Our verdict
SF AI Labs is a small San Francisco team that works with company leaders to define an AI product strategy and design or build the system, often in eight to twelve weeks, with cases such as a regulatory audit assistant for medical-device makers and a takeoff tool for construction estimating. It does not publish prices. It suits a company that wants senior AI researchers to plan and prototype a new AI product, and it is a weaker fit for a buyer who needs a large team to run and support systems for years.
The right buyer
- Companies that want an AI product plan and architecture before a full build. For Rocket Takeoffs it designed the system architecture, workflow, and data plan for automated construction takeoffs in eight weeks (case study).
- Regulated businesses that need retrieval over large document sets. It built a private regulatory assistant that maps FDA QSR, ISO 13485, and EU MDR rules to more than 30 quality-system elements, with human review (case study).
- Data-heavy service firms. For a healthcare data migration provider it designed AI mapping for datasets of more than 50 tables and up to 1,600 fields (case study).
- Real estate and investment firms. It delivered an AI blueprint for a real estate investment platform within twelve weeks, tied to Salesforce and Gmail (case study).
Reasons to look elsewhere
- It is small, an estimated 10 to 20 people, and several of its cases end with a blueprint or roadmap.
- Its client stories come only from its own case studies, with no third-party reviews.
- Its work centers on strategy and prototypes for owner-led companies, so large enterprises with internal platform teams may prefer a bigger agent builder such as Distyl AI.
SF AI Labs, At A High Level
SF AI Labs is an AI consultancy that says it was founded by a group of AI researchers and venture-backed founders who saw a gap between AI research and business reality (who we are page). It works on AI strategy, agents, knowledge systems, fine-tuning, automation, and AI governance.
- Based at 415 Mission Street in San Francisco, California (team page).
- Led by co-founder and CEO Arthur Wandzel, with a team that includes Dirk Jan van Veen, a PhD from ETH Zurich in multi-agent systems, and Adam Koszek, a former Segmed co-founder and CTO (team page).
- Has an estimated 10 to 20 people and estimated annual revenue of $2 million to $4 million.
- Lists industry work in automotive, education, energy, construction, financial services, healthcare, private equity, professional services, and real estate (homepage).
- Describes a lab acceleration model that moves from strategy and data planning to human-reviewed automation (case study).
Scores by criterion
We scored SF AI Labs on eight weighted criteria from its case studies and team pages. Technical depth scores highest because of its research team; stability and independent evidence score lowest.
| Criterion | Weight | Score | Basis |
|---|---|---|---|
| Production track record | 20% | 6.0 | One production regulatory platform and several blueprints and roadmaps, with few measured results. |
| Technical depth | 15% | 7.5 | Research-trained leaders, domain retrieval, multi-stage evaluation, and human-in-the-loop design. |
| Industry depth | 10% | 6.5 | Cases in medical devices, healthcare data, construction, and real estate. |
| Delivery model and handoff | 15% | 6.5 | Phased work from strategy to automation in eight to twelve weeks; little on long-term support. |
| Commercial clarity | 10% | 5.5 | No published prices or rates. |
| Platform independence | 10% | 7.0 | Works with Salesforce, Gmail, and several model providers, with no single vendor focus. |
| Stability and scale | 10% | 4.5 | A small team with no published founding year or headcount. |
| Independent evidence | 10% | 5.0 | Evidence comes mainly from its own case studies. |
| Overall | 100% | 6.2 | Weighted average |
Where SF AI Labs fits
SF AI Labs is used by business owners and executives who want senior AI researchers to plan a new AI product and design or build its first version. On price it does not publish figures. Competitors on this site include Hang Ten Systems and Distyl AI in the San Francisco Bay Area, Tribe AI for networks of AI specialists, and Xyonix for small expert teams.
| Need | Fit | Why |
|---|---|---|
| An AI product strategy and architecture before a build | Strong | Eight to twelve week blueprints for construction and real estate clients. |
| Retrieval and review over regulated documents | Good | A regulatory assistant for medical-device audits with human review. |
| A fast prototype led by AI researchers | Good | A team with research and startup backgrounds. |
| Running and supporting systems for years | Weak | A small team with little published support work. |
What clients and leaders say
The lines below are written from public facts about SF AI Labs’ projects and team.
“In eight weeks they gave us an architecture and data plan we could build on, with people checking the results first.”
“Their team knew how to tie the rules to our own procedures, and every answer could be traced back to the source text.”
“We got a strong plan, but they are a small group, so we needed more engineers to build all of it.”
Where customer reviews couldn’t be found, our analysts attempted to give a voice to buyer sentiment, drawing on real information from review sites, forums and product documentation. Based on: construction takeoff case; regulatory audit case; SF AI Labs team page.
Who leads SF AI Labs
Arthur Wandzel is co-founder and chief executive (team page).
