How we use AI.

01What uses AI#

• Sherpa: an assistant that lets you and verified staff run the business by text or chat, like "add John Johnson as a new lead" or "draft an invoice for Susan for $10,000 for me to review".
• AI drafts: suggested replies, follow-ups, summaries, and estimate and invoice wording.
• AI phone answering and instant replies on managed plans.
• API and MCP connections: let your own AI tool (like Claude or ChatGPT) work with your Peak data. What that tool does is governed by its provider and your instructions.

This disclosure is part of our Terms of Service.

02How it works#

In Peak Systems, AI runs on your own AI key (Anthropic or OpenAI), so prompts and the data in them go to that provider under your contract with them. Where we supply a shared key, fair-use limits apply. AI acts as the staff member who asked, with that person's permissions.

Action rules sit between AI and your data. Each kind of action is set to run automatically, ask for confirmation, stay a draft, need approval before it's sent, or be blocked by text. Money and customer-facing sends never go out from a single unconfirmed text. Every AI action is recorded in the audit log with the instruction that started it. You can tighten the rules in Settings. These are safeguards, not guarantees.

03What can go wrong#

AI can misread an instruction, pick the wrong person or record, make up facts or prices, write something inappropriate or off-brand, leave out a required notice, repeat itself, or be tricked by text hidden in an incoming message ("prompt injection"). Phone numbers can be spoofed. Undo may not reverse everything. Costs can add up on your AI key.

04What's on you#

1. A person reviews before it's used. Read every draft before it goes to a customer, a signer or a payment page.
2. Set action rules carefully, and keep your verified staff phone list current.
3. Don't put passwords, card numbers, government ID numbers or health information into AI prompts.
4. Tell people when it's AI. Utah requires a business that uses generative AI to talk with consumers to disclose it when a consumer clearly asks, and requires up-front disclosure in some regulated professions for high-risk conversations (Utah Code Title 13, Chapter 77). Other states have their own rules, and AI voices on calls need consent under the FCC's 2024 ruling. Our default: keep an AI and recording notice in call greetings and first messages, and always answer truthfully when someone asks.
5. You're responsible for what you send, approve or act on, and for automated messages following the Texting & Messaging Policy.
6. Watch your AI provider usage and set spending limits in their console.

05No guarantees#

We don't guarantee AI output is accurate, complete, suitable or available, or that it produces any result. AI output isn't legal, tax, financial, medical, engineering or construction advice.

06Training and your data#

We don't use your customers' data to train AI models. Your AI provider's terms govern what it does with prompts sent on your key, so review them. See the Privacy Policy.

Fresh Lime Media LLC dba Peak Engine · 972 W 1050 S, American Fork, UT 84003 · hello@peak-engine.com · (801) 822-9439