About Us

We know accounting software.
We built the last one.

The Founders

👤

Co-Founder 1

👤

Co-Founder 2

Part of the founding team that created the online accounting category, then went off to do other things. Between them, over twenty years reimagining accounting for small businesses, building integrated financial services, launching the partner ecosystem, and scaling it to millions of businesses worldwide. But as they watched QuickBooks continue to grow, they felt it drifted from its original mission: simplifying the complex and being a fierce advocate for accountants. With the shift to AI and agentic services, they saw the chance to start over and to build an accounting platform where AI isn't a feature but the foundation, where simplicity for the business owner and depth for the accountant aren't trade-offs, and where the starting point is a blank page.

Claude

AI Co-Developer

Built by Anthropic, Claude serves as Imprest's co-developer — writing code, designing architecture, and solving problems alongside the founding team. We believe in transparency: this product is built with AI, and we're proud of it.

AnthropicArchitectureFull-StackCode GenerationTesting

Claude

by Anthropic

Built with modern technology

Rails 8.1 + Next.js powering a full-featured accounting platform.

Backend
Rails 8.1
36,390 lines
Frontend
React 19
40,440 lines
Database
PostgreSQL
Build
Next.js 16
Styling
Tailwind CSS 4
Data
SWR
Async
Solid Queue
Codebase composition69,000+ lines total
Ruby 36%
TS/TSX 49%
Migrations 3%
Docs 7%
Config 5%

Infrastructure at a Glance

The building blocks behind a full-featured accounting platform.

🗄️
66
Rails Models
🔌
139
API Endpoints
⚛️
91
React Components
🪝
5
Custom Hooks
🤖
17
AI Agent Types
📦
140
DB Migrations
💎
38
Ruby Gems
📦
24
JS Packages

Key Integrations

61+ app connections across banking, payments, payroll, e-commerce, and more.

Banking
Plaid
10,000+ financial institutions
Payments
Stripe
Invoice payments & processing
Email
Gmail
Invoice & receipt scanning

AI Models via Anthropic

Claude powers all AI features — from transaction categorization to expense matching agents. We pick the best model for each task.

ModelProviderRoleAccess
Claude Sonnet 4.6AnthropicPrimary agent modelAnthropic
Claude Haiku 4.5AnthropicParsing & categorizationAnthropic
GeminiGoogleAvailable via adapterGoogle AI

Our Ethos

The principles that guide how we build, lead, and operate.

Nine is Enough

We don't think any company needs more than 9 employees. With the right people, the right tools, and AI as a force multiplier, a small team can outperform organizations 100x its size. We're proving it.

People are Product

People, not process, build products. People mistakes are poison — make resolving them a priority. When picking your own professional opportunities, over-index on the people you work for. Their track record is in the public domain and it's crystal clear whether they value the product.

Purposeful Friction

Friction is a controlled weapon at your disposal — don't shy away from it, create it. Purposely design the right type of friction through org design and leadership team composition so the right critical decisions get surfaced directly.

Constructively Contrary

Organizations cycle from energized and depressed, neither are helpful, but the biggest risk is that success breeds complacency. Be calm during the roughest seas, but the most important thing is to be paranoid when everyone around you is calm and content — that's when the sloppy work happens.

No Handlers

Be nicer, humbler, and lower maintenance with each success — don't have an entourage, don't talk through people, don't work behind people's backs and never assume your position comes with assumed power. Every day it needs to be earned. With every gatekeeper you add, the less you understand the lay of the land.

Extreme Awareness

When people start to laugh at your bad jokes, you have a problem. You used to get feedback or at least authentic responses; now you have to be highly aware on your own. You have to read the room, from nervous silence to an awkward beat. Spend time playing back what happened during the day to find those cues.

Instinctual Calls

Data is for optimizing, big calls require pattern recognition, strategy, and conviction (and they are all fuzzy). Leaders are there because they have the courage to take action against an unknowable future. Be ready for the smart well-intentioned risk averse people to provide a steady stream of paper cuts — absorb it ... no revel in it.

Edit, Refine, Reduce

Your job is to only do a couple of high impact things perfectly every year and that means choosing the right small handful of things. Simplifying is one of your most foundational jobs — do less than you think you can if you want to do great things.

No Parroting

Learn from other people, don't blindly emulate them. The best product people are artists with unique voices. They learn and observe from others, but recognize what makes them and their situation unique and different. That said, do copy the best most forward-looking products as step one.

Cut Corners

Great product people thoughtfully cut corners — the very best cut more corners. When a team has demonstrated the ability to work smart, fast, and beautifully — give them rope and let them operate. Circle back to understand why they cut that corner. Operationalize their corner cutting for broader consumption.

Storytellers

Talking is a skill not unlike building. Some are amazing at it, but it can be just talk. Builders may not value talk, which makes them harder to identify — especially if you value and have relied on your own ability to tell stories. You have to meet builders where they are and always be skeptical of great talkers. Great talkers burn the most money.