The Carrier's Numbers Beside Yours
Record what the carrier quoted, from a fully insured renewal ask to stop-loss premiums, deductibles, and contract basis, and see it next to the figures your projection uses, with the gap in dollars and points.
Stay up to date with the latest features and improvements to PlanVantage.
9 releases, January 2026 to September 2026
September 2026
Latest
Built for Every Funding Type
September's release speaks each funding arrangement's own language and puts the carrier's numbers beside yours. Fully insured and level-funded dashboards get decks and exhibits built for how those contracts work, renewal projections record what the carrier quoted and what you delivered, and Creditable Coverage adds the actuarial equivalence test. You can also choose the AI model and how hard it thinks, catch contradictory plan designs before they reach a client, and create an account straight from the connector.
Record what the carrier quoted, from a fully insured renewal ask to stop-loss premiums, deductibles, and contract basis, and see it next to the figures your projection uses, with the gap in dollars and points.
Trend now follows the months of experience you actually loaded, with a per-period override. Level-funded renewals add projected components by plan and tier, the implied current expected claims, and an aggregate corridor applied after credibility blending.
Refreshing a projection keeps the experience months and inputs you entered by hand, and the Refresh Data dialog tells you what each option brings over before you click.
PlanVantage records each renewal export you deliver, flags a projection whose numbers no longer match what the client received, and asks you to confirm before you send a replacement.
Fully insured and level-funded dashboard decks now tell their own story: premium against claims for fully insured lines, and the claims fund and rate components for level-funded lines.
The year-over-year slide answers why gross claims changed, and its drivers now include the effect of members moving between richer and leaner plans.
Mark an exhibit column as the carrier's quote so it stands apart from your options. Exhibit rows use each funding arrangement's own labels, such as Total Premium for fully insured or Claims Funding for level funded.
When a plan does not pass the simplified method, run the actuarial equivalence test: set one group of assumptions, value every plan against the standard Part D benefit at once, and export the results to Excel.
The AV Model flags plan designs that contradict themselves, such as a copay above the out-of-pocket maximum, drug tiers out of order, or a primary care visit that costs more than a specialist visit. Extraction raises the same flags before you import.
Define a contribution group relative to another, such as $50 less or a percentage of it, and its amounts update whenever the anchor group changes.
Upload a census workbook with several tabs, choose the sheet to read, and preview its column headers first.
AI Fill reads a fully insured renewal exhibit onto its dashboard line, and gap-fill proposals for multi-line files name the dental or vision line they apply to.
When extraction converts a value on the way in, such as a PMPM rate restated as PEPM, the review screen says what it did. A plan document that describes a gatekeeper or narrow network prompts you to calibrate the manual rate yourself.
Choose which AI model to use and how hard it thinks (Quick, Balanced, or Thorough) from Settings or right in the chat composer. Document extraction follows your choice too.
The assistant reads several attachments at once, previews grid changes before it writes them, remembers what earlier chats changed on a scenario, and leaves rate proposals on the Rates tab for your approval.
New users can create a PlanVantage account from the connector's sign-in screen, with a setup guide to follow. Guided workflows, such as building a projection or updating experience, are now built into the connector.
Refinements across the app that make everyday work smoother.
August 2026
Ask, Review, Approve
In August the AI assistant became the front door to your work, and you stay in control of what it changes. A new home page puts an Ask bar next to your recent work, files you attach in chat load through the same reviewed extraction as an upload, and rate changes wait for your approval. Renewal projections got easier to trace, with an export that matches the screen and carries live Excel formulas, workpaper rows for the credibility blend, and assumptions you can reset. Stop-loss, plan design, and sharing each took a step forward too.
After you sign in, PlanVantage opens on a home page with your recent work, links shared with you, and your recent clients, plus an Ask bar where you drop files and say what you want done.
Drop claims reports, rate sheets, or plan documents into any assistant chat, even without a message. The assistant saves them to the client and loads them through the same reviewed extraction as an upload, and from the all-clients chat it can set up a new client first.
