Replies with context
Before generating a reply, the assistant synchronizes the conversation history it can retrieve from Bonzo. New stage assignments and campaign replies do not start as blank threads.
Built for mortgage teams on Bonzo
Every lead gets a thoughtful next step.
BrokerBot reads the full conversation, answers eligible inbound texts, follows up when a thread goes quiet, coordinates calls, and keeps Bonzo stages current. Your loan officers stay in control while the assistant handles the repetition.
Plans start at $50/month per company.Every loan officer is included, with usage and AI processing shown separately.See full pricing
No per-seat fees
Human handoff built in
Public and private sandboxes
Lead conversationHistory synced
M
Maria Alvarez
Previous Bonzo history imported
We talked a few months ago. I’m ready to look at the numbers again.
Absolutely. Last time you were mainly looking to lower the payment and keep some cash available. Is that still the goal?
Yes. Could Cole call tomorrow afternoon?
Yes. I can help line that up. Does 2:00 or 3:30 work better?
Reply preview →
Bonzo-nativeProspects, stages, notes, and text history stay connected.
Stage-aware behaviorProactive, reactive, or pipeline-only by pipeline and stage.
Per-officer assistantsSeparate voice, booking link, rules, and controls under the approved company model.
Deterministic send checksTiming, opt-outs, and outbound safety run before delivery.
Operational snapshot
Real activity. Measurable follow-through.
A current view of how quickly BrokerBot responds, moves conversations toward appointments, and keeps coverage working beyond office hours.
- 31minutes
- Median from the bot's first message to a booked appointment.
- 154seconds
- Median response time when a lead texts in.
- 189
- Appointments booked in the last 30 days.202 all-time
- 31.8%
- Of BrokerBot messages were sent outside standard 9–5 business hours.
- 22.5%
- Of worked leads replied.1,530 of 6,793
- 48,417
- Leads under management.
Snapshot supplied for BrokerBot in August 2026. Medians and counts reflect observed activity; individual team results will vary.
One continuous workflow
From first text to clean handoff.
BrokerBot is not a separate chatbot bolted beside your CRM. It works inside the lead history and pipeline structure your team already uses.
Works the pipeline
Set behavior at the company, loan-officer, pipeline, or stage level. A stage rule wins when the same pipeline needs different treatment across its lifecycle.
Coordinates appointments
Uses the officer’s scheduling settings, Calendly link, blackout dates, and availability guidance while recording the next action for the team.
Adapts without losing control
Give every officer their own tone and standing instructions. The managed rules remain locked beneath those additions, and a human text makes the bot step back.
1
ReadHistory, notes, stage, officer settings
2
DecideReply, follow up, update, or stay silent
3
GuardTiming, opt-out, duplication, instruction leak
4
ActSend through Bonzo and record the outcome
Fails closed
Opt-outs, internal prompt leaks, unsafe send conditions, and duplicate work are blocked before the outbound network request—not left to the model’s judgment.
Precise control
The right behavior for each stage.
A fresh internet lead and a funded borrower should not receive the same automation. Configure how the assistant behaves down to a single Bonzo stage.
- ✓Proactive: open, answer, and follow up with eligible leads.
- ✓Reactive: reply only after the lead texts first.
- ✓Pipeline-only: read and update stages without sending messages.
- ✓Stage settings override pipeline, officer, and company defaults.
Cole Williams · Pipeline modesStage → pipeline → LO → company
Bonzo stageModeWhy
New leadProactiveFast opening
RespondedReactiveLead-led
Follow-upProactiveKeep warm
ApplicationPipeline onlyHuman owned
Plans and AI models
Compare what your team values most.
Every plan includes the full product and every loan officer. The plan determines the AI model, fixed monthly company fee, and per-message charge.
$50/moEssential · GPT-5.6 Luna · + $0.01 per BrokerBot-sent message
$150/moPro · GPT-5.5 · + $0.02 per BrokerBot-sent message
$400/moMax · Claude Fable 5 · + $0.05 per BrokerBot-sent message
AI processing is additional and shown separately in estimates: Essential and Pro use OpenAI processing charged through BrokerBot, while Max uses your Anthropic key and Anthropic bills that processing directly. Monthly fees are per company, with no added per-seat platform fee. Benchmark claims come from BrokerBot’s July 2026 internal evaluation of 1,012 graded conversations; they are comparative test results, not a promise of future outcomes.
Where your leads’ data goes
American AI models. Every message, every plan.
Your borrowers hand you income, credit, equity and property details. Those conversations only ever reach AI companies based in the United States. There is no offshore model anywhere in the path, and no cheaper foreign model waiting behind a future cost-cutting switch.
US-basedEvery model BrokerBot runs on is built and operated by a US company — OpenAI and Anthropic. They are named in our privacy policy as the only AI providers that ever see a conversation.
Not training dataConversations are sent under the providers’ business API terms, which do not use your borrowers’ messages to train their public models.
Your key on MaxOn the Max plan the assistant uses your own Anthropic key, so processing runs under your company’s own agreement and is billed to you directly.
Ask any AI vendor two plain questions: which model answers my leads, and what country is that model’s company in? A surprising number of tools cannot answer either one in writing, because the cheapest way to run an AI assistant is to route it to whichever model is cheapest this quarter — including models operated outside the United States. BrokerBot answers both in writing: OpenAI and Anthropic, both US companies, both listed in the privacy policy. We do not route your conversations to Chinese AI providers or any other offshore model, and if our provider list ever changes, it changes in the privacy policy first.
K
Kai · BrokerBot
Simulated conversation
Browser sandbox · avoid sensitive informationSame response guardrails
Test before live
A testing ground that shows its work.
Use the public sandbox for a quick conversation. Signed-in teams get a private testing ground that can use their live brain, loan-officer identity, stage, pipeline, state, scheduling rules, and configured model.
- ✓The public test never texts a phone number or changes Bonzo.
- ✓The private test can simulate a new lead or a multi-turn thread.
- ✓Administrators can preview a specific AI model before requesting a tier change.