How a Build Works

A joint practice from The Reagent Co. and Fullstacked. ← Back to how we work

How a build works

We build AI-enabled tools to save your team hours and to show your CEO better results over time.

Once a process is written down and working, we build a tool that does the repetitive parts. A person approves every high-stakes decision. Most builds take 4 to 8 weeks, because we don’t hand over a tool until your team has tested it on real work, twice. You own the tool, and we train your team to run it.

The build, start to finish

Before we build

A tool only works if the process underneath it works.

A tool built on a messy process makes the mess faster. So before we build anything, we ask two questions. Is the work written down and followed the same way each time? And does it happen often enough for a tool to pay off? Where your process lands tells you where to start.

How often it happens →

Happens often · Not written down yetAudit and redesign first

Worth fixing, but a tool would only speed up the mess. We fix it on paper first, then build.

Example: performance reviews, when every manager runs them differently

Happens often · Written down and followedBuild now

Ready, and frequent enough for a tool to pay off. We go straight to the build.

Example: hiring, for an organization filling 20 or more roles a year

Happens rarely · Not written down yetStart with an audit

Probably not your first tool. An audit shows whether it’s worth fixing, or whether another process is costing you more.

Example: replacing a senior leader, when there’s no succession plan and every search starts from scratch

Happens rarely · Written down and followedBuild something simpler

A rare process won’t save many hours, but it’s easy to forget and hard to defend. A simple tool keeps the steps, the files and the reason behind each decision in one place. It runs the same way next year, even if the person who ran it has left.

Example: a once-a-year pay review, where every decision needs a reason you can show later

Written down and followed the same way each time →

If both answers are yes, we build. If not, we’ll tell you, and point you to the right first step.

Most builds take 4 to 8 weeks.

Every build includes two rounds of feedback from your team on real work. Beyond that, the biggest factor is how quickly we get access to your systems and approval from your IT team.

About 4 weeks

Simpler builds

The tool works alongside what your team already uses, like forms, spreadsheets and email, and system access comes quickly.

Up to 8 weeks

Larger builds

The tool connects to more complex systems, like your HR or hiring system, IT approval takes longer, or your team needs more time between feedback rounds.

How it works

Four to eight weeks, three questions answered.

Week 1

Plan and get access

What will the tool do, and what will a person still decide?

We agree which steps the tool runs, and which decisions a person approves
We set the starting numbers from your audit. If there was no audit, we measure your current process for two weeks first, so there’s a real “before” to compare against
We request access to your systems and approval from your IT team
We agree how your private data is handled, before we touch any of it

You get: a one-page plan: what the tool does, what people still decide, and the numbers we’ll track.

Your time: a 60-minute planning session, and an introduction to your IT contact.

First round

Build, and the first round of feedback

Does the tool get the same answer your best person would?

We build the tool around how your team already works
Your team uses it on real work, side by side with your current process
We fix everything they flag before anyone relies on it

You get: a working tool, tested on your own cases, with its results checked against your team’s.

Your time: the people who will use the tool try it on real work, and someone checks its results against their own, about an hour a week.

Second round

The second round of feedback, then handoff

Can your team run it without us?

Your team uses the improved tool on real work again, and we fix what’s left
We train the people who will use it, and hand over a playbook your team keeps, so new staff can learn it without us
We turn on tracking, and put three checkpoint dates on the calendar

You get: a tool your team has tested twice on real work, a playbook, and the checkpoint dates to prove it’s working.

Your time: one training session for the people who will use the tool.

What we ask for

What we need from you.

The person who owns the processTo make decisions with us and check the tool’s results
An IT contact and system accessThe sooner we have access, the sooner the tool is ready
Your rules for private dataSo the tool follows them from the first day
Real past cases to test againstFor example, your last 20 offers, so we can check the tool against what actually happened
The people who will use the toolFor one training session, and their honest feedback

What you get, and why it matters

What you get at the end of the build.

The tool

Built around how your team works

It does the repetitive parts of the process. A person approves every high-stakes decision.

Why it matters: You own it, with no ongoing dependence on us. Your team gets its hours back for the work only people can do.

A playbook and training

So your whole team can run it

A short written playbook your team keeps, plus training for the people who use the tool.

Why it matters: A tool only one expert understands stops working when that person leaves. A playbook keeps it running.

Three checkpoints

Proof of what it’s worth

At the first two, we adjust the tool and the process based on what we see. The third proves the return, in dollars your CEO cares about.

Why it matters: When budgets get tight, the tools that survive are the ones with numbers behind them.

How we measure what it’s worth

You’ll know exactly what the tool is worth.

The audit measures the “before.” The tool records its own results every time someone uses it, so your team doesn’t track anything by hand. After handoff, we meet at three checkpoints, every two to three months across the first year. At each one, a report shows what the tool is worth to your team and to your CEO.

What your talent team sees

Hours back every month
Offers, reviews and paperwork done faster
Fewer manual steps and fewer errors

Starts in the first month

What your CEO sees

New hires who perform and stay
Reviews that catch the right people early
Fewer costly people mistakes

Shows up over the first year, sometimes longer

Time saved pays for the tool. Better people decisions keep paying after that. The tool tracks both from the first day. At the last checkpoint, we put the results in dollars, in a number your CEO cares about.

When you’ll see what

Some results show up right away, like hours saved and faster turnaround. Others, like whether new hires stay, can take a year or more. So at every checkpoint, we track the early signs that the bigger results are coming. We set the checkpoint dates up front, so a quiet few weeks after handoff doesn’t look like failure.

When	What you’ll see
The first month	Effort: hours saved, and how often the tool is used
Checkpoints 1 and 2	Early results: new hires still there at 90 days, and first-review scores
Checkpoint 3, and the year after	What your CEO cares about: new hires who perform and stay, and fewer costly people mistakes

We track effort, like how often the tool is used, only as an early sign of the result, never as the goal.

Sample report: salary offers

What we measure	Before	After	How it’s tracked
Time to set a salary for an offer	Up to 3 days	24 hours	Every request is time-stamped when it comes in and when it’s answered
Roles decided by the written rules	None. Every offer went to one person	80%	The tool records which rule set each salary
People who can approve	1	2 to 3, named in advance	The tool records who approved each offer
Hours your team saved	Measured from the first month		Time spent per request, before and after

These numbers come from our salary offer example, based on a redesign Helen led at a previous organization, with details changed to protect confidentiality. Once a tool runs the process, it keeps tracking them at every checkpoint.

Before you say yes

What you’ll want to know.

Who makes the final call?A person approves every high-stakes decision. The tool prepares the work, and your people decide.
What happens to our private data?It stays in your own systems and follows your rules. If we ever need to take data out, we take only what the work needs, with names removed, and we delete it when the work ends. We agree on all of this in writing before we start.
Who owns the tool?You do. It keeps running without us.
What if it gets something wrong?Your team tests it on real work in two rounds before handoff. After that, a person checks every high-stakes result, and the first two checkpoints are there to fix what we find.
What does it cost to run each month?We give you the exact monthly cost before we start building.
How long does it take?Most builds take 4 to 8 weeks, including two rounds of feedback from your team on real work. Builds that connect to more complex systems take closer to 8.

Results from tools we’ve built

8×faster role banding, with 85% agreement with senior HR leaders’ judgment

15×faster workforce reassignment analysis, with 100% agreement on direction in a blind peer review

3tools still in use by a former team, with no one from our side maintaining them

Not sure your process is ready? Start with an audit.

In three to four weeks, we tell you which steps are ready for a tool now, and which need fixing first.

Book a call → See how an audit works → ← Back to how we work