Read how the Notch team brings engineering-grade discipline to no-code AI ops - so teams can iterate fast with near-zero mistakes in production. 🦾
Notch@notchCX
Shipping AI agents is easy. Shipping updates safely is the hard part 👉
English
Elool Jacoby
134 posts
Elool Jacoby รีทวีตแล้ว
Proud of the work we’re doing at Notch to turn operational know-how into AI driven performance.
Read the full article: forbes.com/sites/keithfer…
English
@HarryStebbings In 2026, AI support will split.
Most of the well-funded players are chasing “AI agents” that still depend on human scaffolding and FDEs- brittle systems that OpenAI could wipe out in one release.
@notchCX is building a real foundation to replace them entirely.
English
One market I cannot figure out in a world of AI is customer support. You have:
Sierra
Decagon
GigaML
Parloa
Forethought
Kore
Cresta
All of these have raised over $100M.
and then oldies like
Intercom
Zendesk
Help me out, how does this market shake out?
English
Excited to be part of @TechCrunch's 60 AI Disrupters list!
Notch@notchCX
We’re proud to share that Notch has been named one of the AI Disruptors 60 by Greenfield Partners and @TechCrunch. This list celebrates companies pushing AI from hype to hard results - and that’s exactly what we’ve set out to do since day one.
English
@justinmrobbins Sometimes the real innovation is understanding the business…
It’s surprising how often simple issues, like a duplicate charge, slip through automation. 🤷🏻♂️
At @notchCX we’ve found that a few well-placed deterministic rules can make all the difference.
English
Tipped my Uber driver. The app charged me 2x.
Support? A chatbot loop that apologized, redirected, & then abruptly ended the chat.
No next step. No human. Just…closed.
Automation should enable recovery, not block it.
#CX #CustomerExperience #AI #ServiceDesign
English
We're just warming up 🦾
Lightspeed@lightspeedvp
Customer support teams have had to choose between using AI and great CX. Until @NotchCx came along, which is why we are leading their $15M seed round. Their platform resolves up to 87% of tickets, cuts enterprise costs by 50%, and has already processed 10M+ tickets for 30+ clients. The best part? Their customers pay only for resolved cases. Congrats to Co-Founders @RafaelBroshi, @EloolJacoby, Yuval Peled, and the team building the autonomous enterprise. Cc: @TalMorgenstern
English
@aakashgupta Thank you @aakashgupta
Don't forget to include @notchCX in the customer service section 🦾
English
A 21 year old marketer is getting me 10M+ social impressions a year. Using a simple 3 step AI process:
He:
1/ Compiles daily transcripts from my meetings
2/ Fine-tuned an AI model to convert them to content
3/ Tracks each metric obsessively
I asked him to create a 5 min Loom so all of my portfolio companies can copy this process.
Want the loom? Comment “want” to get it.
English
Sources:
NVIDIA’s “Small Language Models are the Future of Agentic AI”
arxiv.org/pdf/2506.02153 arxiv.org/abs/2506.02153
English
Bottom line
Using heavy LLMs for every agent action is like buying a supercomputer for high-school math. Pick the right model size for the job and you can slash costs while maintaining outcomes - then reserve premium LLMs for where they actually move the needle.
English
SLMs are the future of AI agents - and here’s how to put that into practice 👉
English