Owner-led software consultancy · Stockholm

We build the systems that run your operations.

Owner-led software consultancy in Stockholm. We diagnose the real bottleneck, build the working system, and keep the critical parts aligned — custom software, AAOS, integrations, infrastructure, and agentic AI.

Request a bottleneck audit → See the services ↓
An engineer inspecting a humanoid android on a diagnostic table

What we do

Six ways we remove the bottleneck.

One team, one owner, one standard: a working system you can run, not a slideware promise. Fixed-price starting points below.

01

Custom software, web & mobile

Internal tools, customer portals, SaaS features, and dashboards. Scoped discovery, delivered in milestones you can test.

from 95,000 SEK · 1,050–1,250 SEK/hr

02

AAOS & embedded Android

Build, review, and harden Android Automotive and embedded Android apps. The owner has shipped a fleet of AAOS Car App Library apps to Google Play.

1,250–1,500 SEK/hr · 45,000 SEK Play-readiness review

03

System integration & automation

Connect the systems that still need copy-paste. n8n workflows, API integrations, document flows, and CRM/ERP glue with review gates.

from 65,000 SEK · 1,100–1,400 SEK/hr

04

Cloud, DevOps & self-hosting

Set up and harden Linux, Docker, reverse proxies, mail, observability, and backups — including local and private AI infrastructure.

from 75,000 SEK · 1,200–1,500 SEK/hr

05

Technical audits & code review

Architecture, delivery risk, security posture, and AI governance. You leave with a written decision memo and a prioritized backlog.

from 25,000 SEK · 45,000–75,000 SEK standard

06

Agentic AI — our specialty

Diagnose. Build. Supervise & Align. Deterministic agents with LLM-as-a-Judge review, file-based audit trails, and an EU data posture.

25,000 SEK setup · 4,900 SEK/mo supervision

Why the name

We sit with the machine.

In Westworld, the engineers never treated their robot as a finished product. They sat with Delorus, listened to how it thought, and kept refining it. A mind that isn’t tuned is just a very expensive toy.

We think the same way about your systems. Software drifts when nobody owns it. We listen for the real bottleneck, build the system for exactly that work, and keep supervising the parts where drift costs you money. We don’t deploy and walk away. We diagnose, we build, and we keep the critical parts aligned.

An android's exposed endoskeleton on a diagnostic table, an engineer leaning in to examine it.

The work

The method is the same. The work isn’t.

How a bottleneck becomes a working system, across four very different operations.

Specialized Clinical Care

Prior-auth & recall adjudication

Bottleneck
18+ hrs/week lost to manual insurance prior-auth forms, payer rejection callbacks, and unreturned patient recall cycles.
System
An intake-and-adjudication pipeline that parses clinical notes, maps them to statutory payer billing codes, files authorizations, and re-engages lapsed patients conversationally.
Outcome
0% clerical rejection on initial insurer submissions; 32% more recovered 6-month treatment cycles, no receptionist intervention.

Heavy Machinery & Automotive

Warranty claims that clear the first pass

Bottleneck
8–14% of warranty revenue lost because technician work orders lack manufacturer-specific defect phrasing, triggering OEM claim rejections.
System
A technician-to-OEM translation agent that ingests diagnostic logs, cross-references manufacturer warranty databases, and prepares audit-proof repair orders before submission.
Outcome
85% drop in first-pass OEM warranty clawbacks; billing cycles from 12 days to same-day dispatch.

Aviation & Complex Logistics

Crew-rest cascades caught before dispatch

Bottleneck
Minor delays trigger cascading crew rest-period violations, forcing dispatchers into multi-hour scrambling and preventable ground delays.
System
An autonomous telemetry-and-roster supervisor that intercepts delay feeds, computes multi-leg crew-legality cascades ahead of drift, and stages reserve reallocations before violations occur.
Outcome
Cascade failures flagged 90 min before dispatch; 40% reduction in hotel and emergency-reserve standby overhead.

Enterprise Hospitality

After-hours events that never bounce

Bottleneck
35% of high-value event and private-dining inquiries arrive after hours or during peak rushes, bouncing to competitors within a 2-hour window.
System
A reservation-and-event agent wired to POS table inventory and deposit gateways, negotiating group minimums, issuing hold agreements, and securing deposits in real time.
Outcome
100% of after-hours event inquiries captured; 28% more booked group-banquet revenue with zero front-of-house phone disruption.

Illustrative examples of the method in practice, not client results.

Swedish trust shield

Built for high-compliance operations.

The guarantees an institutional buyer needs before a system touches a real workflow.

A

Data sovereignty

Your data stays in the EU. We run in eu-north-1, Stockholm, inside a private VPC. Nothing leaves your region, and nothing is shared with a third party. Ready for BankID / Freja eID and a pre-drafted Swedish DPA (PUB-avtal).

B

Zero model training

We never train or fine-tune a model on your data. Your inputs and outputs are processed for the task and nothing more. A contractual guarantee, not a promise.

C

Determinism & audit

Agents run on pinned model versions, fixed prompts, and a fixed temperature, so a given input produces the same decision. Every decision is logged to file-based audit trails in .json and .md. When a regulator asks what the system did and why, you have the record.

Wired into the systems you already run

  • SAP
  • Salesforce
  • Postgres
  • Jira
  • REST & Webhooks
  • Fortnox
  • BankID
  • DMS / EHR

Working together

An ongoing partnership, not a one-off build.

We run this as a Design Partner Program: we diagnose, build your first system, then keep supervising and aligning it as a monthly engagement. The first 90 days look like this:

  • Weeks 1–2: Diagnose. We sit with your team and find the one task worth automating.
  • Weeks 3–6: Build. We ship a specialized system for that task, wired into your real workflow.
  • Weeks 7–12: Supervise & Align. We run a continuous LLM-as-a-Judge pipeline over every interaction, catch drift and edge cases, and re-align the system. Then it keeps improving.

Prefer to start small? Try the BRF off-plan calculator → — a free 30-second diagnostic for housing associations, no form required.

Let’s find your bottleneck.

A free 30-minute conversation. We listen first, and you leave with a clear picture of where a working system would pay for itself.

No deck, no pitch. We listen first.