Is Follow Up Ace safe?

"Is Follow Up Ace safe?" is the right question to ask before connecting any AI to your CRM — and it deserves a concrete answer, not a trust-us paragraph. Below is exactly how Ace handles the four things agents worry about most: whether it will message clients on its own, what happens to your contact data, whether it can produce something that breaks Fair Housing law, and whether the company behind it is legitimate. Every claim here is backed by the code that runs the product.

This is the plain-English safety overview. For the technical security deep-dive — contact-email and phone tokenization in the Ace chat, AES-256-GCM encryption, and code-enforced Fair Housing scanning — read how Ace protects your Follow Up Boss data.

Will Follow Up Ace message my clients without my approval?

Not by default. Out of the box, Follow Up Ace writes the draft and you are the one who clicks send. That is the single most important safety property of the product, and it is by design: Ace's job is to surface the next move and write the draft, and sending stays with the agent unless an admin deliberately turns something on.

The product's recommendation engine flips the usual AI contract. Instead of "the AI writes and sends a message," Ace surfaces a recommended action and pre-stages a draft text, task, or email that the agent reviews in the composer and sends manually (verified: chat-app/utils/whatsNextGenerator.js — the generator produces pre-staged action chips the agent clicks to send or draft, never an automatic send). The exceptions are things an admin deliberately turns on: Automations (follow-up rules the admin switches on one at a time), a graduated Ace Autopilot playbook (which starts as drafts for review and sends on its own only after the admin graduates it), and Home Almanac emails to a homeowner who has opted in. Automations and Autopilot are scanned for Fair Housing language before they send, and Autopilot sends and templated Automation emails carry the AI disclosure.

Why this matters for safety: an AI that auto-sends can embarrass you, message the wrong person, or send something tone-deaf during a sensitive moment in a transaction — and you'd find out after it happened. An AI that drafts and hands off keeps a human in the loop by default, and anything that sends on its own is something an admin chose to switch on. You get the speed of AI with the judgment of an agent.

What happens to my Follow Up Boss data when Ace uses AI?

When you connect Ace, it reads your Follow Up Boss data to do its work. The safety question is what form that data is in when it crosses into an AI model, and who can see it.

In the Ace chat, emails and phone numbers are tokenized before the model sees them

In the embedded Ace chat, Follow Up Ace tokenizes email addresses and phone numbers before the request reaches the language model (verified: chat-app/utils/piiAnonymizer.js — anonymizeString() replaces matched emails and phone numbers with placeholder tokens against a server-side replacement map). The mapping between a token and the real value stays server-side and is never sent to the model; the response is de-tokenized after it comes back. If the model provider logged the input, it would see tokens for those values — not your clients' real emails and phone numbers.

The scope is deliberate, and we state it plainly: names are not tokenized, and this applies to the embedded Ace chat only. Voice, automated drafting, typed classification and anything you ask Claude or ChatGPT through the connector receive the CRM text that task needs. Every vendor that processes CRM text is named on our Sub-processor page.

Data is encrypted with AES-256-GCM

Sensitive stored values are encrypted using AES-256-GCM (verified: chat-app/utils/encryption.js — crypto.createCipheriv('aes-256-gcm', ...)), an authenticated encryption standard that protects both the confidentiality and the integrity of the data. This is the same class of encryption used to protect data in regulated industries.

No public model is trained on your data

Your contacts, conversations, call notes, and pipeline data are not used to train the language models behind Ace. Our contracts with OpenAI, Anthropic, Google and ElevenLabs exclude your data from training their models, and the underlying API models (OpenAI's API) operate under API terms that do not use API inputs for training by default — a separate policy from the consumer ChatGPT product. TypeSafe, which classifies notes and replies, is listed on our Sub-processor page with the status of its data-processing terms (written terms are still being documented). Ace's own scoring models learn from de-identified outcome patterns pooled across accounts (stage history, response times, whether a lead closed) — never names, contact details or message content.

Can Ace's AI write something that violates Fair Housing law?

