Your enterprise systems, mounted to AWS AI.
Hardpoint connects Amazon Quick and Amazon Bedrock AgentCore to 250+ business systems through hardened, governed MCP connectors built with CData. It deploys into your AWS account and plugs into AWS’s AI services without custom connector code.
MOUNT SERVICENOW via CData TOOL incident.search [read] WRITES OFF until enabled AUDIT ON → CloudWatch + S3
Illustrative tool names. Each deployment exposes only the tools you allow.
noun. A reinforced mounting point on an aircraft’s wing, built to a standard interface so that any compatible payload can be attached, locked, powered and put to work without redesigning the aircraft.
AWS built the aircraft. Hardpoint is where your business attaches.
Amazon Quick and Bedrock AgentCore supply the models and the runtime. Your ERP, CRM, ticketing and accounting systems hold the data. Hardpoint is the hardened, standard mount that connects them.
AWS supplies the AI. The connectors are left to you.
Amazon Quick ships with a short list of built-in connectors and lets you add your own MCP servers. Bedrock AgentCore gives you a runtime and a gateway. On most projects, the connection to the systems the business actually runs on still has to be built.
One MCP server per system
Authentication, pagination, rate limits, schema mapping and error handling are written again for every source on every project.
Community servers aren’t enterprise-grade
Open-source MCP servers rarely come with least-privilege scopes, audit trails, secret rotation or anyone accountable for patching them.
The pilot doesn’t carry to production
Connectors improvised for an Amazon Quick pilot usually have to be rebuilt before the same workload can run on AgentCore.
Hardened, governed connectors, ready to deploy.
Hardpoint turns CData’s connector library, which covers more than 250 enterprise systems, into a hardened MCP layer for AWS, then deploys it into your account and supports it.
250+ enterprise sources
SaaS applications, databases, data warehouses and legacy systems, including QuickBooks Desktop and Enterprise and anything with an ODBC or JDBC interface, all behind one consistent interface.
Deployed in your AWS account
Runs inside your VPC, keeps credentials in AWS Secrets Manager, encrypts with your KMS keys and reaches Amazon Quick over private VPC connectivity.
Governed by default
Every tool starts read-only. You allow each tool per connector, scope access per user and log every agent-to-data call for audit.
One connector, pilot to production
The endpoint you register in Amazon Quick for a proof of concept is the same one you register behind Bedrock AgentCore Gateway in production. Nothing is rebuilt.
Guided setup
Pick your sources, authenticate, choose the tools agents may use, and deploy. There is no connector code to write or maintain.
Supported end to end
The Hardpoint team installs, integrates and supports each deployment, and scopes the first connector with you.
Four waypoints from subscription to a working agent.
Subscribe
Subscribe to Hardpoint through AWS Marketplace, so it bills against your existing AWS account and commitments.
Deploy
A CloudFormation stack provisions Hardpoint in your AWS account and VPC. Credentials go into Secrets Manager.
Connect
Authenticate each source system, then choose which tools each connector exposes and to whom.
Register
Add the Hardpoint endpoint to Amazon Quick as an MCP connector, or behind AgentCore Gateway for production agents.
Reference architecture. Deployment topology, including connectivity to on-premises sources, is confirmed for each customer during onboarding.
What “hardened” means here.
Before an agent can reach a system of record, security and audit have to be satisfied. Hardpoint is built around the controls they ask for, and runs the same checklist on every deployment.
- Private by defaultHardpoint runs in your VPC with no public endpoint required. Amazon Quick reaches it over VPC connectivity.
- Least privilege per toolTools are read-only until you enable writes, and each connector exposes only the tools you allow.
- Identity-scoped accessRequests carry the calling user’s identity, so agents see only what that user is permitted to see.
- Secrets stay in AWSSource credentials live in AWS Secrets Manager, encrypted with your KMS keys, and never sit in agent prompts or configuration files.
- Every call on the recordEvery agent-to-data request is logged with who asked, which tool ran and which source answered, and the log lands in CloudWatch and S3 in your account.
- Supported and patchedConnectors are maintained by CData, and each deployment is supported by Hardpoint.
The projects that stall without connectors.
ServiceNow in front of Amazon Quick
Give Amazon Quick agents governed access to incidents, requests, CMDB and knowledge articles, so IT teams can triage and search in plain language on AWS.
Legacy accounting, finally in reach
QuickBooks Desktop and Enterprise and on-premises ERP systems are hard for AI to reach. Hardpoint brings them in through supported drivers instead of one-off code.
Questions that span systems
“Which renewals this quarter have open P1 tickets?” An agent can answer by querying CRM, ERP and ticketing together through one endpoint.
Pilot to production without rework
Standardize proof-of-concept work on Hardpoint in Amazon Quick, then move the same connectors behind AgentCore when the workload goes live.
Common questions.
Is Hardpoint a CData product?
Hardpoint is built with CData connectors. CData supplies and maintains the connector library. Hardpoint packages it for AWS, hardens the deployment, is preparing its AWS Marketplace listing and supports each deployment.
Does it work with Amazon Q?
Amazon Q Business is now part of Amazon Quick. Hardpoint registers with Amazon Quick as an MCP connector and with Amazon Bedrock AgentCore Gateway as an MCP target.
Does my data leave my AWS account?
Hardpoint is designed to run in your own AWS account and VPC. Queries go from Hardpoint to your source systems and back to the AWS AI service that asked. The exact topology, including any on-premises connectivity, is documented for each deployment.
Can agents change data, or only read it?
Every tool is read-only until you turn on write access for that specific tool. Writes are logged the same way reads are.
How is it priced?
Hardpoint will be sold through AWS Marketplace, so it can bill against your AWS account and, where eligible, count toward your AWS commitments. Early-access terms are set with each design partner.
Why is it called Hardpoint?
A hardpoint is the reinforced, standardized mounting point on an aircraft wing where a payload is attached, secured and powered. Hardpoint is that mount between AWS’s AI services and your business systems.
Mount your first system.
Early access is open to organizations running, or about to run, Amazon Quick or Bedrock AgentCore. Tell us which systems your agents need to reach and we’ll scope the first connector with you.