How we use AI to automate our BizOps

I run finance and business operations at Adapt. I spend my days in spreadsheets, workflows, and reporting.

At Adapt we use Adapt. And that has fundamentally changed my job. The work itself is the same, but getting to insights that can move the needle has sped up.

I am not an engineer, and I don’t know how to write code. But with Adapt, I now have the capabilities of a data analyst and a data engineer available to me whenever I need them, and it has shifted what is possible in my role.

No more schlep work

I love spreadsheets. It’s the backbone of what I do. I have strong opinions about the merits of Sheets vs. Excel (unpopular opinion: I prefer Sheets).

Forecasting, scenario planning, revenue design - I enjoy hashing out financial models in columns and rows. That part of my job is not changing anytime soon, and I don’t want it to.

However, the part that used to slow me down was getting to a first version of a model. I don’t enjoy the drudgery of laying out the structure, writing the initial formulas, and building something I can start reviewing and stress-testing.

That setup phase could eat hours before I even got to the real work, and if the structure was wrong it meant a painful reset.

Now I walk through what I want with Adapt. I tell the agent about the inputs, the assumptions, and the structure I am looking for. We go back and forth on the layout, purpose, and use case.

Then it builds me a first draft. So, now I go from a blank page to an output in a few minutes. All the source data I need is built out and included. That alone can be half the battle; creating different reports in your CRM or data warehouse just to get the specific data points you need and then getting them into your model.

Recently, I needed to model out volume discounting for our pricing tiers. I knew the process to run: set up tier breakdowns, layer in annual commit discounts, and stress-test margin floors at every level.

I described the structure to Adapt and had a workbook back in minutes with credit tiers, discount schedules, margin floor analysis, and a summary view I could use to pressure-test the whole thing. My time was spent tweaking assumptions instead of setting up the views.

The output is still mine to own and refine and is set up the way that I want it. I am still the one pressure-testing the assumptions and making sure the final version makes sense. Adapt gets me off the ground faster than I could on my own.

Business intelligence without the BI tool

Reporting has always been another pain point. Either I maintain something by hand in a spreadsheet, refreshing exports and updating formulas every week, or I put in a request for a proper dashboard and wait for it to get prioritized.

The spreadsheet route works but does not scale. The engineering route scales but I do not control the timeline or the output.

With Adapt, I now build my own. I describe what I want to track, what the data sources are, and how I want it laid out. Adapt builds a working dashboard I can deploy and share with the team.

I don’t write SQL or front-end code, and I don’t tweak any WYSIWYG tool. Adapt even follows our brand guidelines automatically, so the output looks right without any design work. It’s amazing. I go from "I wish I had visibility into this" to a live, shareable view in a single session.

We have a leading indicator of customer retention that summarizes accounts with specific levels of usage, users, and integrations. I spun up a dashboard around that metric in a few minutes that tracked it over time against our broader account base, segmented by how closely each account matches our ideal customer profile. It was a quick visualization that gave us an immediate understanding of how we are progressing on our product-market fit journey. It’s connected to our BigQuery instance, so it’s a live tool.

The biggest benefit of using Adapt has been keeping up with the speed of change. When priorities shift or a new metric becomes important, I do not file a ticket with a data engineer. I open the dashboard in Adapt, describe what I need, and it is live. The iteration cycle is minutes, not sprints.

Freedom to ask better questions

Dashboards cover recurring metrics. But most of my questions come up once, need an answer fast, and move on.

I know what data lives in BigQuery, HubSpot, and Stripe. I have a mental model of how those tables connect and what questions they could answer. What I do not have are any SQL skills to make it happen.

Previously, any new analysis meant either a spreadsheet built off bulk exports or looping in someone technical. Now I describe what I need and Adapt handles the rest. It writes the query, runs it, and brings back the answer. It joins tables that would have taken me a long time to figure out on my own. I can ask a question and have an answer in seconds that used to take days, if it got prioritized at all.

I can also send Adapt on quick sniff tests. After enough time in a role like this, you start to understand trends in your business and expect certain outcomes. Chasing down an outlier or anomaly used to be a long process. Now I can tell Adapt briefly what feels wrong, what I expect, and have it chase down the root cause.

A recent example: we have a customer whose usage drastically increased week over week. They had been growing steadily before, but saw a step function jump one week. Adapt dug in and highlighted that they had set up scheduled tasks in some of their public Slack channels that was driving additional usage. We reached out to the customer to flag it and learned that it was a new part of their workflow that they were excited about.

The thing that surprises me most is how quickly it compounds. I have started asking better questions. I know that one question is not going to spin someone up and burn their whole day. With Adapt, I stopped approximating and started checking actual data. That shift in speed changes how I make decisions.

How I trust the output

The natural question from anyone in finance is: how do you know it’s accurate?

I check. Every time Adapt writes a query, I can see it. I do not read SQL fluently, but I can follow the logic well enough to spot when something is joining on the wrong field or filtering out records it should not.

I cross-reference outputs against known benchmarks. If a number looks off from what I would expect, I push back and we dig in. The same judgment I would apply to a report from a colleague applies here. Adapt is fast, but I still make the call on whether or not the answer makes sense.

Where the moat moved

My personal moat used to be knowing where all the data lived, the caveats that had to be called out, the edge cases that needed to be checked, the bodies buried in the closet. That knowledge still matters. But it is no longer bottlenecked behind my ability to extract it myself or my ability to get time on someone else's calendar to extract it for me.

The most valuable skill I bring to this work is not technical. It is knowing what questions to ask and understanding the business well enough to know what the answers mean.

The bottleneck was always the gap between having the right question and being able to get the answer. For me, that gap is gone. And once it is, you realize how much of your day was spent working around it.

If you’re in finance, ops, or any analytical role where the gap between knowing the right question and getting to the answer slows you down, get early access to Adapt.