Tripling Engineering Productivity with AI in 16 Months
Here's a must read for founders/CEOs/engineering leaders 👇 Darragh + team answered 70+ questions about how we've tripled productivity in our product/engineering org over the past 16 months using AI.
Model Architecture Dictates Reasoning Style: Commit vs Explore
Wenlong Wang from our Fin AI group published a great paper at ICLR 2026 on Looped Transformers. With the speculation that Claude's mysterious Mythos is built on a looped transformer (i.e. a model that thinks by quietly looping over its...
Triple Engineering Productivity with AI: Essential Tools & Changes
🚨Attention CTOs + Founders…🚨 We are hosting a proper deep dive into what it takes to triple engineering productivity with AI: the tools you must build, the changes you must make, and the new standards you must set. If you’re...
Fin.ai Launches Top‑performing Open AI Support Agent
There is, of course, one AI Agent for Customer Support with public docs, self serve sign-up, public pricing, a CLI, API + more. It's the highest performing one, and we share all our ideas+research too. Product → fin .ai Research → fin...
AI-Powered Search Transforms Online Shopping Experience
We've just released @Fin_ai for E-commerce solving one of the biggest pain points for buying online. We've finally broken free of the limited land of faceted filters and actually recreated a great in-store experience. The biggest reason buying online sucked...
Survive AI by Rebuilding Platforms for an AI‑First Future
Great point by @eoghan here: surviving AI isn't just • build ✨AI✨ features (2023), • or a Copilot(2024), • or even an AI agent (2025), • it's also... rebuilding your platform for an AI-first world (2026) Systems of record simply won't...
All 8,000 AI Users Simultaneously Demand Universal Reliability
One interesting thing about running a very modern AI product (@fin_ai) with 8,000 customers is that literally all of these users exist at the same time. And you kinda have to make it work for everyone. https://t.co/PM1r4IyQw0
Doubled Intercom Productivity in Nine Months – Full Breakdown
🚨Here's a must read🚨 We 2x-ed our productivity in Intercom in 9 months. Our CTO @darraghcurran shares all the gory details, charts, costs, and more in this thread
Fin’s New “Ask A Teammate” Automates Approval Workflow
"We need a someone to approve this" is a very common statement in Customer Support (e.g. a large refund, or info that needs verification). Fin now supports this with our (just shipped) "Ask A Teammate" feature. Fin will now go...
Scaling AI Feature From One to 8000 Customers
Thrilled to see this one ship to everyone. It's one thing to hand roll an AI feature once for one customer. Rolling this out reliably to 8000 customers is a different type of challenge. Kudos to Pratik & Team...
Stateless or Solo AI Leads to Dogfight Chaos
A lot of AI rocketships end up in the dogfight category because they're either stateless: easy/free to hot swap one for another (and another) or single player: I use [X] and my teammates use [Y,Z] and it just doesn't matter at all....
AI Era's New Table Stakes for Adaptation and Competition
This is a great article about the new table-stakes of adapting to, and competing in, the AI era.
Startups Inflate ARR: Not Annual, Not Recurring, Not Revenue
Startups misreporting ARR is frustrating as an infrequent angel investor who relies on trust (no time to inspect) ~Every startup claiming 0→$[X]M ARR in [Y] months always forget to clarify that it's either a) not really Annual b) not really...
Halve KV‑Cache Memory by Sharing Attention Across Heads
Our AI group published a novel finding from pre-training research In short: you can cut KV-cache memory in half by sharing most of the attention structure across heads + keeping small per-head differences—without hurting model quality or speed I'll explain this as...
LLMs Dictate Tech Stacks; Every Business Needs a CLI
LLMs decide what languages and libraries we use. They’ll also decide what products a business runs on. They'll prefer ones that work with them. Every business now needs a CLI. fin(.)ai/cli https://t.co/ycezh9XYWd