Ask for a rate increase and the assistant applies it through the rate engine. Specific rate overrides arrive as an approval card showing current and proposed rates by plan and tier, and nothing changes until you click Approve.
When a claims workbook spans several years, the Documents pane shows the years, months, and lines it covers, compares it with the dashboards you already loaded, and offers to fill missing months into a new dashboard version when you choose.
Each uploaded document shows a status chip for each kind of data it holds, with Re-extract, Delete data, and Stop on each. Uploading the same file again asks whether to keep the earlier extraction or run it again.
Census files with unusual headers or blank plan cells now load. PlanVantage drops columns that look like personal identifiers and tells you it did, matches plan names to the client's plans, reads tiers from the values, and asks you to pick a column only when it can't tell.
The renewal projection export now matches the screen, with live Excel formulas for percent changes, annual totals, stop-loss and fee subtotals, the credibility blend, and the fully insured build-up.
With blending on, the renewal tab shows Experience, Manual, and Blended claims PEPM rows with the credibility weight on each. With blending off, it suggests a blend when credibility is low.
Fully insured Carrier Renewal inputs are grouped into trend, credibility and manual, pooling, and retention, with a total retention row. Risk margin is entered once, as a retention item.
Edit any assumption in place and reset it to the PlanVantage value in one click. Each projection shows whether it is in sync with its source dashboard, and a refresh keeps months that exist only in the projection.
Default pooling points, expected pooling charges, and expected large claimants now come from one consistent model that trends cost and utilization separately and reflects the group's own cost level and plan richness.
Enter aggregate stop-loss premiums and claims factors as single and dependent rates on dashboards, projections, and scenarios. Renewal stop-loss rows are also quicker to edit, with separate specific and aggregate increases.
Model mail-order Rx copays as a multiple of retail for a 90-day supply, and per-day inpatient or skilled nursing copays that stop after a set number of days, in a cost-share editor that reads as a plain sentence.
The AV Model gains the undo and redo, settings menu, and drag-to-reorder plans that scenarios have, and utilization adjustment is now one value per plan that you set.
The Share dialog puts the link, its on/off switch, and quick expiry choices in one place, and the people who can open it in another. Only people on that list, plus you, can open a shared dashboard.
Through the MCP connector, your AI agent can now check a multi-year file against loaded dashboards and apply the gap fill you choose, apply rate increases and stage overrides for approval, stop one data type's extraction, and work with single and dependent stop-loss rates and per-day copay limits.
Refinements across the app that make everyday work smoother.
July 2026
Renewals In, Decks Out
July brought fully insured renewals into PlanVantage end to end and put the AI assistant on every page. Load a carrier's renewal into a projection with AI, compare suggested, carrier, and override values side by side, and export to Excel with live formulas. Dashboards gained a Slides tab you can arrange, present, and share through a live link. A new Creditable Coverage module handles Medicare Part D determinations, and document intake now reads Excel, Word, and PowerPoint files.
Every fully insured renewal assumption (trend, pooling point, pooling charge, retention, manual rate, and credibility) shows a suggested value, the carrier's figure, and your override, with the difference under each result.
Pull a carrier's fully insured renewal, current rates, level-funded components, claims experience, large claimants, stop-loss terms, and census enrollment into a projection with AI, and review each value before it is applied.
Export a projection to Excel with live formulas on the experience, large claimant, and claims projection sheets, so you can trace the math and change an input in the workbook.
Turn a renewal projection into a status-quo scenario in one step, with rates, fixed costs, and contributions carried over, and start modeling plan changes right away.
A new Slides tab lets you reorder or hide slides, choose which lines appear, and add your own title and text slides, with an AI-drafted summary to start from. Present full screen or download a PDF.
Share a plan-year deck through a live link that always shows current data, with an optional expiration date.
The year-over-year slide breaks the change in cost into drivers, such as enrollment growth, plan and tier mix, trend, and fixed costs, and the multi-year slide compares each year with market trend.
