Woodgrove Life Insurance • Powered by Regisseur
From first inquiry to policy in force — in days, not weeks.
Follow Priya Anand from her first conversation with Woodgrove's AI advisor through lab results, an identity-verified medical review, automated underwriting, one human sign-off, and a confirmed policy — all transparent, every step on record.
20
Automated pipeline steps
1
Human approval — an underwriter signs off before issuance, by design
10
Real emails sent automatically
3
Secure document uploads via portal
32
Screens captured from a real run
Priya Anand
36 years old, UX design lead, non-smoker. Applying for $750,000 of 20-year term life to protect her daughter Zara and husband Deepak.
ApplicantWoodgrove's automated pipeline
Intake verification, parallel orders (lab / background / MVR), lab analysis, APS intake, risk classification, policy generation — all without manual routing.
Automated reviewDr. Marcus Feldman
Priya's attending physician. Receives a secure, identity-verified portal link and uploads the APS directly — no paper, no email attachment.
Attending physicianPriya's journey — one application, end to end
Priya applies
AI advisor + form
→
Automated review
Intake, orders, analysis
→
Lab + verified APS
Secure, identity-checked uploads
→
Preferred Plus
Risk classification
→
Policy issued
Real policy document
→
In force
Family protected
About this walkthrough
Every screen below is a real screenshot from an automated, end-to-end run of Regisseur, captured live against a running system — not a mock-up or staged demonstration. Priya Anand is a fictional applicant; all data is illustrative. The case ran on Woodgrove Life's demo workspace, using the same All-Agent Autonomous underwriting pipeline that a real applicant submission would use. The AI advisor chat responses are from a real language model call. The lab results and physician statement were submitted via the real, identity-verified secure-portal upload flow, the outbound emails were sent through the platform's real mail delivery path, and the policy document was generated by the system's real document engine. The email images in "The Communications" show the verbatim content of messages as actually sent and received — delivery was routed to a monitored test inbox (standard practice for a demonstration run), and the received text is shown in a neutral mail-reader frame.
Chapter 1
Applying for Coverage
Priya Anand speaks with an AI advisor, fills her application, and submits — all in one visit
Day 1 — application submitted
Priya Anand visits Woodgrove Life's website. Before filling out a single form field, she's greeted by an AI insurance advisor — someone who can answer her questions, explain her options, and help her find the right coverage for her family.
Priya asks about protecting her family. Before quoting anything, the advisor asks targeted qualifying questions — household income, how long her dependents need support, mortgage and other debts, and any existing coverage — so its recommendation fits her real situation rather than a one-size-fits-all number.
Priya toggles to the structured form to review and confirm her details. Every field she'll fill — applicant information, coverage choice, beneficiary designation, and attending physician — is on one clear page.
Priya's application: $750,000 of 20-year term life coverage, primary beneficiary Zara Anand (minor — guardian Deepak Anand), contingent beneficiary Deepak Anand (spouse), attending physician Dr. Marcus Feldman. The full form also includes health, lifestyle, and eligibility questions further down the page.
Application submitted. Priya receives her reference number and a link to track her application's progress — no login required. Woodgrove's team has been notified; the automated review begins immediately.
Chapter 2
Review Begins — Automatically
The moment Priya hits Submit, Woodgrove's automated review is already running
Day 1 — within seconds of submission
Seconds after Priya submitted, her case appears in Woodgrove's operations view. The case is already active; the automated review pipeline has started. No one had to manually open a ticket or route the application.
Early in Priya's case, three parallel post-intake requests resolve at different speeds. Lab Results Received and APS Document Received are both awaiting external upload (amber), while Background & MVR Analysis has already completed (green). Downstream nodes stay pending until their inputs arrive — the pipeline waits on the real world rather than guessing.
Chapter 3
Behind the Curtain — How Woodgrove Built This
Each agent runs a deterministic pipeline; each document is onboarded once and auto-mapped forever
Every agent in Woodgrove's underwriting pipeline is configured as a deterministic sequence of steps — not an open-ended LLM that can do anything it likes. The Risk Classification Agent: read the applicant's medical history and lab results, run the actuary's rule table, let an LLM make the judgment call, write the risk class back to the case. Each step is visible, auditable, and controlled. The LLM is a tool for judgment, not the driver.
