Des Traynor's Recent LinkedIn Posts

It's been a *long* time since we've done one of these! So why raise now? A: The Customer Agent is coming. The future will simply *not* be a dozen different disparate agents talking across each other, pissing off your customers and giving internal teams orchestration, hand off, and integration headaches. That would be a nightmare for everyone. The future will be a single seamless Customer Agent. That's what we're building. Come join us!

Support teams have always found it incredibly hard to keep docs/help centres up to date with policy & product changes, but it's more important now than ever before. Docs are load-bearing infrastructure in the AI era. Hence one of my favourite features in Fin is Content Suggestions. Fin will recognise out of date docs and suggest a diff that you can easily approve/tweak, meaning the first 95% of the work is now done for you.

Cool to see Fin and Intercom featured in The New York Times. I’m not fully subscribed to the “All SaaS is Dead Immediately because OpenClaw” narrative but I am fully subscribed to the “AI Is The Only Future for Software Companies” (which, granted, nets out in the same place: drop everything and build AI)

When you're buying AI it's really important to verify that the product actually works well for lots of people at scale. Promises and marketing hype are all well and good, but the proof is very much in the pudding. So when we tell people Fin has > 7,000 happy customers, this is what we mean. It really works, at real scale, for real businesses, delivering real outcomes. If you'd like to join the list, just DM me, mail me, or contact the team.

🚨Attention startups🚨 We're re-launching our AI Startup pack today with 54 great companies offering you $500K in value This is every product a startup needs, free. Examples: Granola, Cloudflare, Miro, Replit, Superhuman, PostHog, Snowflake, Linear +many, many more. Details in the comments

More details on our 2× initiative here, if I could offer you generalized advice on AI Engineering and how to get the most from it, it's just 2 things 1. Make and Communicate a firm decision that you want a massive measured boost in productivity. (Ours was to 2× output in 1 year from June 2025) 2. Take it very, very seriously.

We're sharing some product updates in a fun new casual style these days, hear from Brian, Christine, and Dawn P. Here's the key feature callouts CX Score - AI tells you what your customers are *actually experiencing* (1:30) CX Reasons - AI will also tell you *why* Curated Topics: AI clusters your topics, but you can also curate them (5:57) Trending Topics: AI identifies new emerging topics from your customers (8:30) Fin over Slack: I don't need to explain this (11:22) Audience Control: Control who sees Fin, when, where, etc. (12:28) Fin Changelog: Stay up to date with everything that changes in your Fin setup (13:43) What started as a little AI Agent has grown into a fully featured and massively powerful Agent Platform.

A big failure pattern in the Web 2.0 era was making it extremely easy to do things that sadly no one wanted to do. A big failure pattern in the AI era is automating entire processes+workflows that actually no longer need to happen. It's akin seeing the full power of the internet in the 90s and thinking "shit this is gonna make it really easy to send faxes"

Tonight we launch our Berlin office properly with a fun Intercom and Fin event with a few hundred locals, some fun presentations, and great food and drinks. (Starting at 6pm tonight in w3.hub) I’m really excited to get to know the tech scene here in this incredible city, and I’m looking forward to visiting often.

I had a wide ranging chat with Eléonore about Intercom, Fin, and a variety of topics * Why did we need to drop everything and run to AI immediately, and do so regardless of the cost. * Building reliable AI through empirical processes is extremely different to building bread&butter SaaS. * Mistakes we (I) made along way * Lots more anecdotes along the way. Link in comments

"Businesses will tolerate any amount of chronic pain that will long term kill them just to avoid any amount of self inflicted acute pain that will immediately hurt them. That's why adapting to AI is hard" Wouter Teunissen asked some great questions starting with my failed PhD and catching all the way up to our plans for Fin this year (including Customer Agent and beyond).

Here Brian Scanlan shares a fun update on our 2x initiative. We are 9 months into it and Claude Code is authoring 90% of our PRs.

