whole hiring pipeline, and tested every number before anyone relied on it.
Finance was rebuilding the same Excel reports every month. I set up a pipeline into Tableau, so the P&L, balance sheet, budget vs actual, cashflow and ageing now refresh on their own.
SAP, BambooHR and Greenhouse never quite agreed with each other. I brought them together in one tested, documented dataset in AWS RDS, which saved over 100 hours of manual work a year.
Workforce planning only ever looked backwards. I built models that use past trends to show where headcount is likely to go, so resourcing calls can be made earlier.
Some reports don’t need a person to run them at all. I build workflows in n8n and Workato, including AI agents in Slack and Teams, that deliver them where people already are.
At the agency, every client wanted to know where their budget was actually working. I built dashboards in Data Studio and Zoho Analytics that showed them, channel by channel.
The visuals here are mockups with made up numbers. The real dashboards hold confidential employee and client data, so they stay private.
These are live Tableau Public dashboards built on public or made up data. Use them the way a manager would: filter, hover and drill in.
The tools change from job to job. These habits don’t.
Agree what each number means before building anything. If two teams define turnover differently, the dashboard is already wrong.
Build tested SQL and dbt pipelines, so every source lands in one dataset people can rely on.
Lead with the answer. The chart is there to back it up, not to make people go looking.
Put AI assistants in Slack or Teams, so a quick question gets a quick answer.
Control who sees what, stay GDPR compliant and write it all down. That’s how people come to trust the numbers.
Computer Science gave me the engineering side and an MBA gave me the business side. Working across finance, marketing and people data taught me how to join the two.
Each role added something new. Finance taught me to get the numbers right, the agency taught me to explain them, and FIRST is where I took on the engineering and AI.
I lead People Analytics, working with HR, Payroll, Finance and senior leadership. I own the data end to end, from the source systems through to the AI assistants people use every day.
Agency life meant lots of clients, and every one of them asking where their budget should go next.
Where it all started. A finance team, month end deadlines, and numbers that had to tie out.
Each job taught me something I still use every day.
My first analyst job was in finance at Hi Inc., reconciling bank statements, VAT, AP and AR at month end. It taught me that a report is no use if the numbers don’t tie out. It’s also where I built my first automated Tableau dashboards on NetSuite, and finally stopped rebuilding revenue figures by hand.
What stuck: get it right firstAt WHM Global I worked with lots of clients, and they all wanted to know where to spend next. I dug into their channel and campaign data, built dashboards in Data Studio and Zoho Analytics, and learned to explain the results to people who don’t live in spreadsheets.
What stuck: keep it simpleWhen I joined FIRST, people data was spread across SAP, BambooHR, Greenhouse and a lot of spreadsheets, and no two reports agreed. I sat down with HR leaders to agree what each metric meant, built tested dbt pipelines into AWS RDS, and swapped the manual reports for Tableau dashboards. That saved over 100 hours a year.
What stuck: fix the foundationsOnce the data was solid, I started wondering why anyone should have to open a dashboard at all. So I built Slack assistants and an Analytics Agent with Claude and Workato. Now a manager can ask a question in plain English and get a secure answer in about 2.4 seconds.
What stuck: answer the questionI’m still finding new ways to bring finance, people data and AI closer together. The aim hasn’t changed since that first finance job: less time looking for numbers, more time acting on them. If that sounds like your world too, let’s talk.
Whether it’s a dashboard, a pipeline or an AI assistant, I’m always happy to chat. Send me a message, or use whatever’s easiest for you.