Woodgrove's form library, where inbound documents are mapped to structured fields for the pipeline to read. The Lab Results Report template shows 11 fields extracted straight from an AcroForm PDF — that mapping is what lets an uploaded lab result flow into the case without anyone re-keying it. (The row selected here, Lab Analysis Summary, is a generated-output template, so it carries no inbound field mapping.)
Chapter 4
Priya Checks In — and Lab Results Arrive
Priya sees her application is in review; the exam lab uploads results securely
Day 2 — Priya checks status; lab results received and analysed
Priya follows the link from her confirmation to check where things stand. The status page shows her application is under review — underwriting is in progress. No login needed, no phone call. She can see where she is in the process and what happens next.
The exam lab receives a secure link and uploads Priya's lab results directly into the case — no email attachment, no manual intake, no re-keying. The results arrive securely and the review pipeline resumes automatically.
Woodgrove's system has analysed Priya's lab results (done). Biomarkers, lipid panel, BMI, blood pressure — all reviewed automatically against actuarial benchmarks. An initial risk profile has been generated. The attending physician statement request to Dr. Marcus Feldman is now active; risk classification and underwriting are next up in the graph, not yet started.
Chapter 5
Underwriting, Identity Verification, and Policy Generation
The APS arrives behind a verified-identity gate; automated risk classification and underwriting complete — then a human underwriter reviews and approves before the policy is generated
Day 3 — risk class: Preferred Plus; underwriter signs off; policy schedule generated
Dr. Marcus Feldman, Priya's attending physician, receives a secure, identity-verified link. He uploads the attending physician statement directly. The medical records go straight into the underwriting process — no unsecured email, no manual scanning, no transcription.
Because the attending physician statement carries protected health information, Woodgrove's portal will not show it — or accept an upload against it — until the recipient proves they are who they say they are. Dr. Feldman's office sees a verification gate first: a reference code, sent through a separate channel, is required before anything else on the page unlocks.
Identity confirmed. Only now does the upload control unlock — the same secure link cannot be used to view or submit medical information without first passing this check, even if the link itself were somehow intercepted.
Risk Classification is actively running (blue), with Risk Classification Routing still pending below it. The agent reads Priya's medical history and lab results, applies the actuary's rule table, and lets the model make the judgment call — the routing decision that follows depends on the risk class it lands on.
Risk Classification Routing is complete, and it has routed around the standard-review branch: Adverse Action Notice and Medical Director Review both show SKIPPED — explicitly bypassed for a clean Preferred Plus profile, not silently dropped. Downstream, Policy Document Generation has reached AWAITING REVIEW (amber): the agent produced a draft policy, but the node is gated for a human operator's sign-off before anything is delivered — the checkpoint the next scenes walk through.
The documents Woodgrove's agents have generated for Priya's case: the Lab Analysis Summary from the automated lab review, and the Policy Schedule — $750,000, 20-year term, policy number assigned, effective date and premium calculated. These were produced by the system's agents, not typed by a person.
Policy Document Generation is the last checkpoint before a policy is issued. Woodgrove's operator sees the node status as 'awaiting_review' — even though the assigned agent, Policy Issuance Agent, is itself configured to run autonomously. The node-level gate overrides the agent's own autonomy setting: a human must look at this one before money and paper go out the door. This is a deliberate design choice, not every node in this pipeline stops for a person — most do not — but the ones that generate or deliver a binding financial document do.
Before approving, the operator sees exactly what the agent would act on: model used (Claude Haiku 4.5, version-pinned), and every case field the pipeline read to reach this point — applicant identity, address, DOB, all Priya Anand's own data, no trace of any other applicant. The reviewer isn't clicking blind; the full pipeline context is one scroll away.
A real operator (woodgrove@demo.com) clicked Approve — the production Review-tab button, not an API shortcut. The task moves to complete, the autonomy gauge on this node reads 1/1 approved, and downstream nodes (policy delivery, signature) become eligible within seconds. The underwriting rationale is visible here too: Preferred Plus risk classification, no debit factors, clean biomarkers.
Chapter 6
Priya's Portal — Decision & Policy
Priya checks her portal and finds her decision, her policy, and her documents waiting
Day 4 — policy issued and visible in the portal
Priya logs in to the Woodgrove applicant portal — no password, just the link from her confirmation. Her dashboard shows her application: $750,000 term life coverage. One place to see everything; no confusion, no clutter.
