Deployment Process

Cloud Deployment

ClimaMind qualifies most projects for cloud deployment first and keeps approved BMS writes under explicit review with the required safeguards, approvals, and operating evidence in place.

Parties & Responsibilities

Who participates in deployment

Clear responsibility split is required because cloud services, approved BMS integration, site operations, and M&V evidence move at different speeds.

Customer

The building owner, facility operator, energy team, or asset manager responsible for the site and business approval.

  • Define business goals, operating constraints, and acceptance criteria.
  • Authorize access to approved data, BMS stakeholders, and site context.
  • Approve deployment boundaries, write permissions, fallback expectations, and final acceptance.

ClimaMind

The AI HVAC optimization provider responsible for data review, modeling, control strategy design, deployment support, and performance analysis.

  • Assess data quality, train and validate models, and identify feasible control strategies.
  • Prepare point mapping, control recommendations, risk registers, and operating evidence.
  • Support deployment, monitor performance, and produce customer-readable M&V and operating reports.

System Integrator

The BMS or controls contractor responsible for site-side integration and controls implementation support.

  • Confirm point names, histories, trend configuration, BMS topology, and equipment mapping.
  • Configure approved connectivity, read access, write paths, alarms, overrides, and fallback behavior.
  • Support commissioning and confirm that BMS-side logic matches the approved control boundary.

Optional M&V Partner

An independent measurement and verification partner used when the customer wants third-party review or contractual savings evidence.

  • Review baseline method, measurement boundary, adjustments, and evidence requirements.
  • Check savings calculations against the agreed M&V plan and available data.
  • Provide independent review input for acceptance when the commercial arrangement requires it.

Deployment Process

A clear handoff at every step

From initial qualification through ongoing optimization, the customer provides defined site inputs and approvals, and ClimaMind returns defined engineering deliverables. The project advances only when both sides agree that the current step is complete.

Customer includes the site owner or operator and an authorized SI/BMS contractor where applicable.

01

Initial Qualification

Customer → ClimaMind

  • Provide current plant diagram with major equipment and piping
  • Provide representative BMS screens showing points and controls

ClimaMind → Customer

  • Review plant topology, telemetry quality, and deployment-readiness constraints
  • Check visible points, trends, control access, and security dependencies
  • Identify whether the approved deployment path is ready to proceed

Joint decision

  • Proceed to connection and technical scoping
  • Request specific missing information
  • Pause until site conditions are suitable

02

Connection & Scope

Review data requirements

Customer → ClimaMind

  • Define business goals and target operating outcomes
  • Provide current equipment, sequence, and controls documentation
  • Provide available trend exports and meter history
  • Name the customer project lead and SI contact

ClimaMind → Customer

  • Define the systems, data path, and outcomes in scope
  • Specify required data, points, sampling, and approval checkpoints
  • Document gaps, owners, next actions, and approved write boundaries
  • Propose the access, fallback, and M&V approach

Joint decision

  • Approve the technical scope and boundaries
  • Confirm inputs are sufficient to begin modeling

03

Model Readiness

Customer → ClimaMind

  • Provide validated historical data for agreed signals
  • Answer outstanding site questions
  • Confirm schedules, limits, and operating constraints

ClimaMind → Customer

  • Build a model calibrated to site behavior
  • Validate model performance against held-out data
  • Evaluate feasible control strategies offline
  • Define writable points, limits, and exclusions

Joint decision

  • Confirm the model is ready for commissioning
  • Approve the proposed control scope and boundaries

04

Commissioning & Advisory

Customer → ClimaMind

  • Provide approved connectivity, operator contacts, and BMS stakeholders
  • Coordinate the review window and SI availability
  • Confirm the approved read/write point list
  • Document fallback and manual override paths

ClimaMind → Customer

  • Run commissioning and advisory review against the approved boundary
  • Map approved points to equipment and functions before authority expands
  • Verify data quality, alarms, limits, and supervised write prerequisites
  • Run in shadow mode before any active control path is approved

Joint decision

  • Complete the commissioning readiness checklist
  • Confirm whether the approved deployment path is sufficient for the scope
  • Proceed directly to acceptance and ongoing optimization when the approved path is sufficient

05

Site deployment

Learn about local deployment

Customer → ClimaMind

  • Provide approved network and BMS access
  • Coordinate the site window and SI availability
  • Confirm the approved write list
  • Confirm customer authorization for supervised BMS writes

ClimaMind → Customer

  • Verify the approved BMS boundary for supervised writes
  • Verify supervised write paths, fallback behavior, and operator-visible status
  • Test alarms, limits, safety responses, and manual-override handling
  • Document field evidence for acceptance and ongoing optimization

Joint decision

  • Complete the commissioning readiness checklist
  • Require customer authorization for controlled BMS writes

06

Acceptance & ongoing optimization

Review acceptance criteria

Customer → ClimaMind

  • Provide operator feedback from live operation
  • Log issues, overrides, and unexpected behavior
  • Share agreed BMS and meter evidence

ClimaMind → Customer

  • Summarize availability and issued control actions
  • Review comfort, equipment, and operating constraints
  • Calculate results using the agreed M&V method
  • Track remaining issues, owners, due dates, and optimization follow-ups

Joint decision

  • Sign acceptance against agreed criteria
  • Agree the review and service cadence for ongoing optimization
  • Require new approval before expanding scope

Network Architecture & Data Flow

Where data lives and who can reach it

ClimaMind Optimize runs either in the cloud or on a client at the customer site. In both modes the customer BMS initiates requests and keeps control authority, each customer is isolated, and data leaves the site only with customer approval.

Network and trust boundaries

Customer site / control boundary

BMS

  • Initiates requests; reads results and executes control
  • Falls back on failure or stale results

On-site client and dashboard

  • BACnet/IP within the controlled site LAN
  • Local disk, local training, offline login
  • Uploads require customer approval

Customer browser

Console login

  • Business functions only
  • No container or bucket admin access
  • On-site uploads appear in project Files

AWS / ClimaMind management boundary

Shared EC2: client runtime

  • A separate container per customer
  • Per-customer identity and S3 access
  • Cloud BMS entry over HTTPS / mTLS

S3: one bucket per customer

  • Cross-customer access is denied
  • Operating data, results, and project files
  • Versions and retention per contract

Console / API services

  • Console → api.climamind.ai
  • Password + MFA; tenant authorization
  • Dashboard reads the customer's own S3 data

Private database

  • Accounts, projects, access, file metadata
  • No operating data or optimization results

Training and model management

  • Trains only on authorized customer data
  • Exact model versions are approved before release
  • Customer models are exclusive to that customer

Internal operations

AWS Client VPN + corporate identity + MFA

  • Customers never use this path
  • Resource access is granted by job role
  • A VPN connection grants no blanket access
  • Admin access is logged and denial-tested

Data flow by deployment mode

Each mode keeps the BMS in charge of control and keeps every customer's data in that customer's own bucket.

  1. 01 Cloud: the BMS initiates requests and executes control

    Customer BMS

    • Sends inputs, retrieves results
    • Keeps fallback and control

    Customer client container

    • Shared EC2, separate containers
    • Computes optimization results

    Customer S3 bucket

    • Operating data + results
    • Runtime reaches its own customer only
  2. 02 Console: authentication, authorization, dashboard, and files

    Customer / Console

    • Email + password + MFA
    • View data, start/stop, manage files

    api.climamind.ai

    • Customer and project authorization
    • Database holds identities and metadata

    Customer S3 bucket

    • Feeds the cloud dashboard
    • File upload, download, versions
  3. 03 On-site: data stays local; uploads need customer approval

    On-site BMS

    • BACnet/IP requests and result reads
    • Client responds; BMS controls

    On-site client + local disk

    • Local dashboard: view, start, stop
    • Offline login; uploads by consent only

    Customer S3 bucket

    • Approved uploads: raw data + results
    • Visible in project file management
  4. 04 Training and model delivery

    • Cloud: authorized S3 data → training → versioned models and evaluations → ClimaMind approval → staff deployment to the designated client.
    • Without upload approval, data, training, approval records, and model delivery stay on site; remote approval requires customer authorization.
    • Customer models are exclusive to that customer; general training needs separate contract terms. Approval does not trigger deployment, and the client never switches models on its own.
Security and compliance details

Additional Concerns

Topics handled alongside the main stages

These topics are not a separate deployment phase. They are checked at the stage where they affect risk, approval, acceptance, or the decision to proceed.

Deployment Packet

Download the deployment technical overview

Use this PDF for customer, facilities, IT/OT, and SI review. It summarizes the deployment path, responsibilities, security boundary, fallback behavior, acceptance, and long-term support.

PDF · Technical overview · Prepared by ClimaMind

Download PDF

ClimaMind AI Advisor

Ask about ClimaMind, HVAC controls, or how optimization works.