# Softblues

> Softblues is a reviewed AI agency in London, United Kingdom, listed in the AIAgencies.eu register and last assessed 2026-09-23.

Canonical page: https://www.aiagencies.eu/agency/softblues
Generated: 2026-09-28 · Curated by Gabor Kiss · No paid placements — inclusion, score, and rank are never sold.

## Facts
- Location: London, United Kingdom
- Website: https://softblues.io
- Hourly rate: €45-85/hr (directory-listed)
- Team size: 10-49
- Founded: 2014
- Services: AI Automation, AI Agents, AI Consulting, AI Development
- Industries: Healthcare, SaaS & B2B, E-commerce, Legal
- LinkedIn: https://www.linkedin.com/company/softblues

## About
Softblues is a London AI implementation firm founded in 2014, building multi-agent pipelines, retrieval-augmented support agents, and voice agents for mid-market companies. It is a registered Anthropic Partner Network member.

## Documented clients
Documented engagements: Roof Maker, SofiaHR, Key Capital, Agrotop, and AutoBI. All 5 carry a source URL on file.

## Curator assessment
Reviewed 2026-09-23 by Gabor Kiss against published evidence.

Softblues labels each flagship case study with the stage it reached, and the labels hold up: Roof Maker's eight-agent voice line is live, answering every call 24/7 and writing HubSpot tickets, while SofiaHR's assessment pipeline—ten fine-tuned Gemini Flash models validated at 90%+ agreement with expert analysts—runs in beta across 10 clients and 300+ interviews a week.

Of the eight flagships, one is live at a named client, two are in beta, one is the firm's own internal system, one is a Claude rollout and training program, and three stopped at an audit, a design or a proposal—the stack detail is deep, the production record short.

Where a system is live, the reported results are capability measures—24/7 coverage, 97% speech-recognition accuracy, sub-two-second replies at Roof Maker—rather than calls resolved or cost removed.

The weaker fit is buyers who need a named reference system proven at volume, or an in-person London team—the site describes a distributed European team working remotely with clients.

### Key strengths
- SofiaHR documents its training data and validation: about 20,000 labeled phrases per dimension drawn from 2,000+ expert-coded interviews, checked against held-out interviews before go-live
- Stack specifics throughout, monitoring included—Roof Maker names Aircall, Deepgram Nova 3, GPT-4.1-mini, ElevenLabs Flash v2, LiveKit and Langfuse for tracing
- Candid evidence labels: the Key Capital study states it reports no measured savings, and the homepage calls its value objectives "not results we have delivered"

### Good to know
- One flagship names a client running the system live—ask to speak to Roof Maker or a SofiaHR beta client before committing to a production build
- The back catalog files conventional builds, such as AutoBI's fleet SaaS and AgroTop's farm platform, under AI Product Development with no AI in the stack—judge AI depth on the flagships
- No team page names the engineers—ask who builds, who operates the system after handover, and where the people doing the work are based

---
Source: AIAgencies.eu — the curated register of European AI agencies. Rate provenance: "agency-confirmed" = disclosed via submission or claimed profile; "directory-listed" = Clutch or two agreeing B2B directories (lone unverified sources are never written). Full method: https://www.aiagencies.eu/methodology · Rate index: https://www.aiagencies.eu/rates