Kindling

Software for the work your ERP doesn't cover.

We build warehouse, order and back-office tools for distribution and service teams in Bulgaria and the EU — and we keep working on them after launch.

  • No middlemen

  • Honest scope

  • Phased rollouts

  • Built to outlast

  • Support after launch

  • Warehouse systems

  • Order tools

  • Back-office tools

  • ERP integrations

  • Customer portals

  • Ongoing support

We're a small team of engineers in Sofia who left full-time chairs to work directly with the people doing the work. Whether it's a warehouse that still runs on Excel or a back office held together by copy-paste, you talk to the engineers who build your system — from the first call to long after launch.

Warehouse systems

Order & back-office tools

ERP integrations

Ongoing support

Warehouse systems

Order & back-office tools

ERP integrations

Ongoing support

How we work

You know the price and the timeline of the build before it starts: both are fixed at the end of the diagnostic sprint, not after the first call.

Step 1

Free process review

30 minutes

We look at one process and tell you honestly whether custom software makes sense.

Step 2

Diagnostic sprint

1–2 weeks, fixed price

Process map, scope of the first module, clickable prototype, exact timeline and price. Credited against the build if you continue.

Step 3

Build the first module

Fixed scope and timeline

We build the module agreed in the sprint and roll it out in phases.

Step 4

Support and development

Monthly

We keep the system running and extend it as your operation changes.

Direct contact, no agency overhead

You talk to the engineers actually building the work. No account managers, no project managers, no relay. The first email is from a developer and so is the last.

Honest scope before code

We won't quote until we understand the process. The process review is free; the price and timeline of a build come out of the fixed-price diagnostic sprint. If custom software isn't a fit, we'll say so and point you somewhere better.

Production-grade defaults

Sensible architecture, typed code, tests where they earn their keep, deployment your team can actually run — rolled out in phases so the operation keeps moving.

AI used where it helps

We integrate Claude, OpenAI, and open-source models when the use case is real — semantic search, agents, structured extraction. We don't bolt LLMs onto things that don't need them.

Long-term partnership over one-shot delivery

We keep working with our clients after launch — our Intercars and Agriventures projects are both ongoing. We optimise for what your system looks like in year two, not what looks fast in week three.

Still running operations on spreadsheets? Let's look at one process together.