Technical support for XSparks deployments.
XSparks provides technical support for active deployments through a dedicated engineering channel. For technical issues during production: [email protected]. The frequently asked questions below cover six categories of common requests.
Frequently asked questions, organized in six categories.
Getting started
How do I set up my first AI agent?
Most engagements start with an Educate workshop or a Solution Accelerator Assessment. We map your operational pattern, identify the highest-impact use case, and scope the first deployment. From there, the Accelerator Fast Path takes you from assessment to managed operation in 4–6 weeks.
What infrastructure do I need?
AIOM runs on infrastructure you already pay for. We compose customer platforms, ecosystem AI, and our own technology into a working system. No proprietary stack required.
How long does initial deployment take?
Standard deployments run 4–6 weeks under the Accelerator Fast Path: assessment (1–2 weeks) → customize + deploy (3–4 weeks) → managed ops.
Products & services
What AI services does XSparks offer?
See the full catalog at services.html. Briefly: Think (Educate, Strategic Opportunities Mapping, Data Readiness Assessment, Solution Accelerator Assessment), Build (Pilot, Beta, Custom Agent Engineering, Data Transformation, MCP Server Development, Custom Model Training, HITL, Agentic Orchestration, Governance, Legacy AI), Operate (Managed Services, Center of Excellence, Scale).
Can I customize agents for my specific use case?
Yes. Solution Accelerators (Technician AI Twin, CallReady, AI Configuration Agent) are productized and customized to your operation. When the accelerator pattern doesn't fit, we run Custom Agent Engineering against your operational reality.
Do you offer implementation support?
Yes. Forward-Deployed Engineers staff into your team for the build and stay through Operate. The same team that builds the system runs it.
Data management
Where is my data stored?
On infrastructure you already control. AIOM is deployed into your stack; your data does not move to a proprietary XSparks platform.
How do I export or delete my data?
Your team owns 100% of what we build. Export and deletion follow your existing data governance procedures. Contact your engagement lead or [email protected] for specifics.
What are your data retention policies?
We retain operational data only as long as necessary to provide service or as your contract specifies. See Privacy Policy for details.
AI & training
How do custom models get trained on my data?
Layer 5 of the AIOM, the Intelligence Layer, is where models meet enterprise context. We fine-tune foundation models on your proprietary data and deploy in your environment. Continuous learning loops are part of the build.
Can agents learn from feedback?
Yes. Human-in-the-loop checkpoints feed back into the training cycle. Agents get smarter every interaction; failure modes become predictable; automation expands as confidence grows.
What's included in model updates?
Managed Services covers ongoing model tuning, performance monitoring, cost optimization, and rollouts of newer foundation models via the Multi-Model Orchestration Engine.
Troubleshooting
Why is my agent returning unexpected results?
Most often, the issue is at AIOM Layers 1–2 (data) or Layer 4 (workflow), not Layer 5 (the model). Check the observability dashboard first; escalate via [email protected] if the issue persists.
How do I monitor agent performance?
Layer 7 of the AIOM, Reporting & Tools, surfaces dashboards for accuracy, throughput, escalation rates, and cost per task. Your engagement comes with monitoring built in from day one.
What should I do if an agent fails?
Failure modes are designed in. The HITL spine catches the agent before it acts on a low-confidence decision, escalates to a human, and logs the failure for retraining. If a failure occurs outside that envelope, contact [email protected] immediately.
Feature requests & feedback
How do I request a new feature?
Email [email protected] or surface the request through your engagement lead. Feature requests feed into the AIOM roadmap.
How can I provide feedback on the service?
Quarterly business reviews are part of every Operate engagement. For ad-hoc feedback, email [email protected].
Where can I see the product roadmap?
The AIOM roadmap is shared with active engagement clients quarterly. Public roadmap items (Agent Builder, future Solution Accelerators) appear on /aiom.
Still stuck. Email us.
Technical issues: [email protected]. Sales / engagement: [email protected]. Privacy / legal: [email protected].