← Back to blog
22 April 2026·4 min read

Inside a BIDLA build: from messy process map to live AI app in 2 weeks.

Two weeks from kickoff to a live internal app used daily by one team. Here's exactly what happens on each of those days.

buildprocessshipping

“Can you just build us an AI tool that does X?” is the wrong opening question. The right one is: what does X look like today, who does it, where does it break, and what would “done” feel like on a Tuesday afternoon?

A BIDLA build is two weeks long by default. Here's the rhythm.

Before: a typical engagement start

  • Client has a vague idea: “we want AI to help us with something”
  • No documented version of the current process exists
  • Several people do it slightly differently without realising
  • Nobody has measured how long the process actually takes
  • There's no written agreement on what “done” means

What we build (the 2-week cadence)

Week 1, map and design.

  • Day 1 to 2: shadow the people who actually do the work
  • Day 3: whiteboard the current process end to end
  • Day 4: find the 3 to 5 steps where AI earns its keep
  • Day 5: approve the design, lock the scope

Week 2, build and ship.

  • Day 6 to 8: build the internal app, small and focused
  • Day 9: internal testing with the user team, not just the sponsor
  • Day 10: ship to production, walk-through, support hand-off

By the end of week two you have a live app on a secure link, used by one team, doing one thing really well. Iteration starts week three on real usage.

After: what changed

  • 2 weeks from kickoff to a live app in production
  • The process is documented for the first time, often surprisingly
  • The team uses the new app within 24 hours of ship
  • Real data is flowing into iteration decisions by week three
  • The sponsor has something concrete to point at internally

One design choice that mattered

We do not build the “nice to have” version in week one. We build the boring, “does exactly the thing” version and ship it. Features that felt essential at kickoff frequently turn out to be unnecessary once the team has the core working.

Iteration on a live system is cheaper and far more honest than trying to design the perfect thing on a whiteboard. Week three tells you more than weeks one and two combined.

Want a workflow like this?

Book a 30-minute AI workflow audit.

We'll map the first 1 to 3 automations in your business that should “just run themselves,” plus what they'd be worth if they worked.