"It's only good at the simple stuff" That's the most common persistent misperception about Agentic Support. Today Fin powers things like * complex Fintech support involving account transactions, * messy B2B queries involving multiple back-end changes * refunds, changes to addresses, repeat orders * powerful multi-step conditional policies (e.g. VIP customers get unlimited refunds but only if ...) And today we're releasing a shitload more features to make it even more powerful. Watch it live here → https://lnkd.in/dszPaNPz

Fin will now literally manage your docs for you - it'll let you know what's wrong, out of date, duplicated, or missing. It'll also write the docs for you too. All you need to do is approve the changes and they're sorted. Here's how it works: 1. Customer asks about X 2. Fin doesn't know, so hands over to a CS agent 3. Fin reads their answer 4. Fin writes/updates an article about X. 5. Fin never hands that question over again. Check out Christine & Brian's video to see how easy this is.

I had a fantastic chat with Kyle Poyar on all things Marketing, Intercom, Fin, AI, and lots in between. All new topics for once!

Cool prototype and great accompanying blog post by Emmet here.

The 2026 Customer Service Trends Report is out. Every year we interview over 2,000 CS leaders to discuss what's happening in AI & Customer Service, and publish a full 50 page report of everything we've learned. 4 Key ideas for me: 1. it's still early - most orgs are not yet at mature deployment (e.g. fully AI-First, massive resolution rates, etc.) 2. those who are fully deployed are getting ahead of their competitors 3. chat is still the first place of deployment but Voice is growing very fast (this resonates with what we've seen with Fin Voice) 4. there is a lot of support whitespace (e.g. people deploying support in places where they couldn't previously afford to) Link to the report in the comments

If you work on B2B AI products, or are buying AI and want to avoid inheriting an unmaintainable and fast-degrading shitshow, read this piece by Fergal. It's excellent. Reminder: it's not PMF if it's a different prompt/product for every new customer, it's just ✨ sparkling consultancy✨

2026. You're probably thinking about your Customer Support volume & plans. Len (COO at Podia) adopted Fin and got the following outcomes 1. 64% resolution rate 2. much happier customers (see the screenshots) 3. much happier team It's truly win-win-win. This is what's available to your business, not months away, but today. → Fin .ai ← (or DM me if you have questions)

If you're running or working in a SaaS business and wondering how to pick a path forward, you should read this short piece by Eoghan about the pain, the progress, and the mandatory creative destruction necessary to survive. Also some fun new numbers to share about our business 📈

If you're in Paris, swing by Fin Labs and say Hi! Si vous êtes à Paris, passez chez Fin Labs pour nous dire bonjour. (Also: cool product launch tonight, link in comment)

Patrick McKenzie is one of the sharpest thinkers in the game, 6 years at Stripe where he remains a Strategic Advisor, and he is also a former Customer Support rep. We spoke about * customer support not as a cost center, but as a complex system shaped by organizational structure and incentives. * metrics and their unintended consequences: how turning useful measures like average handling time or NPS into targets warps behaviour, incentivising shorter calls or superficial satisfaction instead of actually helping people. * the human side of scaling support, from founder-led, bug-fixing missions to quotas and standardized workflows, why support teams systematically lose talent when there’s no real career ladder beyond managing tickets. * AI’s role today — not replacing support teams, but doing the “easy 60–70%” of tickets reliably and creating new career paths like “AI support specialist,” where context, automation, and deep tooling matter far more than typing speed. It was a wide ranging chat with one of my favourite speakers, who I first interviewed on our own blog ~12 years ago.

One thing most people don't realise about Fin: Fin fully resolves complex queries (refunds, amendments, upgrades etc) end to end, no humans needed. Example: it can discuss a refund with a customer, calculate if it is eligible according to the refund policy, and execute the refund via your payment platform, wholly owning the issue without it ever touching a CS agent. More than Fin will tell you what data connectors or integrations you should install to make it ever more powerful. Check out the video by Christine below

A lot of people think Fin is only necessary when you're "at scale" and until that point you should do your own support. I understand the thinking, but it misses out of what your customer's experience is. The reality is Fin lets you offer instant accurate replies in every language and timezone, and that's what customers actually want (for many query types). Here we see Jake from Siro, a young start (doing extremely well), using Fin for 250+ tickets per month, while leaving the knottier more open-ended issues to their team. Another advantage of starting AI early in your functions is that you scale things the right way first time.

Fin *fully* automates 81% of our own support volume. It's totally possible to hit these numbers, make 2026 your year for it. fin.ai