Compare two claims-experience versions cell by cell, including large claimants. Enter data with Excel-style copy, paste, and fill, run AI Fill from several files at once, and spot versions that still have months missing.
A new Creditable Coverage module tests each plan's prescription drug coverage against the CMS simplified determination for Medicare Part D and exports the results to Excel.
Scenarios gain an Inputs tab that understands funding type: model specific stop-loss with lasers and admin fees by tier or plan, pull those terms from a document with AI, and hide the tabs you don't use.
Open the assistant from any page and @-mention the scenarios, dashboards, and projections you mean. It asks before deleting anything, you can undo a run in one click, and chat history is grouped by client.
A new Help center has tutorials matched to your plan, a glossary of benefits terms, and a keyboard shortcuts page, and the assistant answers how-to questions from those tutorials.
Upload Excel, Word, and PowerPoint files and PDFs up to 300 pages. Funding extraction now captures admin fees by tier or plan, which plans each fee applies to, and aggregate stop-loss claims factors.
Each client's home page opens with recent items, a plan-year health summary, and an enrollment snapshot, and long client lists load much faster.
The MCP connector now offers a smaller, consistent set of find, get, manage, and delete tools, plus new tools for monthly experience updates, dashboard metrics, Creditable Coverage, AV models, and loading data into projections.
Refinements across the app that make everyday work smoother.
June 2026
The 2.0 release
PlanVantage 2.0 is the largest release in the platform's history: multi-line benefits dashboards with independent funding and presentation-ready slide decks, full renewal projections with credibility blending and stop-loss modeling, a Med/Rx actuarial-value split, contribution surcharges and credits, a unified AI "Load Data" intake for every document type, and an AI agent layer you can drive from your own assistant through the connector and API. A new four-tier plan ladder (Free, AV, Scenario, Full) plus an Enterprise governance overlay rounds it out.
Build client-ready experience dashboards spanning medical, dental and vision, including multiple medical lines, each with its own independent funding arrangement.
Track claims experience as dated versions and let AI populate them straight from carrier reports, with completion and credibility surfaced as you go.
Generate presentation-ready slide decks for each line of coverage, deep-linkable by slide and line, with cover pages derived from the data.
Chain renewals across plan years so each year builds on the last instead of starting from a blank sheet.
Project renewals with credibility blending between a group's own experience and pooled or manual rates, with full control over the blend.
Model ISL and ASL with multiple premium-input granularities (composite, by tier, by plan, by plan-and-tier) and apply specific-stop-loss lasers per member.
Level-funded builds carry arbitrary AI-captured funding components (commissions, fees, and more) pulled straight from carrier reports, with the funding rate enforced as the sum of its parts.
Compute medical-only and Rx-only actuarial values independently, with the Rx AV gated on creditable-coverage rules, useful for Medicare and carve-out analysis.
Layer surcharges and credits (spousal, tobacco, wellness, and custom) onto contribution strategies as first-class, normalized modifiers on both the current and proposed sides.
A single "Load Data" wizard ingests any of seven document types (plan designs, rates, contributions, fixed costs, claims experience, funding, and census) into a scenario, a dashboard, a brand-new dashboard, or an experience version.
Extraction runs through one shared dialog with confidence flags and an apply step, so what the model proposes is always yours to review before it lands.
An open MCP connector and scoped API keys let your own AI assistant work inside PlanVantage: load a folder of renewal documents, build a dashboard, or draft a projection through natural-language workflows. The connector is agent-agnostic; the bundled plugin ships a load-a-folder workflow today.
Organize work in folders, share across plan sponsors, and track supplemental carriers alongside primary coverage.
Four capability tiers (Free, AV, Scenario, and Full), each including everything below it, plus an Enterprise governance overlay with SSO and audit logs. AI is included across every tier, with no per-token bills.
May 2026
Tuned for Throughput
Smarter document extraction with multi-level rules and confidence flags, per-user processing queues so heavy uploads stay responsive, a richer migration estimation flow, and a cleaner deductible/OOPM model.
