Start with four questions
Stage, vertical, how you use AI, and what is forcing the purchase. That is enough to sketch the right stack.
Answer all four questions and your recommended stack appears here.
The startup core, in two packages
Management Liability
D&O plus EPLI, usually with Fiduciary and Crime attached. This is the package your term sheet and your board care about.
Tech Professional
Tech E&O blended with Cyber, with Media and IP where content and AI output create claims. This is the package your customers audit.
AI risk is the flagship
Every large platform sells a version of the core above. AI liability is where the market is actually moving: dedicated standalone programmes at Lloyd's, endorsements that affirm AI, and, on the other side, new ISO GenAI exclusions arriving at 2026 renewals. We built Helix around that split. Start with the AI insurance hub or the buyer's guide.
A dedicated standalone AI liability policy from a Lloyd's coverholder built only for AI risk. It sits alongside Tech E&O and covers the AI specific claims that general forms are starting to exclude or sublimit.
Access via our wholesale partners
A standalone, claims made GenAI liability policy for companies that deploy generative AI in their products and operations. It is not built for companies selling AI models; those need a builder product such as Tech E&O with AI cover or a standalone builder policy.
Access via our wholesale partners
Embroker's management insurance suite: one application for D&O, employment practices, fiduciary, and Tech E&O/Cyber, bought bundled or standalone. Its AI endorsement is one of the most specific in the startup market, including a pre claim algorithm removal benefit.
Access via our wholesale partners
Industries we build for
Companies building foundation models, AI infrastructure, and applied AI products.
Companies whose AI agents take actions: booking, buying, coding, moving money, operating tools.
B2B and B2C software companies, the core Bay Area startup segment.
Payments, lending, banking infrastructure, and financial data companies.
Digital health, telehealth, health data, and clinical AI companies.
Robots, drones, autonomous vehicles, and AI that acts in the physical world.
APIs, infrastructure, data platforms, and tools sold to engineering teams.
Connected devices, sensors, and hardware led startups.
Biotech platforms, diagnostics, research tools, and life science software.
Our second vertical: professional services firms on the same professional lines desk as tech.
Built by a founder, for founders
Eric Ullrich spent about ten years co-founding and operating a multi site startup incubator in the Bay Area, then pivoted to tech API sales when Covid hit, selling infrastructure to the same CTOs and founders he now insures. Helix is the book he is building for the long term, which is why the advice on this site names competitors, paper, and sublimits plainly. More about Eric.
Frequently asked questions
What does a seed stage startup pay for insurance?
In the market, a seed stage software company's full stack (D&O, EPLI, Tech E&O and Cyber, GL, Workers Comp) generally runs $3,000 to $10,000 per year. A California Series A programme generally runs $30,000 to $60,000 per year all in. General market ranges, not quotes.
Are you a broker or a carrier?
A retail broker. We shop your programme across carriers and specialty programmes, including broker only AI markets like Armilla and Testudo, accessed via our wholesale partners. The direct platforms can only sell you their own programme.
Can you place AI liability insurance?
AI risk is our flagship vertical. Access to the dedicated programmes on this site runs via our wholesale partners while our own appointments are set up ahead of launch, and we will tell you the access route for every programme we recommend.
When should we buy D&O?
At the first priced round or first outside board seat. Your term sheet will usually require it at closing.
Not sure where your programme stands?
Send the renewal, the term sheet, or the customer clause. We will read it and tell you what it actually says.