Help: AgentGPT keeps timing out on long-running data enrichment tasks

I'm currently evaluating AgentGPT for a client project that involves enriching a large dataset of B2B leads with firmographic and technographic data. The workflow is conceptually straightforward: feed a CSV of company names/domains into an AgentGPT agent, have it call a series of external APIs (Clearbit, Hunter.io, internal systems via REST), and output an enriched dataset.

However, I'm consistently encountering a critical operational flaw: the agent reliably times out on tasks that require more than 4-5 minutes of sequential API calls and data processing. This isn't a network latency issue on my end; it's the AgentGPT runtime itself terminating the session. The task will be progressing, I'll see log entries for several API calls, and then it simply stops with a generic timeout error.

My current agent configuration is below. The goal is sequential enrichment: validate domain, fetch company data, find key contacts, then stack technologies.

```json
{
"name": "DataEnrichmentAgent",
"goal": "Enrich lead data from provided CSV using external APIs.",
"tasks": [
"Load and parse the input CSV file.",
"For each record, validate the domain via Clearbit Domain API.",
"If valid, fetch full company data from Clearbit Company API.",
"Then, find email addresses for the domain using Hunter.io API.",
"Finally, check against internal tech stack database via POST to our endpoint.",
"Compile all enriched data and generate a new CSV output."
]
}
```

The problem manifests during the loop processing individual records. With a batch of just 50 leads, the agent times out around record 30. I've attempted to mitigate by:

* Breaking the initial CSV into smaller batches (10 records).
* Simplifying the task list to the absolute minimum steps.
* Ensuring all API calls are as concise as possible.

The batch splitting helps marginally, but it defeats the purpose of automation if I must manually manage and trigger 10 separate agent runs for 100 records. Furthermore, it introduces complexity in reassembling the final dataset.

My core questions for the community are:

1. **Architectural Workaround:** Has anyone successfully designed an AgentGPT workflow that delegates the *actual* long-running, iterative work to an external, persistent service? My thought is to have the agent's primary task be to invoke a single webhook to an external service (like a Flask app or a dedicated Zapier/Make/Workato scenario) that handles the batch processing, and then have that service call back with the result. Is this supported, or does the timeout clock run even while an external webhook is processing?

2. **Platform Limitations:** Is this a known, hard limitation of the AgentGPT platform (community or pro tiers)? Is the timeout configurable, or are long-running processes simply outside its intended scope? The documentation is unclear on maximum runtime.

3. **Pattern Alternatives:** If this is a platform constraint, what alternative agentic workflow platforms have you used for reliable, batch-oriented data processing tasks? I require something that can handle at least 20-30 minutes of runtime for complex data pipelines.

The lack of clarity on operational limits is a significant hurdle for adopting this tool for real-world data integration tasks. Any insights or shared experiences would be invaluable.

- Mike