Priya's portal shows her the underwriting outcome in plain language, and right below it, her policy — $750,000 of 20-year term life. Everything she needs, in one view.
Chapter 7
Policy in Force
Priya signs the policy receipt; coverage is confirmed and the case is complete
Day 5 → automated to hours, not weeks
Priya receives a secure link to sign and return her policy receipt. She uploads the signed document directly through the portal — the whole policy acceptance is handled digitally. No printing, no scanning, no postage.
Every step of Priya Anand's application has been processed. The autonomous pipeline is complete: intake, exam ordering, lab analysis, APS review, risk classification, underwriting, policy generation, policy delivery, receipt of signature, and final policy issuance — all done. Priya Anand is now a Woodgrove Life policyholder.
The complete underwriting record for Priya Anand's case: risk class, decision, policy number, coverage amount, and the full trail of every automated step. Every decision is documented, every data point recorded.
Chapter 8
Governance & the Communication Record
Woodgrove controls which channels can act — and messages sent on the case stay on file
Always on — channel policy + a working audit trail
Woodgrove's administrators decide, in one place, which channels are allowed to act on a case. Acting on a case by replying to email is an explicit opt-in — off by default — so the security and compliance owner signs off on the whole policy here. A channel can only take an action once it has been turned on, and even then only from a verified sender who already holds the right permissions.
Regisseur keeps a timestamped, attributed log of everything sent on Priya's case — every email delivery, secure portal link, link-open, and identity-verified health-information access, each row showing who sent it, who opened it, and when, with no personal health information exposed in the log itself. Worth knowing: an earlier run of this same demo surfaced that real email deliveries were missing from this view. That gap was fixed the same day, and this screenshot is from a run after the fix — the demo is the quality control.
Chapter 9
On the Record
Every action on this case — agent, human, and message — in one exportable audit trail, plus what the whole book actually costs to run
Always on — full auditability
Woodgrove's compliance team opens Priya's completed case and expands the Audit Trail Export: 113 events on the record — every graph event, document, and decision, timestamped and attributed to the system, an agent, or a named person — with one-click JSON/CSV export for a downloadable, attorney-ready copy. Nothing about the case's history is hidden from an auditor. (Re-captured after the automated run caught the panel mid-collapse; same case, same live system.)
Woodgrove's operations dashboard shows real spend, not a marketing estimate: total workspace cost for the week/month, and the AVERAGE cost per case across the whole book, plus any outlier cases that ran unusually expensive. This is measured from the actual LLM calls the pipeline made — not a claim about any single case's cost. (This view was re-taken later the same week, after a routine housekeeping pass: earlier demo iterations had left stale test cases, escalations, and a fired alert on the board. All of it was resolved through the product's own escalation, alerting, and archive paths — nothing was edited in the database to make the screen look better.)
A per-agent cost breakdown ("Agent Spend") — which of Woodgrove's automated agents are actually driving spend over the last 7 days, and how much each one costs to run. This is the same real measured spend, broken out by agent rather than by case. Also visible: the Workspace Doctor — 223 automated configuration health checks across every process template, form binding, and integration in the workspace, all passing at capture time.
The Record
The Communications
Every one of these actually sent — real emails, from the real send path, landing in a real inbox
To: Priya (applicant) • Sent: Jul 16, 2026, 10:23 PM UTC
The moment Priya submits, Woodgrove's system sends a real confirmation email — not a placeholder, an actual message through the platform's mail provider. It tells her the application was received and that she'll be notified as each step completes.
To: Dr. Marcus Feldman (physician) • Sent: Jul 16, 2026, 10:23 PM UTC
Woodgrove's system emails Priya's physician directly, with a secure, identity-verified link to upload the attending physician statement. A parallel email goes to the exam lab requesting Priya's lab results, subject "Lab Results Upload — Priya Anand — Woodgrove Life". Neither requires a phone call or a fax.
To: Priya (applicant) • Sent: Jul 16, 2026, 10:30 PM UTC
The final email in the chain: Priya's policy is active. Along the way she also received her policy documents by email and a secure link to confirm receipt — this is the last message in a real, end-to-end automated sequence, not a single staged "success" screen.