This is a real estate–specific safety risk that generic AI tools ignore. An AI that drafts client messages and property descriptions can accidentally produce steering language or reference a protected class — exactly the kind of wording that creates Fair Housing liability.

Follow Up Ace scans AI-generated content for Fair Housing problems in code (verified: chat-app/utils/complianceGuard.js). The scanner checks for references to protected classes (race, color, religion, national origin, sex, familial status, disability) and for steering language — phrases like "great for families" or "close to the church/mosque/synagogue" that create steering risk regardless of intent. The patterns are covered by regression tests, so a future code change can't silently weaken the guardrail without a test failing.

For the full picture of how Ace handles Fair Housing — including HUD's 2024 guidance on AI in housing and NAR's Articles 2 and 12 — see the Follow Up Ace compliance page.

Is Follow Up Ace security-certified?

We don't hold — and are not pursuing — a third-party security certification, and we won't imply otherwise. Our controls are designed in line with recognized security frameworks: encryption in transit and at rest, least-privilege access, HMAC-verified requests, audit logging, and API terms that never train on your data. We document what we actually do on the compliance page rather than borrowing a badge.

If a certification is a hard requirement for your brokerage's vendor approval, the honest answer is that we can walk your team through our real controls, but we can't hand you an attestation. We'd rather tell you that plainly than imply a certification we don't hold.

Is Follow Up Ace a legitimate Follow Up Boss integration?

Yes. Follow Up Ace is a certified integration in the official Follow Up Boss Marketplace (verified: chat-app/public/compliance/index.html — "Listed in Follow Up Boss's official integration directory"). FUB Marketplace certification means the integration meets Follow Up Boss's standards for API usage, security, and integration quality. It is an active listing that can be revoked — not a one-time badge — so it reflects ongoing standing, not a historical approval.

How do I cancel Follow Up Ace, and what happens to my data?

You can cancel Follow Up Ace yourself from your account — there is no retention call or cancellation gauntlet. Connecting and disconnecting are self-serve. Because Ace works inside Follow Up Boss through an authorized connection rather than copying your CRM into a separate product you have to migrate out of, removing Ace removes its access. For specifics on data handling and retention after cancellation, see how Ace protects your Follow Up Boss data and the privacy policy.

Follow Up Ace safety at a glance

Frequently asked questions

Does Follow Up Ace read all of my Follow Up Boss data?

Ace works inside Follow Up Boss through an authorized connection so it can read the contacts, conversations, and pipeline data it needs to make recommendations. It does not copy your CRM into a separate product. In the Ace chat, contact emails and phone numbers are tokenized before the model sees them, as described above.

Can Follow Up Ace send a text or email on its own?

Not by default. Ace produces drafts and pre-staged actions, and the agent reviews and sends them. Autonomous sending exists only where an admin turns it on: Automations, a graduated Ace Autopilot playbook, or Home Almanac emails after the homeowner opts in.

Does Follow Up Ace hold security or HIPAA certifications?

No — we don't hold or claim third-party security certifications; our controls follow recognized security frameworks and are documented openly. Ace is not a HIPAA product — real estate CRM data is not protected health information.

What makes Follow Up Ace safer than connecting ChatGPT to my CRM directly?

Pasting client data into consumer ChatGPT means real names and contact details leave your control with no tokenization, no Fair Housing scanning, and consumer-grade data terms. In the Ace chat it tokenizes emails and phone numbers first, scans drafts for Fair Housing risk, keeps a human on every send unless an admin turns on autonomous sending, and runs under API terms that don't train on your inputs. See Ace vs. ChatGPT for Follow Up Boss for the full comparison.

Go deeper on security and compliance

• How Ace protects your Follow Up Boss data — encryption, tokenization, and our certification posture (no third-party security certification held or pursued) in detail.
• AI and data privacy in real estate CRMs — the privacy landscape for agents.
• Follow Up Boss security features explained — what FUB itself provides.
• Follow Up Ace compliance — Fair Housing, HUD 2024 guidance, and NAR Articles 2 & 12.