Statement of Work · August 2026
A 30-day build placing a guided enrollment experience in front of members ahead of the September 1 window — layered on the platform NROLL NOW already operates.
Summit Alliance Financial, LLP provides the benefits packages, tied to the payroll service. NROLL NOW — of which Summit is part owner — provides the benefits administration platform this work builds against. What is built becomes the foundation of Bookah, a new entity formed through this engagement. Other companies sit alongside them in the group and the wider vendor set; August is scoped to what can be achieved now, and built so those pieces can be added later.
An enrollment window opens September 1 — a multi-hundred-person group behind it, and 1,500 – 2,500 truckers to reach. Today that volume goes to a call center: the most expensive channel, and the hardest to scale on short notice.
NROLL NOW already owns what matters — ten years of rules, rate tables, eligibility logic and carrier configuration. What is missing is a modern, assisted way into it. That is all this engagement builds.
The four items scoped here — Enrollment, Web Self-Service, AI Agent, Call Center — occupy the center of that diagram, between the NROLL NOW benefits administration platform and the wider product and service provider ecosystem.
That position is the point. These four are not another spoke on the wheel; they are the connective layer through which employers, employees and key vendors reach what NROLL NOW already operates. Built well, the same layer serves every product and every vendor added to the right-hand side later — which is what makes it an asset rather than a feature.
| Remains in the NROLL NOW platform | Built in August, reading from it |
|---|---|
| Rules and eligibility tables | Member questions answered against those rules |
| Rates, states, premiums, coverages | Cost explained to the member in plain language |
| Calculations | Results presented and narrated, never recalculated |
| Brochures and plan documents | The same documents, delivered interactively |
| System of record | Enrollments written back; the layer holds no authoritative store |
| Carrier and product configuration | The member-facing shelf, driven entirely by that configuration |
A single direction of truth. Where the two could ever differ, the NROLL NOW platform is authoritative by construction, because the assisted layer maintains no independent copy.
One rules engine — NROLL NOW's. Rates, eligibility and coverage are computed by NROLL NOW and never recomputed. Documents may be indexed ahead of time for speed; a decision never is.
NROLL NOW is the system of record. Bookah is the system of engagement. Coverage, rates and the enrollment itself are NROLL NOW's, authoritatively. The conversation, consents, tickets and audit trail sit with Bookah — records of the interaction, not of the coverage.
Restricted retrieval. Responses come from NROLL NOW's own material and APIs. Never from a general-purpose model's own knowledge.
Reusable, without leakage. The layer is built to serve other clients later — that is what makes it an asset. Tenant boundaries are absolute: NROLL NOW's rules, documents and members never cross into another instance.
The list NROLL NOW circulated is the scope of this month. It is set out below as six workstreams rather than thirty-four lines, because several of those lines are the same requirement asked more than once — and it closes with a map showing where every single one of them is delivered.
Four of the workstreams — Enrollment, Web Self-Service, AI Agent, Call Center — come straight from the centre of the Strategy and Objectives model. Enrollment is the outcome; those three are the channels to it. Two more, Platform & Integration and Operations & Evidence, carry the requirements about how it connects, how it is run, and how it is proven.
They form a cost ladder. Web self-service is the least expensive channel — the member completes it alone. The AI agent is next — it catches the member who would otherwise abandon at the point of confusion. The call center is the most expensive, and is reserved for those who genuinely need a person.
The objective is not to remove the call center. It is to ensure that everyone who reaches it needed to — and that nobody arrives there simply because a screen went unexplained.
The list NROLL NOW circulated is the specification for this month. Each line below maps to the workstream above that satisfies it. Thirty of the thirty-four are delivered in August; the four that are not are marked, and the reasons are set out in section 04 rather than buried.
| Requirement | Delivered by | Status |
|---|---|---|
| NROLL NOW is the system of record | Platform & Integration | Delivered |
| Enrollment written into NROLL NOW | Platform & Integration | Delivered |
| Closed–restricted retrieval | AI Agent | Delivered |
| API with NROLL NOW, including brochures | Platform & Integration | Delivered |
| Integration between platforms | Platform & Integration | Delivered |
| Security level | AI Agent · compliance controls | Delivered |
| Calendar invite supported | Call Center | Delivered |
| Ticket system for unanswered questions | Call Center | Delivered |
| Outbound robo calls | — | Excluded by choice |
| Sell the AI agent as revenue to others | AI Agent · ownership | Delivered |
| Build timeline with testing | 1 – 31 August | Delivered |
| Health and benefits people on staff | Yours, not ours — by design | Not provided |
| Own the instance after build | AI Agent · ownership | Delivered |
| AI screen share with the employee's screen | Enrollment | Delivered |
| Voice-only support | AI Agent · hands-free | Delivered |
| No lag time | Platform & Integration | Delivered |
| Volume issues covered | Platform & Integration | Delivered |
| Use the ticket system | Call Center | Delivered |
| Case set-up time short | Platform & Integration | Delivered |
| Back-up for future IT and dev support | Operations · the Editor | Delivered |
| Robo calls supported | — | Excluded by choice |
| Global script process, no hallucinations | AI Agent · restricted retrieval | Delivered |
| All APIs tested for data and PDFs | Platform & Integration | Delivered |
| Functional application with UAT | Operations & Evidence | Delivered |
| An existing AI agent demo | Live now | Delivered |
| Weekly synchronisation audit report | Operations & Evidence | Delivered |
| Analytics dashboard | Operations & Evidence | Delivered |
| Dedicated admin centre | Operations & Evidence | Delivered |
| Performance and SLAs defined | Operations & Evidence | Delivered |
| Sprints and stories mapped | Operations & Evidence | Delivered |
| A development team of eight people | Not applicable — see §04 | Not provided |
| Integrate with merchant services | Platform & Integration | Delivered |
| Discovery and SOW done | Complete | Delivered |
| Use existing AI experience | Throughout | Delivered |
Two lines are excluded deliberately rather than by phasing — outbound automated calling is a liability decision, not an engineering one. Two are honest noes: we do not employ licensed benefits staff and would not place them in this path if we did, and we are not a team of eight. Section 04 argues why neither should decide this.
