Claim your program
We already hold guidelines for many programs, so yours may already be listed. Claim it, confirm it, or take it down.
For MGAs, MGUs, program managers and carriers
Publish your appetite once, confirm the rules we draft from your guidelines, and fill your referral queue with risks you actually want. Program administrator software that learns from your own underwriters.
Early access. We are onboarding program owners now. Join the waitlist to claim yours.
What you get
MGA appetite management, without the email chase.
Everything a program needs to publish appetite, work referrals and learn from its own decisions.
Rules drafted from your guidelines
We draft carrier appetite rules from the underwriting guidelines you already have. You confirm each one. Nothing goes live until you do.
A referral queue with reason codes
Close calls land in an underwriting referral workflow sorted by how close they are. Quote, ask, or decline, and every decline carries a reason code.
A model that learns your soft appetite
Your written rules say what you will not write. Your underwriters' decisions say what you actually like. The model learns the difference from your own queue.
Rule suggestions
When your decisions keep landing the same way, we suggest a rule. You accept it, edit it or ignore it.
Distribution lanes
Run each program in the lane it actually uses: carrier-direct, MGA or MGU, so appetite and referrals route to whoever holds the pen. Built for E&S program distribution as well as admitted.
Verified productionComing
Production matched to statements and payments the parties reconcile themselves, so your numbers are confirmed, not self-reported.
Pooled learning is opt-in
By default your program learns only from your decisions. Opt in to share anonymized features and get the stronger network model back. Your decisions stay yours.
A human confirms every decline
A confirmed rule can decline and it names the rule. Every other decline of a referred risk is made by a person on your team.
Governance and bias testing
Every score is explainable and versioned. Models are tested for bias before and after each release, and you can see why any risk scored the way it did.
License
Licensed per program.
Program owners license the console per program, tiered by size, with a discount for opting in to pooled learning. Recording a decision is always free. Agents pay per call, so a busy queue never costs you gas.
Included with every program
- Program console and listing
- Referral queue with reason codes
- Guideline ingestion and rule suggestions
- Portfolio and appetite analytics
- Explainable, versioned scores
FAQ
Questions from program owners
My program may already be listed. How do I claim it?
We already hold guidelines for many programs. Join early access with your work email and we will verify you, then hand the program listing to you to confirm, edit or take down.
Do I have to write appetite rules from scratch?
No. We draft rules from the underwriting guidelines you already have. You review and confirm each one. Nothing goes live on your behalf until you confirm it.
What happens to a risk that is close to the line?
It lands in your referral queue, sorted by the closest calls. Your underwriters quote, ask a question, or decline, and every decline carries a reason code.
Will the model decline risks on my behalf?
A written rule you confirmed can decline, and it names the rule. Everything else is scored, and a person on your team confirms every decline of a referred risk.
Is my decision data shared with other programs?
Not unless you opt in. By default your program learns only from your own decisions. If you opt in to pooled learning, anonymized features are shared and you get the stronger network model back, plus a license discount.
Can carriers, MGAs and MGUs all use it?
Yes. Each program runs in a distribution lane: carrier-direct, MGA or MGU, so appetite and referrals route to the party that actually holds the pen.
How is the program console priced?
Program owners license the console per program, tiered by size, with a discount for opting in to pooled learning. Recording a decision is always free.
How do you guard against unfair outcomes?
Every score is explainable and versioned, rules are human-confirmed, a person owns every decline, and models are tested for bias before and after each release.
Early access
Claim your program.
We probably hold your guidelines already. Tell us who you are and we will hand you the listing to confirm.
We only use your email to invite you to early access.