Layer extraction instructions at the plan sponsor, scenario, and document level. Rules cascade so you can set defaults once and override only where needed, with a new UI to manage the hierarchy.
Every extracted plan field now carries a confidence signal. The redesigned Extraction Notes tab surfaces flagged values with a summary view, making it easy to spot fields that need a closer look.
Improvements to how documents are scanned produce more accurate plan lists on complex multi-plan PDFs and benefits summaries.
Subsequent extractions on the same document return noticeably faster, especially during back-and-forth refinement and reprocessing.
Document and contribution extraction now run on a per-user queue. Upload a batch and items process in order without one heavy user blocking another. Queued items show a clear status while they wait their turn.
Per-sponsor document limit raised from 50 to 1,000 to support customers with deep historical libraries and large book reviews.
Cancel a long-running agent conversation mid-loop instead of waiting for it to finish. Canceled and failed states are now visually distinct, and errors surface inline with detail.
When running a migration, choose which contribution option to use for the estimate. The selector stays visible (disabled when no options exist) with clearer helper text so the flow is obvious from the start.
Migration estimates now produce more realistic enrollment shifts, especially when the proposed lineup differs meaningfully from the current plans. Employees move toward better-value options instead of staying anchored to look-alikes.
The deductible and out-of-pocket-max model has been refined. The "non-embedded" type replaces the older type-family concept with clearer, unbundled semantics: what you select is what you get, with no hidden coupling between deductible and OOPM behavior.
Subgroup comparison settings (baseline group, comparison toggle, and group management permissions) now carry through scenario clones, contribution option imports, and undo/redo snapshots. Comparison columns also appear in contribution model exports.
When rate override mode is on, Status Quo assumptions remain editable so you can run override comparisons without losing your baseline inputs.
Auto-calculated fixed costs in the summary are now held constant at the base PEPM across scenarios, so changes elsewhere don't cause unintended fixed-cost drift.
Download the active census directly from the census dialog. Boolean criteria dropdowns now correctly display false and 0 values instead of falling back to "Select".
Enter-to-next-cell navigation extended to difference, comparison, and cost-share percentage cells, matching how rate and contribution tables already work.
Stronger selection shading on editable result cells, long names now stay readable on hover instead of being clipped, and a tighter header card layout throughout.
New tool for updating plan document extraction rules from the MCP connector. Queued status, group management permissions, and processing group fields exposed to agents.
API documentation expanded to cover plan document rules and account endpoints. Routine dependency upgrades across the platform.
April 2026
Contribution Refined
A redesigned contribution comparison experience with group difference columns, an option library for cross-scenario imports, and bulk paste operations. Plus ongoing platform maintenance and infrastructure improvements.
Compare contribution amounts between any two groups side by side. Select a baseline group and see dollar and percentage differences for every tier, with styled headers and disabled baseline cells.
Import contribution options from other scenarios within the same plan sponsor. The library shows all options, both AI-calculated and manual, with compatibility indicators and automatic enrollment distribution.
Paste tabular data directly into rate and contribution tables. The new bulk update endpoint processes multi-cell changes in a single request.
Core rate model calculations extracted into pure functions for testability and clarity. Manual rate method now has its own explicit dispatcher path. Fixed a spread HSA bug uncovered during the refactor.
Rate override values are now correctly preserved when cloning scenarios, both through the standard clone path and the current-to-proposed rate plan copy.
Census tier mapping now uses scenario-specific tier names instead of defaults. Partial disambiguation results are persisted so you can resume where you left off. Upload and mapping statuses are tracked independently.
17 new endpoints added to the OpenAPI specification covering SSO, MFA, SIEM, admin, and contribution copy operations. Plan design exports now sort by rate plan order in sections view.
All numeric cells in rate and contribution tables now use JetBrains Mono for consistent digit alignment and easier scanning.
Sticky horizontal positioning for section header labels, improved table padding, better text wrapping for long group names, and various border fixes across contribution and plan design tables.
