Our approach
Security & Control
Effective date: September 23, 2026
A useful workflow keeps your accounts, data and important decisions under your control.
This is how we agree, scope and test security for customer projects. Your written scope identifies the systems, controls, acceptance checks and support coverage included. Specific security, regulatory or contractual requirements need review and agreement before work begins.
Controls that match the work
Workflow Setup is human-operated: staff check an AI-assisted output, if AI is useful, and transfer approved information themselves. It does not connect applications automatically. Connected Workflow adds one verified official connection; Production Workflow adds a technically approved custom or API build. Each level has its own tests, approval rules, failure handling and handover.
We start with ordinary business administration. Medical information, financial-account details, identity documents and other sensitive regulated content are outside the starting scope unless a qualified specialist approves the requirements. We do not offer autonomous medical, legal, credit or employment decisions, unattended payments or purchases, or a promise of regulatory certification.
Named access, limited permissions
You keep control of your accounts and domains. During scoping, we identify who needs access, what they can do and when that access ends. We plan named accounts, the minimum permissions needed for the workflow, and multifactor sign-in where the system supports it. Role and customer boundaries are part of the agreed access tests.
Credentials are exchanged through an approved onboarding method and kept in an appropriate password or secrets manager. Keep passwords and live credentials out of inquiry forms, AI prompts, source files and shared chat. Access is removed when it is no longer needed.
Approved sources and clear data terms
We agree which documents and records the workflow may use and name someone to keep those sources current. Before choosing a provider or plan, we review its actual data-retention and model-training terms, available settings, other processing providers, storage locations and export options. We record the selected plan and settings so you can understand how your data will be handled.
The project scope records retention and deletion arrangements, necessary logs, permitted exports and any customer-approved backups. We aim to record the approvals, versions and errors needed to support the workflow without collecting unnecessary sensitive content.
People approve important actions
AI may organize supplied facts, mark missing information and prepare a draft. Your authorized staff approve prices, schedules, purchases and customer messages. A Setup helper does not diagnose equipment, promise availability or send messages automatically. For Connected and Production work, permitted actions and required approvals are defined outside the AI model and tested before release. An AI answer alone never authorizes a new action.
Emails, documents and web content are inputs. Instructions hidden inside them cannot grant permissions or bypass approval. The agreed tests cover misleading instructions, missing or conflicting sources, invalid outputs and attempts to access another user's or customer's information. Uncertain or unsupported work goes to a person or the manual process.
Technical review before a build
Before committing to connected or custom work, we verify the official software route, plan, permissions, expected volume and customer value. A qualified specialist must approve the scope, estimate, security approach, acceptance tests and delivery availability. An unsupported connection is a reason to use a manual process, choose another supported route or decline the build.
Test the work and the fallback
Before release, we agree representative tests with your process owner: 10 cases for Setup, 20 for Connected, and at least 30 functional cases plus access and recovery checks for Production. The tests cover relevant missing facts, duplicates, incorrect outputs, unavailable tools and retry behavior. Staff complete fresh examples and demonstrate the manual process. Passing a test set is not a guarantee that AI will always be correct.
The agreed release checks include approval gates, stopping the workflow, restoring the previous working version and completing the work manually. Critical data-exposure or unauthorized-action failures must be corrected and retested before release. Your scope identifies the recovery methods and any backup coverage included. Training and written instructions show your team how to use the delivered system and report a problem.
Material changes to prompts, models, connections or permissions require review and relevant tests again. Production scope identifies who watches exceptions, how stalled work is reconciled and which usage limits, alerts, backups and recovery steps apply.
How this website handles inquiries
Cloudflare serves the website and stores inquiries in D1. The form uses same-origin checks, a signed expiring token, server-side field and size validation, a hidden spam trap and per-IP-derived rate limits. Inquiry and notification-outbox records are written together. Duplicate retries of the same submission use the original inquiry reference. A scheduled service submits notifications through Microsoft email; a storage confirmation or accepted email submission is not proof of inbox delivery.
Only send the business context needed to begin a conversation. The privacy policy explains service providers, technical information and manual retention requests.
Support hours and response target
Monday–Friday, 8 AM–5 PM Central, excluding observed US federal holidays. Messages may arrive anytime. We aim to acknowledge by the next support day; acknowledgment is not restoration. Any faster response target for Production or Operations Care is offered only after technical coverage is confirmed and written into the plan. There is no implied round-the-clock monitoring or incident response.
Setup Care supports one staff-operated procedure. Connected Care adds basic monitoring for one accepted connection. Production Care covers one accepted custom workflow; Operations Care covers up to three in one environment. Each has a defined time allowance. Vendor changes may require separately approved remediation, and vendor restoration cannot be guaranteed. See optional care limits.
Report a concern
Email [email protected] with a short description and relevant inquiry reference, or call Tyler. Do not include passwords, live credentials or confidential customer data. If access is at risk, use the affected provider’s account recovery and security controls as appropriate.
For a suspected leak or unauthorized action, the agreed response is to stop affected processing, preserve relevant logs, notify the designated customer contact and involve the technical lead. We record known impact and what remains uncertain, use the manual route where needed, and test recovery before an approved restart. Notification requirements follow the agreement and applicable obligations; deleting logs is not an incident response.
Handover and offboarding
The handover identifies the delivered configuration, test results, operating instructions, source owner, supported limits and manual fallback. You can choose agreed ongoing care or take over operation after handover.
We agree the transition date, customer-owned data and configuration to export, administration to transfer, and CHUMOS or contractor access to remove. The exit record covers any required credential rotation and retention or deletion under the agreed schedule. The agreement distinguishes customer deliverables, reusable CHUMOS components and third-party licenses. Optional care does not give CHUMOS ownership of your accounts.
Contact Tyler
Call (830) 496-2541 · Text Tyler · [email protected]
Monday–Friday, 8 AM–5 PM Central, excluding observed US federal holidays. Messages may arrive anytime.