Activity
Mon
Wed
Fri
Sun
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
What is this?
Less
More

Memberships

AI Automation Society

417.3k members โ€ข Free

AI Automation Agency Hub

327.7k members โ€ข Free

AI With Anushka

1.6k members โ€ข $23/month

AI Automation Skool

2.4k members โ€ข Free

AI Automation Club

751 members โ€ข Free

The AI Advantage

126.2k members โ€ข Free

AI Automation (A-Z)

162.2k members โ€ข Free

Chase AI Community

69.9k members โ€ข Free

1 contribution to The AI Advantage
I just built this production ready 'Support Gmail Agent' with the help of Claude code
What started as a demo is now almost production ready, let me walk you through it: I built this for a Imaginary Tech company called 'Gadgets & More Electronics' (Yeah I know name is not sounding like name) First of all the workflows, There are actually Two workflows: 1.The first one is for document ingestion. Where I have converted the FAQs & The policy pdf into chunks & uploaded them into Pinecone vector database. 2.The second one is our Support Gmail agent. Now let's talk about it what it does:- ~It checks for new emails every 15 minutes, takes only unread emails ,newer than 10 minutes. ~Then it detects whether the emails are automated or not, if it is Automated then workflow will stop but if it is not automated it will continue ahead. ~Then it extracts the customer's name from the emails to make it personalized to them. ~Then it will classify the query or emails whether it is Support relevant or not if not the workflow will stop right there but if it is then the workflow will proceed ahead. ~Then comes our support Agent which has our Pinecone vector database as a tool where our policy and FAQs are stored, it knows how to handle each topic, Personalization ,Tone and brand voice ,Output formats, rules, Confidence scores based on the retrieval of data. ~Then there is Agent response checker which will see whether The agent actually found the policy or not? because sometimes you know how Agent make up and hallucinate. ~If the policy found it will create a draft so that a human can review and send it and if it does not found the policy it will create a fallback draft. This is the entire workflow, will drop demo videos today. let me know whether we can add some new features,or add some more guardrails......Just let me know i am welcoming all thoughts and perspective. Harsh Singh
I just built this production ready 'Support Gmail Agent' with the help of Claude code
1 like โ€ข 2h
@Art Stevens No not now , i have just built it and tested it myself with 15-20 different type of queries ranging moderate to complex but if anybody wants to use and test it and give some feedback i will be more than happy
1 like โ€ข 39m
@Art Stevens thanks
1-1 of 1
Harsh Singh
2
15points to level up
@harsh-singh-4383
I help Business Implement Ai in the right way & Saving their Time, Cut Their Cost , and Grow

Active 32m ago
Joined Feb 13, 2026
Powered by