9 new tools added covering exhibit fixed costs, contribution copies, and scenario operations. Bulk update tool for paste operations. Improved OAuth discovery and endpoint routing for the Claude.ai connector.
Routine infrastructure updates and dependency upgrades across the platform.
March 2026
Smarter AI, Faster Loops
Enhanced AI infrastructure with OpenAI Responses API integration for faster multi-turn agent conversations, GPT-5.4 support, and improved agent reliability across all providers.
Agent tool-use conversations now use server-side state via previous_response_id, eliminating redundant token usage on follow-up iterations. This reduces cost and latency for multi-step agent operations.
Full support for OpenAI's latest GPT-5.4 model with reasoning effort parameters for agent requests.
Each agent loop iteration now logs individually with tool call details, enabling better debugging and cost attribution for complex multi-step operations.
Agents can now access tools across plan, rate, contribution, and summary models in a single conversation, enabling complex multi-model operations.
New agent tool for setting scenario tier names directly from chat, plus automatic enrollment redistribution when tier structures change.
Agent loops now detect when consecutive write operations keep failing and bail out early instead of burning tokens on repeated failures.
Routine dependency upgrades across the platform.
Added census, export, M&A, and MCP guidance to agent system prompts for more contextually aware responses.
February 2026
AI Agent & Exhibits
This release introduces AI-powered chat agents for hands-free data editing across all four models, a redesigned exhibit comparison view, global undo/redo, and dozens of UX refinements including optimistic updates throughout the interface.
A multi-turn agentic chat system that can read and modify your scenario data through natural language. Ask the agent to change deductibles, update rates, adjust contributions, or modify summary settings: it reads the current state, makes changes, and verifies the results.
A new exhibit mode that renders side-by-side scenario comparisons with current/status quo baseline columns and dynamic scenario columns.
A single undo/redo stack captures all changes across plan designs, rates, contributions, and summary settings. Snapshots are taken automatically before the first mutation and after each successful change.
Quickly copy contribution amounts from a current plan to its proposed counterpart with a single click from the Rate or Contribution tabs.
Mutations across all models now reflect instantly in the UI before the server responds, eliminating perceived latency.
New display toggles for rate model views including hide/show options for plan and group total rows with hover eye-off buttons.
Improved migration dialog with re-estimate and dismiss buttons for stale warnings, plus direct apply option from the census view.
Plan and tier dropdown selectors in the census criteria editor with LLM-powered inference for automatic criteria suggestions.
Password setup flow for new users and admin-initiated account creation.
Exhibit fixed cost rows are now dynamically generated rather than cached, and orphaned exhibit-scenario links are automatically cleaned up.
Admin endpoint to clone plan sponsors across user accounts including all documents, censuses, and file references.
January 2026
Census & Comparison
Comprehensive census data management, current vs. proposed plan comparison, bulk data export capabilities, and numerous UI improvements to streamline your benefits analysis workflow.
Upload and analyze your employee census data directly within PlanVantage. The system uses AI-powered mapping to automatically match your census file columns and values to your scenario's rate plans and tiers.
The Plan Model now supports organizing plans into Current and Proposed sections, making it easier to compare existing benefits against renewal options.
Export all your data across plan sponsors and scenarios for external analysis or reporting.
When your proposed plans use a different tier structure than your current plans, a new Maintain Base Rate option ensures consistent employee-only rates while recalculating family tier ratios.
Streamlined navigation with back button, cleaner scenario name display with inline editing, and reorganized action buttons for Share, Export, Census, and AI Chat.
Tab-style navigation between Plan Designs, Rates, and Contributions with consistent styling across all model views and improved loading states.
Section headers now show enrollment totals, improved group and option management dialogs, and better handling of contribution group criteria display.
Plan Model exports now properly handle current/proposed sections with appropriate sheet organization.
Improved data loading for scenarios with large numbers of plans.
New endpoints for census management, AI model configuration, and bulk data export.
Have questions or feedback? Contact us