You will get a custom AI agent or MCP server that connects AI to your systems safely
Project details
Connecting AI to your systems is powerful and, done wrong, dangerous. I build the version that's safe.
I'm an ex-Datadog SRE and I build MCP servers and AI agents that let Claude, ChatGPT or any client talk to your systems, read-only and human-in-the-loop by design. The AI reads your data and drafts answers or actions, but it never makes a state-changing move on its own. That safety boundary is the whole point: an agent that can quietly change production is a liability, not a feature.
Typical use: expose your observability, database or internal APIs as read-only tools so your team can ask questions in plain English and get reliable, auditable answers. Built in Python on the MCP SDK, with an allowlist and tests that prove no write tool exists.
I've published my work open source so you can judge the code before we even talk: github.com/antoniopablo/datadog-mcp-server and github.com/antoniopablo/incident-triage-agent. Tell me the first system you want to connect and what you'd ask it.
I'm an ex-Datadog SRE and I build MCP servers and AI agents that let Claude, ChatGPT or any client talk to your systems, read-only and human-in-the-loop by design. The AI reads your data and drafts answers or actions, but it never makes a state-changing move on its own. That safety boundary is the whole point: an agent that can quietly change production is a liability, not a feature.
Typical use: expose your observability, database or internal APIs as read-only tools so your team can ask questions in plain English and get reliable, auditable answers. Built in Python on the MCP SDK, with an allowlist and tests that prove no write tool exists.
I've published my work open source so you can judge the code before we even talk: github.com/antoniopablo/datadog-mcp-server and github.com/antoniopablo/incident-triage-agent. Tell me the first system you want to connect and what you'd ask it.
AI Algorithms
Large Language ModelAI Applications
AIOps, Anomaly Detection, Conversational AIAI Development Language
PythonAI Tools
Azure OpenAIAI Models
ChatGPT, GPT-4What's included
| Service Tiers |
Starter
$120
|
Standard
$340
|
Advanced
$790
|
|---|---|---|---|
| Delivery Time | 4 days | 8 days | 14 days |
Number of Revisions | 1 | 2 | 3 |
AI Model Integration | |||
Batch Normalization | - | - | - |
Database Integration | |||
Detailed Code Comments | - | - | - |
Image Upscaling | - | - | - |
MLOps | - | - | - |
Model Deployment | - | - | - |
Model Documentation | - | ||
Model Monitoring | - | - | |
Model Testing & Optimization | - | - | - |
Model Tuning | - | - | - |
Natural Language Processing | - | - | - |
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | - | - | - |
Setup File | - | - | - |
Source Code | - | - | - |
Frequently asked questions
About Antonio
ex-Datadog SRE | DevOps, Kubernetes, Terraform, AWS/Azure & AI Agents
Madrid, Spain - 7:33 pm local time
Most Datadog customers overpay 30–60% — high-cardinality metrics, unindexed log ingestion, forgotten custom metrics, over-provisioned hosts. I've seen the pricing model from the inside, and I know exactly where the money leaks.
What I do: • Datadog cost audits — typical result: 30–60% bill reduction without losing visibility • Migrations to/from Datadog (Grafana, New Relic, CloudWatch, OpenObserve) with zero-downtime overlap strategy • Full observability setup: dashboards, monitors, SLOs, alert tuning (kill alert fatigue) • SRE & platform engineering: Terraform, Kubernetes, AWS/Azure/GCP, CI/CD
I work async-friendly (CET timezone, overlap with US mornings), communicate proactively, and leave documentation your team can actually use.
Beyond observability, I build production AI: MCP servers that let Claude query real systems safely, and LLM agents with a strict human-in-the-loop boundary (they draft and recommend, never take a state-changing action on their own). Two are open-source on my GitHub — a Datadog Cost MCP Server and an Incident Triage Agent — so you can read the code before we talk. Python, SQL, MCP SDK, OpenAI/Claude APIs.
If your Datadog bill makes your CFO nervous, or you're planning a migration — send me a message with your current setup and I'll tell you in the first call whether I can help and roughly what's achievable.
Steps for completing your project
After purchasing the project, send requirements so Antonio can start the project.
Delivery time starts when Antonio receives requirements from you.
Antonio works on your project following the steps below.
Revisions may occur after the delivery date.
Scope the tools and data
We pick the first system and agree exactly what the agent can read and do.
Build the MCP or agent
I build the read-only MCP server or agent in Python, with an allowlist and tests.