ThreoAI Server Connection Error

Product: ThreoAI
Ticket Type: Common Error / User-Facing Issue

---

Symptom

The user sees an error message similar to one of the following while using ThreoAI:

• “Server Connection Error”
• “Failed to connect to the server”
• “Something went wrong. Please try again.”
• The message appears to send, but no response is returned, or the response takes an unusually long time before failing

---

Root Causes

This error has two distinct root causes that result in the same user-facing message:

Cause 1: LLM Provider Outage or Slow Response

The AI model being used (e.g., Claude, GPT-4o, Grok) is hosted by a third-party provider (Anthropic, OpenAI/Azure, xAI, etc.). If that provider’s service is experiencing an outage, high load, or degraded performance, ThreoAI cannot get a response and displays a connection error.

Signs this is the cause:

• The error affects multiple users or all users at the same time
• The error is limited to one specific model (other models may still work)
• Provider status pages show an active incident

Cause 2: Context Window Exceeded

Every LLM has a maximum context window — the total number of tokens (words/characters) it can process in a single conversation, including both the conversation history and the response. When a conversation grows too long, the model cannot process the next message and returns an error.

Signs this is the cause:

• The error only occurs in a specific, long-running conversation

• Earlier messages in the same conversation were answered successfully

• Starting a new conversation resolves the issue immediately

• The user has been pasting large documents or having a very extended back-and-forth

• The conversation ‘gas gauge’ is full:

  

---

Resolution Steps

For LLM Provider Outage

1. Check the provider’s status page:

   • Anthropic (Claude): status.anthropic.com
   • Azure OpenAI (GPT models): azure.status.microsoft.com
   • xAI (Grok): status.x.ai

2. Try a different model. In ThreoAI, use the model selector at the top of the chat interface to switch to a different available model. If the alternate model works, the issue is isolated to the original provider.

3. Wait and retry. Most provider outages are resolved within minutes to a few hours. Ask the user to try again after a short wait.

4. Report to your Synthreo MSP contact if the outage is prolonged or impacting business operations, so your MSP can monitor and communicate updates.

For Context Window Exceeded

1. Start a new conversation. This is the immediate fix. Click New Chat in ThreoAI to begin a fresh session. The error will not occur in a new, empty conversation.

2. Break up large tasks. Advise the user to avoid pasting very large documents (hundreds of pages) into a single message. Instead, break the content into smaller chunks across multiple messages or conversations.

3. Summarize and continue. If the user needs to maintain context from a long conversation, they can ask the model to summarize the key points, start a new conversation, and paste the summary as the opening context.

4. Switch to a higher-context model. Some models support larger context windows. For example, GPT-5.4 supports 128,000 tokens while some models support only 16,384. Your MSP can configure access to higher-context models via Access Rules in Tenant Management.

---

Prevention Tips

• For tasks involving large documents, recommend users upload the document in sections or use a model with a larger context window.
• Users who regularly hit context limits may benefit from switching their default model to one with a higher token limit. Contact your MSP to adjust model access.

---

Escalation

If neither cause above applies and the error persists across multiple models, new conversations, and multiple users, escalate to your Synthreo MSP administrator or open a support ticket with Synthreo.