Looking for n8n Expert to build AI Sales Agent with OpenAI & HubSpot!

Hi everyone, I love this community!

I am looking for a freelancer to help me optimize my current workflow. I have a basic setup running, but it fails randomly.

The Project:
I need to connect a Webhook from my landing page to HubSpot CRM. After the data is received, I want to use the OpenAI node to analyze the lead’s intent and draft a personalized email.

Requirements:

  • Strong experience with n8n self-hosted.
  • Familiarity with PostgreSQL for logging events.
  • Understanding of JSON data structures and Javascript in the Code Node.
  • Experience with error handling (try/catch blocks).

Current Issue:
My workflow works for simple text, but when the API response contains nested JSON arrays, the Split In Batches node breaks.

If you are interested, please send me a DM with your portfolio and hourly rate.

Thanks!

2 Likes

If you need I am available here:

Here is my portfolio:

I’m also a full-stack developer with the right Gen AI experience, which makes me a solid plus for your team[but right now only vibecoding]

Check my recent gen ai projects… I built a native Android automation agent too. It’s worth a look:

I can build complex AI automations directly in code, not just inside n8n.

I recently started posting my n8n work on YouTube with explanations:
https://youtube.com/@blankarray

Fun fact… I even made an n8n workflow to find a few n8n project leads for myself :skull::skull::skull: so I truly believe in what I do…


I started asking My recent clients for honest feedbacks so here is one testimonial: https://www.youtube.com/watch?v=TqBy3SVCHgQ&list=PLAJltY5bp6yiZ3sFBjm7bfrkLXSGtJX8m

Here is my linktree: iamvaar | Linktree

I am open for any remote opportunity such as part-time, full-time, project basis, being a co-founder for your automation agency, hourly basis and any opportunity…

My rate for whatever opportunity is 45$/hr

I am capable of developing custom n8n nodes too..
https://www.npmjs.com/package/n8n-nodes-pdfbro

https://www.npmjs.com/package/n8n-nodes-ocrbro

https://www.npmjs.com/package/n8n-nodes-ttsbro

1 Like

hey @uNavy
I’ve been working with n8n (self-hosted) for a while now and have hands-on experience with HubSpot integrations, PostgreSQL event logging, and JavaScript in Code nodes.

I have sent you my portfolio in dm.

lets connect!

Hi uNavy,

This is a very familiar issue — I’ve handled it multiple times in self-hosted n8n setups.

I’m a Certified n8n Developer (Level 1 & 2) and have built production AI sales and CRM workflows with HubSpot, OpenAI, PostgreSQL logging, and custom Code Node (JavaScript) handling.

Regarding your current problem:
The Split In Batches node breaking on nested JSON arrays is a known pitfall. This should be handled by normalizing the payload first (Code Node / Item Lists), explicitly mapping arrays into flat iterable items, or restructuring the OpenAI response before batching. I also usually add guards to prevent malformed API responses from killing the run.

I can help you:

  • Stabilize the workflow

  • Add proper try/catch + fallback paths

  • Implement Postgres logging

  • Make the OpenAI output deterministic and safe for downstream nodes

Happy to share relevant examples and my hourly rate via DM.

Let’s discuss.
— Muhammad

Moosa, how can we connect?

You can dm me @Parker_Jackson

You’re close — the random failures usually come from inconsistent payload shapes, not the OpenAI node itself. The safe fix is to normalize the nested arrays into a predictable list before Split In Batches, and to guard every “maybe-array” field with a fallback in a Code node.

  • Add a Code node right before Split In Batches to coerce the field into an array (or an empty array) and flatten nested arrays into a single list.

  • Wrap downstream nodes with explicit checks (length, undefined/null) and route “empty/invalid payload” to a logging + alert branch instead of letting the run crash.

  • Log every run to Postgres with a correlation id (webhook request id) + the normalized payload shape so you can replay failures deterministically.

We do this kind of self-hosted n8n hardening at Hashlogics, including HubSpot + OpenAI + Postgres logging patterns like the Automated ESG & Sustainability Engine and Agentic Automations on IRS Escape Plan implemented for QA.tax

My CV:star: Reviews

If you share one sample of the nested JSON (redacted), I can tell you exactly what to normalize and where Split In Batches is choking.

Hi! I’m an n8n specialist with 3 years building production workflows - this is exactly what I do.

I’ve solved the exact issue you’re describing (Split In Batches with nested JSON) multiple times. It usually requires proper JSON parsing before the split node and handling array depth correctly.

Strong experience with:

  • n8n self-hosted (30+ production workflows)
  • PostgreSQL event logging
  • OpenAI node integration and error handling
  • JavaScript Code Node for complex data transformations
  • HubSpot CRM integrations
  • Production-grade error handling with try/catch and retry logic

I can debug your current workflow and implement a stable solution with proper error handling and logging.

Sending you a DM with portfolio and rate.

Email: [email protected]
WhatsApp: +55 11 95174-9848
LinkedIn: linkedin.com/in/iuri-reis-124918181

Hi,

Sounds like a solid workflow you are building there.

I have extensive experience with n8n self-hosted, PostgreSQL event logging, and working with complex JSON structures in the Code Node.

Would be happy to take a look and get things running smoothly for you.

You can reach out to me on my email here

Colin

Hi @Muhammad_Bin_Zohaib , I’d be happy to help optimize and fix your workflow. I have extensive experience with n8n and have worked on projects involving Webhooks, HubSpot CRM, and OpenAI integrations. I understand the issue with the Split In Batches node and can certainly help resolve the error with nested JSON arrays.

Please feel free to share more details, and I’ll be able to provide a solution tailored to your workflow.

:telephone_receiver: Book a quick call:Calendly - Automaxion

Looking forward to hearing from you!

We can help you resolve this in 3 hours. You can book a meeting with us: Calendly

Experienced n8n & AI Architect based in Toronto (EST) :canada:

Hi @uNavy, I specialize in building Production-Ready AI Ecosystems using n8n, ranging from secure RAG pipelines to Multi-Agent strategic systems.

I’ve built several “Headless AI” frameworks that decouple business logic from the UI. You can view my live project gallery and technical portfolio here:

Portfolio & Case Studies

Looking forward to discussing how I can automate this for you! I can be reached at ([email protected])

Thanks,
Ken