Fixed-fee cost & quality audits for B2B SaaS shipping AI — an eval harness you keep, whether you’re proving a new feature works or your bill is climbing.
Fully remote · EU-based · async-first · US & EU clients
Not sure where AI fits in your stack yet? See the AI Opportunity Audit →
B2B SaaS teams — from bootstrapped to Series C — that have shipped an AI feature (customer-facing Q&A, internal copilot, or AI analytics) and aren’t sure the cost, quality, or ROI adds up. If your LLM bill is climbing past $20k/month, the LLM Cost & Quality Audit is the sharp tool; earlier than that, start with the AI Opportunity Audit or the open-source toolkit. Decision-maker: CTO, VP Engineering, Head of AI — or the founder who owns the technical roadmap.
Or — you’re shipping AI features but not sure where the ROI is: ERP exports living in spreadsheets, manual workflows, data scattered across systems. The AI Opportunity Audit maps where AI and automation actually fit — and where they don’t — before you build.
Your board — or you — asks what the AI line item is buying
Hallucination incident in production
You shipped an AI feature months ago — and no one’s measuring if it works
Pre-investor demo or due diligence
Productized, fixed-fee engagements. You always know the scope and the outcome.
You already know roughly what you want done. We scope it: feasibility, effort, go/no-go — before you commit to a full engagement.
You don’t yet know where AI or automation fits in your stack — we map the whole picture and show you where the ROI is. Mapped by a senior data + AI team, not a generic AI consultant.
Productized one-shot fix when one specific thing is broken.
Managed eval-as-a-service for LLM features in production.
Observability dashboards show what happened. This runs the eval, alerts on drift, and writes the report.
When you need the feature built, not just audited — RAG pipelines, agent loops, and LangChain / LlamaIndex integration on your real data warehouse. Built by the same two senior engineers, no junior handoff.
Available for teams ready to build from scratch.
Scope and fee discussed in discovery.
Every engagement is fixed-fee, scoped up front — you’ll have exact scope and pricing in a written proposal, usually within a couple of business days.
AI cost is the hook — but it’s rarely the only thing bleeding. These come up in discovery; we take them on as part of an engagement, or as a standalone Paid Discovery.
The same eval harness we run in audits is public on GitHub (MIT). Self-host it, wire it into your CI, set your own quality bar. When you want it run for you — continuously, with drift alerts — that’s the retainer.
Open-source toolkit, your CI, your quality bar.
Managed eval & cost monitoring, drift alerts to Slack, incident response.
Audit, optimize, ship — the eval harness stays in your stack.
A two-person senior team — data + AI engineering for data-heavy SaaS.
Pure data engineers can’t ship production RAG. Pure AI engineers can’t ship on a real data warehouse. The combination — data engineering at scale plus eval-driven RAG — is genuinely rare. It’s what this team is built on, and exactly what data-heavy SaaS teams need in 2026.
Between us: 8 years of Lakehouse-scale data platforms and production GenAI evaluation — petabyte-scale infrastructure, then production RAG with measurable quality SLAs. Two senior engineers, both with commercial AI experience — one data-and-AI, one AI-focused.
Fully remote, based in the EU. Async-first — US & EU clients, minimal meeting load.
You work directly with both of us on every engagement. No junior handoff, no account manager in the middle.
Engineering scope only — we build the artifacts; your legal and compliance team owns regulatory interpretation.
Public LCQA eval toolkit on GitHub →
Reproducible eval methodology + harness — inspect the code before you hire. MIT licensed.
You’ll get a written proposal within a few days — or an honest no.
Thanks — your message is in.
You’ll hear back within a couple of business days.
No sales sequences. No follow-up cadence. One human reads this.
Your details are used only to respond to your inquiry, never shared. Privacy.