Background
Priya had led engineering at a data startup before co-founding Clio. She knew exactly what six weeks of infrastructure setup looked like: workspace auth for B2B accounts, per-workspace usage tracking for billing, an admin panel for their non-technical ops person, and a backend architecture clean enough for two engineers to work in parallel without stepping on each other. She'd scoped it all out. Then the investor pushed the demo date three weeks forward and the scope stopped being theoretical.
The challenge
The pressure wasn't just time — it was the specific combination of things Clio needed. B2B SaaS has different auth requirements than consumer products: workspaces, team invitations, per-account billing. The team needed those things working correctly, not hacked together, because enterprise customers ask hard questions during evaluations. They also needed an admin panel that a non-technical operations person could actually use — not a database console, not a Stripe dashboard, but a proper interface with user management and subscription controls.
How they built it
Infrastructure week became infrastructure day
Priya's original estimate was two weeks for infrastructure. With ShipAI, it took three days. Auth with workspace support was a configuration task, not a construction task. The admin panel was already built. Usage metering existed and just needed to be wired to their query endpoints. By day three, both engineers were working on the analytics core — the only part of the product that required their specific expertise.
The ops person was in the admin panel on day three
Clio's operations person isn't technical. Priya had expected to spend time building a simplified admin UI for her. Instead, she handed over the admin URL on day three and walked her through the existing views. User management, subscription controls, account history — it was all there. By the end of that first week, the ops person was handling user questions without involving engineering.
Two engineers, one codebase, minimal friction
One engineer owned the analytics query engine. The other owned data ingestion and workspace management. The clear service boundaries in the codebase meant their work rarely overlapped. Merge conflicts were minimal. They didn't need to agree on patterns because the patterns were already established — they just followed them.
Demo day, eighteen days in
The demo was live analytics on real customer data, with per-workspace authentication, subscription-based access controls, and a polished frontend. The investors had moved the date forward expecting Priya's team to scramble. Instead, the product looked more complete than they'd anticipated. The beta launched publicly the following week.
Outcomes
Beta shipped in 18 days against a 6-week estimate
The team compressed their entire launch timeline by two-thirds. Infrastructure that had been estimated at two weeks took three days.
Operations team self-sufficient by week one
Clio's non-technical ops person was managing user accounts and subscription questions independently before the analytics product was complete.
No backend specialist added to the team
Priya had budgeted a possible contract hire for backend work. They shipped a production B2B product without it.
Demo advanced without scope cuts
When the investor moved the demo date forward, Priya didn't cut features — she cut infrastructure time. The demo showed the full product.
In their own words
Every startup I know of this size ends up in the same situation: you need backend infrastructure that would take a specialist weeks to build correctly, but you can't afford or justify a backend specialist at this stage. We got out of that trap. The infrastructure came with the product, not before it.
“We're three people. We don't have a dedicated backend engineer and we probably never will at this stage. We needed something we could all understand, extend, and not babysit. ShipAI is the first foundation where that's actually been true.”
— Priya Ramachandran
Frequently asked questions
How did Clio handle B2B workspace auth?
The Better Auth foundation already supported multi-user sessions and team workspaces. Priya extended the session model to include workspace scoping — a few hours of work following the existing patterns, not a rewrite.
Did the team plan to use AI features from the start?
Not in the beta. Their first release was analytics infrastructure — ingestion, querying, dashboards. AI-assisted analytics summaries were on the roadmap for the next sprint. The AI infrastructure was there when they got to it.
What was the hardest part of the 18-day sprint?
The analytics query engine itself — the actual product differentiation. Priya says that was the only genuinely hard part. Everything around it was handled.