Width.ai
US machine learning shop building custom n8n nodes and document extraction
- HQ
- Richmond, United States
- Founded
- Not published
- Team
- 2-10
- Pricing
- project bands $1,000-$80,000 (n8n directory)
- Regions
- North America
- Stack
- n8n, OpenAI/GPT, AWS, Azure, Pinecone, spaCy
- Industries
- Law Firms, Financial Services, SaaS, Ecommerce, Accounting Firms
About Width.ai
Width.ai is a machine learning shop that also does n8n work, which is a different proposition from an automation shop that calls a model API. The published work reflects that: two GPT-3-based internal HR tools for O.C. Tanner covering bias detection and negative-language rewriting, and a marketplace categorization system stated at 97% top-level accuracy across 5,585 categories running 50M records a month. Custom node development sits alongside computer vision, NLP and RAG.
The corporate record is unusually thin for a US firm. No third-party review profile exists on Clutch, G2 or DesignRush and the n8n directory shows zero reviews. The founding year conflicts between the site (2021) and LinkedIn (2020), no address is published on the site itself, and the privacy policy URL returns HTTP 401 rather than a document.
Our verdict
Width is the right call when the hard part of the workflow is the model, not the wiring: document extraction, classification at volume, retrieval over a real corpus. The O.C. Tanner build is named and the accuracy figure is specific enough to argue about, which puts it ahead of the firms below it on technical evidence. What it cannot offer is any independent confirmation that it delivers well. There is no review profile on Clutch, G2 or DesignRush, the n8n directory shows zero reviews, and a 2-10 person team is a real constraint if the engagement needs more than one or two people at once. The firm also publishes no address on its own site, its founding year differs between its own site and LinkedIn, and its privacy policy returns an HTTP 401, so a buyer cannot read the terms before making contact.
Pros
- Built two GPT-3-based internal HR tools for a named client, O.C. Tanner, covering bias detection and negative-language rewriting
- States a marketplace categorization system at 97% top-level accuracy across 5,585 categories at 50M records a month
- Capability spans computer vision, NLP, RAG and custom n8n node development, so a workflow can be backed by models the team builds rather than only third-party APIs
Cons
- No third-party review presence at all: no Clutch, G2 or DesignRush profile exists and the n8n directory shows zero reviews.
- A 2-10 person team, which is a hard capacity ceiling on a larger or parallel-track engagement.
- The founding year conflicts between its own site (2021) and LinkedIn (2020), and no address is published on the site; the Richmond base is confirmed only via the n8n directory and LinkedIn.
- The privacy policy URL returns HTTP 401, so it cannot be read before contact.
Where Width.ai ranks
- Ranked #9 in AI Automation Agencies
- Ranked #10 in n8n Agencies
Is this your agency? Get your badge.