1. Definition
AI Insights is CANDID’s intelligence layer for referral partner relationships, designed to give LOs deeper visibility into the real estate professionals in their network. It aggregates public data, behavioral signals, and production analytics to create a comprehensive profile of each Realtor, helping loan officers understand who they should be building relationships with and how to communicate with them effectively.
AI Insights enriches each contact with contextual intelligence including professional profiles, personality analysis, production performance, and compliance signals. By combining this data with CANDID’s automation ecosystem, loan officers gain a clearer picture of referral opportunities, stronger conversation starters, and more strategic ways to nurture partnerships. 🤝
The goal is to transform referral partner management from guesswork into data-driven relationship building. Instead of relying solely on memory or scattered notes, loan officers can instantly access structured insights that guide outreach, improve conversations, and strengthen long-term partnerships. 🧠📊
In addition to guiding relationship strategy, AI Insights also enhances the Marketing Hub experience ☁️ by using enriched contact data to deliver more relevant messaging, personalization opportunities, and targeted outreach.
2. How It Works
AI Insights runs automatically when a Realtor Account is created or updated in CANDID, including Accounts brought in through the Support Portal Data Import tab with Enrich turned on. It researches the agent across public sources, verifies what it finds, and writes the results back to the Account. It only fills in fields that are blank, so nothing an LO entered by hand is ever overwritten.
What AI Insights adds to each Realtor Account:
- 🧑💼 Headshot: Searches the web for a professional headshot, scores every candidate on name match, market, source quality, and photo shape, and rejects property photos, logos, and photos of other people. The winning image is copied to CANDID's own storage so it never breaks if the original site changes.
- 🔗 LinkedIn Profile: Finds the agent's LinkedIn profile and only saves it when the profile also matches the agent's state, city, or brokerage email domain. A name match alone is not enough. If the first search comes up empty, a second wider search runs before giving up.
- 🧠 DISC Personality Type: Researches the agent's public bios, social content, and reviews to infer a primary DISC type (D, I, S, or C). The result is quality-checked for confidence and source quality before it is saved, and the Marketing Hub uses it for communication coaching tips.
- 📧 Missing Email and Phone: If the Account arrives without an email or phone, AI Insights looks for a published business email or direct line tied to that exact person (brokerage profile, agent website, Zillow, Realtor.com, license lookup). It never guesses an email from a pattern and only saves values with high confidence.
- 🏢 Brokerage and Parent Account: If the Parent Account is blank, AI Insights finds the agent's current brokerage. It then looks for an existing Account with the Brokerage record type and the same name (on the same LO first). If none exists, it creates one with the LO set, and links the Realtor to it as a child through the Parent Account field. A Realtor already linked to a team under that brokerage is left alone.
- 📊 Production Summary (Last 12 Months): Pulls the agent's recent production to show units closed 🏡, total volume 💰, and average volume per transaction, so loan officers can quickly spot top producers and rising agents. It also captures the agent's buyer vs. listing mix (the share of their deals where they represented the buyer), saved only when the source data is clear enough to trust.
Built-in safeguards:
- Company names are skipped. If the Account name is a company (for example "Smith Realty" or "Parthenon Title") instead of a person, AI Insights sets the Account Type and stops, since person matching would return the wrong people.
- No duplicate runs. Results are cached, so the same record is not researched twice in a short window, and each org is rate-limited to prevent runaway API usage.
- Blank fields only. Every write checks the field first. Existing headshots, LinkedIn URLs, personality types, contact details, and Parent Accounts are left as-is.
Additional intelligence shown on the Account: