Hi Limor! Builder here, not a buzzword guy
I run multi-agent systems on Claude API orchestrated through n8n (self-hosted) daily — recent builds include an automated lead-sourcing pipeline (scraping → AI qualification → personalized outreach → CRM), a multi-agent content pipeline with human-in-the-loop approval via Telegram, and production OCR automation for SMB clients. Airtable/CRM integrations are routine for me. Berlin-based, EU timezone, available now. DM’d you!
Hi Limor,
I’m interested in your n8n + AI Workflow Builder opportunity if you’re still expanding your freelance/remote automation capacity.
My strongest area is building reliable n8n workflows that combine AI/LLMs, APIs, structured data processing, validation, and business logic. I also bring strong experience in process improvement and translating operational requirements into practical automation.
Recent case study:
https://github.com/nassaj2011/n8n-ai-quote-reliability-case-study�
I also have experience with API/webhook integrations and Telegram/Bale bot workflows.
I’d be happy to start with a small paid workflow or test build.
Best,
Naser
Hi — this is closely aligned with the kind of systems I’m building.
My work combines n8n orchestration with AI/LLM integrations, REST APIs, webhooks, databases, and backend services for business automation. I’ve also built a production-style AI automation system with agent tool calling, human approval workflows, audit logging, and n8n-based orchestration.
The sourcing / CRM / internal-data automation use cases you mentioned are particularly interesting to me. Happy to DM you with examples and discuss what you’re currently trying to automate.
Hi Limor — interested in the freelance / flexible-hours path (not full-time exclusive).
I’m Yvinson Pierre (Mesa/Phoenix AZ, US). I’m an operator who combines client-facing diagnosis (full-time sales/service at Dansons/WCB via Safelite) with hands-on AI/ops experiments: process clarity, workflow design, documentation/SOPs, and no-code automation tests (Make / Zapier / HubSpot sandboxes, Personal OS). Im learning n8n and can contribute most on scoping a simple reliable flow, writing the handoff notes, and QA’ing whether it actually runs.
I will not overclaim production n8n client deployments. Best first fit: one small internal workflow (CRM sync, alert, Airtable update, or AI-assisted summary step) with clear acceptance criteria. Available around an active W-2 — strongest Thu/Fri/Sun.
Happy to share a short sandbox walkthrough and discuss a paid pilot slice.
— Yvinson Pierre
Hi Limor,
I’m interested in your n8n + AI Workflow Builder role. I saw that the original post is older, but the thread is still active, so I wanted to check whether you’re still considering remote freelancers.
My current practical project is CarService AI — an n8n-based workflow using OpenAI/LLM, Telegram Bot API, REST API/webhooks and Supabase/PostgreSQL for request intake, validation, routing, statuses and automated notifications.
My strongest areas are n8n workflow design, API integrations, business logic, debugging and turning manual processes into working automations.
Demo project is available on request. I can share the bot and workflow screenshots privately.
I’d be happy to start with one small paid workflow or scoped task so you can evaluate the quality of my work before expanding the cooperation.
Hi Limor — for $49, paid only after delivery, I’ll diagnose one real boundary in your current stack: for example source intake → Airtable, AI classification → CRM, or CRM update → downstream action.
Within two business days you receive a failure/recovery matrix, explicit timeout/retry/idempotency/reconciliation rules, a prioritized patch plan, and executable synthetic tests for the non-happy paths. If I miss the written scope, the charge is $0. The $49 is credited toward a $299 implementation of the audited boundary.
Sample: Sample Diagnostic — FlowPatch
Request: Sign in to GitHub · GitHub
AI assistant drafted this message for FlowPatch Reliability.
Hi Limor,
I’m interested in the freelance/flexible-hours side of this if you’re still adding builders.
The strongest overlap with my work is n8n + AI/API orchestration: multi-step workflows, webhooks, routing/state logic, CRM-style operations, validation, retries, and human handoff where needed. One of my main builds, FlowForge AVA, is a multi-agent lead/customer-operations system built around n8n and API integrations; I also built MailIQ, a multi-workflow AI email automation system.
I’d rather prove fit on one small paid internal-tool workflow than make broad claims. A good first slice could be one sourcing/CRM input → AI enrichment or classification → Airtable/CRM update → review/error path, with a clean handoff afterward.
Portfolio: https://oyekola-systems-portfolio.oyekolaololade698.chatgpt.site/
GitHub: https://github.com/oyekola-ololade
If the role is still active, send me the first workflow you want built and the systems it needs to connect. I can scope the smallest useful version in writing.
Oyekola Ololade
AI Systems & Automation Engineer
“Builders over buzzwords” - so, concretely, the three things that decide whether VC internal tooling survives past the demo:
Sourcing pipelines break by design, not by accident. Anything that scrapes or enriches company data will silently start returning nothing when a source changes shape. The fix is cheap: each run records how many rows it produced, and a run that returns zero where it normally returns forty raises an alert. Without that, your deal-sourcing flow looks healthy for a month while quietly feeding you nothing.
Airtable as a CRM needs one writer per field. Once n8n, a form, and a person can all write the same record, you get silent overwrites and duplicate companies that are genuinely painful to unpick later. One owner per field, dedupe on domain rather than name, and an idempotency key on every automated write.
Claude and ChatGPT should classify and draft, not decide. For sourcing that means the model extracts and scores into a fixed schema, and the thresholds that actually gate a company into your pipeline are rules you can read and change. Otherwise nobody can explain six months later why a company was filtered out.
Background: I build production LLM features with schema-validated output, deterministic checks and an eval set gating prompt changes, plus tool-calling agents and the n8n and backend plumbing underneath - TypeScript, Python, Postgres, Redis queues, AWS. Freelance rather than full-time, evenings and weekends IST with flexible overlap.
One question: is the sourcing data going into Airtable as the system of record, or is Airtable a view over something else? That decides whether this is a workflow problem or a data-model problem first.
Hi Limor — if this is still open, the sourcing/CRM/data side of your n8n + Claude/ChatGPT + Airtable stack is a good fit for my backend/API/data-reliability work.
I won’t claim client n8n case studies I can’t substantiate. A practical test would be one $250 fixed pilot: source or CRM trigger → AI-assisted enrichment/classification → Airtable/CRM review queue → validation/dedupe → source/error ledger, with retries and handoff notes.
That gives you a measurable workflow before any larger freelance arrangement. Public engineering proof: https://fored.id
If useful, send the first internal workflow you want improved plus the current systems and target Airtable/CRM shape, and I can scope the smallest useful version in writing.