Building against the existing platform means working inside the software estate for a month. That access is used deliberately rather than incidentally: alongside the build, August produces an initial scan and review of the services NROLL NOW currently pays for.
| The scan covers | What it produces |
|---|---|
| Systems, vendors and annual cost | A ledger of what is being paid for, by whom, and under what contract |
| Renewal dates and exit terms | The windows in which a decision can actually be made |
| Overlapping and duplicate tooling | Candidates for consolidation, ranked by saving and by risk |
| Licence seats provisioned against used | Seats being paid for and not consumed |
| Integration dependencies | What is load-bearing and must not be touched |
| Capability already inside the platform | Subscriptions duplicating something NROLL NOW already owns |
An initial position, not the full technology diagnostic. It establishes the ledger and a first savings figure; the complete inventory, kill list and annualized savings position follow in September.
A subscription is rented capability: priced by someone else, repriced at renewal, and after ten years of payment nothing is owned. Move that capability inside the platform NROLL NOW already operates and it becomes owned rather than rented — an asset, not an expense line.
Nothing is cut in August. The deliverable is a ranked, costed view of where conversion pays — and the enrollment layer built this month is the first example, replacing outsourced enrollment capacity with capability NROLL NOW owns.
The assisted layer reads from the NROLL NOW platform and cannot be built without access to it. These items sit on the critical path.
| By | Required from NROLL NOW |
|---|---|
| Week 1 | Sandbox access and API documentation for rules, rates, eligibility and product configuration |
| Week 1 | A named technical contact with working knowledge of the platform, available on a scheduled basis |
| Week 1 | Advantage package plan documents, brochures and benefit descriptions |
| Week 2 | Test member records and a test employer group |
| Week 2 | The write path for a completed enrollment, and the validation NROLL NOW enforces |
| Week 3 | A licensed agent and calendar for the escalation path |
Where an item is delayed, work continues against a temporary substitute and is reconciled on access. Rework arising from such delay falls outside this fixed fee.
August delivers the member enrollment path, not the full platform across all user types. The complete platform serves four audiences — member, employer, operator, payroll partner — each with its own surface and permissions. That is the wider engagement.
This release is the member path, proven end to end on the Advantage package. Employer console, operator tooling and partner surface follow later.
Ongoing IT support is not funded by the August budget. The fee here covers the build in this document. What August does deliver is the environment that makes ongoing support cheap to provide — the Editor described in section 04 — so the September conversation is about pricing a support agreement, not about building a capability from nothing.
Security is built to carrier-appropriate standards throughout; formal SOC 2 certification is a later undertaking. Payroll, HCM and the key vendor set sit on the product and service provider side of the model and are integration work for later phases — the layer built here is designed to receive them. Outbound automated calling is absent by design rather than by phasing: this platform serves members who have come to it, and does not initiate contact.
This sprint is deliberately narrow, and it is not the point. Enrollment is the near-term hurdle. The long-term objective is Bookah — an asset with value of its own, built while the immediate problem is being solved rather than after it.
A short, contained engagement lets both parties establish how they work together before larger commitments are made. August produces a working enrollment path, a documented understanding of the existing platform, and a basis on which September can be scoped accurately.
After the two August payments, Sharemeister takes an agreed percentage of enrollments completed through the platform — until $20,000 has been paid, at which point it stops permanently. An earn-out, not an ongoing royalty.
This discounts the upfront cost at a moment when cash matters, and ties the balance to the platform actually working. If it enrolls no one, nothing further is owed. It ends at the cap or at 24 months, whichever comes first.
An enrollment counts if Bookah was part of it, including those a licensed agent completes.
At a rate that makes sense for the client.
This settles the August sprint in full. It sits outside the broader Bookah platform engagement — not credited against those terms, and not part of them.
The percentage should be set against commission earned rather than premium written — premium is the member's money; commission is the group's. Both parties to confirm that basis in writing before the second payment.
This is a discrete first phase, priced so that work can begin immediately rather than await a full commercial negotiation. The cash and the earn-out together settle the August scope in full and stand apart from the longer engagement — neither a draw against it nor a part of it.
The wider platform engagement remains a commitment to build out the broader platform, at a price to be determined — comprising payment for services, equity, and revenue share.
As part of those terms, the parties agree to work out in good faith a bonus-equity package for Sharemeister — a stake that participates both in the revenue the platform generates and in the proceeds of any eventual sale of the company — consistent with the stated long-term objective of selling the platform.
Those terms will be worked out as this first sprint completes at the end of August, with both parties better informed for having built something together first.
Acceptance authorizes commencement and the first payment of $10,000. Work begins on receipt. The earn-out percentage and its basis are to be agreed in writing before the second payment on 15 August.
This Statement of Work covers the period 1 – 31 August 2026 and the scope described at section 04. Terms for subsequent phases are to be agreed separately.
The assisted enrollment layer is an automated tool and is not a licensed insurance agent. A licensed, appointed agent reviews and signs every enrollment before coverage is active. The supplemental and fixed-indemnity products referenced pay a set benefit toward covered events; they are not major medical insurance and are designed to operate alongside, not in place of, comprehensive health coverage. Nothing in this document constitutes insurance, tax or legal advice.