V7

The agentic tax is real. The more agentic the system, the more time and tokens it spends deciding how to solve the problem. Sometimes a tighter workflow gets better results, faster. V7 cofounder & CTO Simon Edwardsson on the tradeoff 👇

AI is easy until the documents get messy. PDFs. Excel files. Emails. Slide decks. Phone calls. Images. John Somerset-Irving explains how V7 processes unstructured insurance data at scale without falling over.

"Make it your use case. Not one a vendor is pushing on you." Our resident AI insurance expert John Somerset-Irving on where insurers should actually start with AI. 60 seconds.

Writing good business faster. That’s the goal. Our resident AI in insurance expert John Somerset-Irving explains how V7 helps insurers get there by fitting around existing workflows, surfacing the right information faster, and reducing the manual grind behind underwriting decisions.

Last week, we partnered with InsTech alongside Google and Optalitix for 'What’s Going On? Reality check: AI and pricing in production.' Our session focused on practical AI workflows already live across the global insurance market. A few key takeaways from the discussion: 👉 The market has moved. Top insurers have completed the exploratory phase of AI, and now they're implementing workflows across multiple use cases and multiple teams, at scale. Document and submission ingestion was just the starting point. The most exciting developments are now happening further downstream: algorithmic underwriting, automated appetite checks, dynamic dashboards for underwriters and claims teams, and increasingly sophisticated, auditable multi-step workflows. 👉 The best partnerships are built on speed and a willingness to challenge the status quo. Work with vendors who understand both insurance and frontier AI deeply enough to rethink how work gets done. Weeks, not months, should now be the benchmark. 👉 Transparency builds capability. At V7, we build workflows transparently so your team understands how they work, and can eventually own and extend them themselves. We are your partner to build an AI centre of excellence internally. That’s what a real partnership looks like.

The AI industry is slowly rediscovering why workflows exist. Top of Hacker News today: “Agents need control flow, not more prompts.” The comments are full of teams who let the model run the whole show, hit a wall around file 30, and quietly added structure back in. That’s what V7 Go is built for. Let models handle ambiguity, reasoning, and unstructured data. Everything else needs: • structure • control flow • verification • audibility • human review https://bit.ly/4dBGF3d

"We publish salary ranges on every single job posting without exception. But transparency doesn't stop at the job posting. Internally, we're fully open about how those numbers are built: what tools we use, what market data we reference, which variables we take into account. Employees can access all of that." At V7, we believe that pay transparency shifts the conversation to the things that really matter, like our mission, culture, impact and growth trajectory. Curious to learn more about our approach, and the challenges it brings? Learn more in our article with Welcome to the Jungle: https://lnkd.in/erdy8Jim

Sticking with the same model for every job is how you end up paying Opus prices for work Haiku could've done in its sleep. V7 co-founder and CTO Simon Edwardsson on choosing the right one. In V7 Go you can swap models per step and benchmark them side by side.

Can you believe it's May already? Here's just some of the features and improvements our team shipped in April. To learn more, check out our full Changelog here: https://lnkd.in/eekyGpcR

The hard part of AI at work isn't the AI. It's trusting it. Our co-founder Simon Edwardsson on how V7 Go solves that, in 32 seconds.