How we build Fyord.

Fyord is an AI employee operating on real customer money, real customer relationships, and real warehouse inventory — always with a person approving. The bar for reliability is high. Here are the principles we work from.

1. 01

   Best-in-class providers and state-of-the-art models.

   We don't pin to one vendor or one model. Fyord uses Anthropic's Claude family — Sonnet for reasoning-heavy work, Haiku for cheap-and-fast triage — and we revisit model selection as new releases land. Our orchestration layer abstracts the model so we can swap upward as the state of the art moves.

2. 02

   Investing in agent runtime and harness.

   Naïve LLM calls aren't enough. Fyord runs on a proper agent harness — structured tool use, retry semantics, deterministic state machines around the non-deterministic parts, and a clear separation between agent reasoning and side-effecting actions. Failures are observable; retries are bounded; nothing happens twice that shouldn't.

3. 03

   Mechanisms for agents to learn your business.

   A generic AI doesn't know that BG-4417 is the same as “regel 45×95” to your customer Ola. Fyord's agent knowledge — customer-specific article aliases, notes and rules your team teaches the agent, suggestions from your reviewed historical orders during onboarding, and your own checklist rules — makes the agent better at YOUR wholesale, not at wholesale in the abstract.

4. 04

   Guardrails and loops.

   Every step is gated. Before approval: does the draft pass your checklist? Are line items recognized? Is the customer resolved? After your approval: did Fortnox accept the order? Do the prices Fortnox set match expectations? When in doubt, no order is created — Fyord drafts a question to the customer for you to send instead. Bounded retry loops with backoff handle transient failures; persistent ambiguity escalates to a human.

5. 05

   Reliability through layered architecture.

   A single agent making every decision is fragile. Fyord uses specialized sub-agents — one for classification (is this an order?), one for extraction (what are the line items?), one for clarification drafting (how do we ask the customer?). An orchestrator coordinates. Excel and CSV attachments are pre-parsed into tables; PDFs are read directly by the model. A separate model evaluates your own natural-language checklist rules against every draft. Human-in-the-loop is built in, not bolted on: every order is reviewed and approved by a person before it is created in Fortnox.