🚀 AIS Value Model · Supplement to the BOS · v0.2 · 🟡 Draft
How an embedded AI Specialist creates visible value in the first 30 days, keeps creating it every month, and feeds what they build back into a library that makes the next engagement faster.
Draft, rebuilt with Catalogue v1. Prepared for Delivery & Client Operations and AI Architecture & Technology. Blended into the BOS after the current review. Companion: REMAP Client Stack, Reference Architecture v0.1.
⚠️ Draft status. Nothing on this page changes the Accountability Chart, the meeting rhythm or the seat charters yet. Where it differs from a seat charter, the charter applies until leadership decides otherwise. See Decisions required below.
Clients tell us the same thing at discovery: we don't know what AI can do for us. Today we answer with a fully custom audit and a fully custom roadmap. It works, but every engagement starts from zero, and nothing a Specialist builds for one client makes the next client faster.
The AIS Value Model changes the default:
The AI Specialist retainer is the constant. The Catalogue is what the client sees delivered against it every month. The AI Audit finds the bespoke work; the Catalogue fills the space between bespoke projects so there is never a month without a visible win.
| 30 days | Seat start to four working items and live infrastructure |
| ~50% | Of every roadmap drawn from the Catalogue; the rest custom from the audit |
| 2 + 1 | Catalogue items plus one project slot per month, from Month 2 |
| 2 | Launch verticals: Real Estate and General SMB. Others earn their way in. |
Five stages that repeat across every engagement and compound across the business.
| # | Stage | Owner | What happens |
|---|---|---|---|
| 1 | Assess | Sales motion | Remote assessment at intelligence.remap.ai; vertical inferred; gated proposal page with three picks |
| 2 | Deploy | AI Specialist in seat | Kickoff Sprint in Month 1 (four items live), then roadmap projects in Months 2–6 |
| 3 | Capture | Business Analyst | What was built: config, outcome, time to deploy |
| 4 | Publish | Marketing | One anonymised deployment piece per client per month |
| 5 | Library | Technology | Promoted into the Catalogue if deployable |
The Library-to-Assess return is where it compounds: every catalogued item is pre-selected for the next client in the same vertical.
Every engagement deploys the same stack. The client keeps the interface they already know; REMAP owns everything behind it.
| Layer | What it is | Who owns it |
|---|---|---|
| Interface | Claude Team, Claude Desktop (via gateway, sovereign), Microsoft 365 Copilot, or Open WebUI by exception | Client-owned and client-billed. REMAP never bills for it. |
| Skills | Named actions the client invokes, identical across interfaces | REMAP-provisioned |
| MCP surface | n8n endpoint, memory endpoint, dashboard endpoint, per client | REMAP; the part that does not swap |
| Automation core | n8n, dedicated instance, with the REMAP workflow library | Private AI environment, client region |
| Memory | Wiki.js + Postgres, MCP-exposed | Portable: the client's when they leave |
| Intelligence | Metabase on a client subdomain (no separate BI build) | Portable: the client's when they leave |
| Model gateway | LiteLLM to Bedrock, client region, zero retention, any model | Private AI environment |
| Secrets | Infisical | Private AI environment |
| Client systems | CRM, mail, files | Connect at the automation core, never directly at the interface |
Grade is set by requirement, not geography. A US clinic under HIPAA and an Australian firm under APRA get the same answer. Skills, MCP surface, n8n and memory are identical in both grades; only the inference route changes.
| Grade 1 · Standard | Grade 2 · Sovereign | |
|---|---|---|
| Path | User in Claude (any device) → Anthropic model under zero-data-retention commercial terms → n8n in client environment → Bedrock in region for heavy work | User in Claude Desktop (developer mode) → LiteLLM gateway (REMAP-hosted, client environment) → n8n → Bedrock in region for all inference |
| What Anthropic sees | The prompt, the skill call, the summary and direct uploads. Not the documents, records or transcripts the workflow processes. | Nothing. No Anthropic-operated server in the path. |
| Status | Default | Live with a REMAP client today. Sovereign work is laptop work. |
Three yes/no questions at discovery set the grade before the Order Form is issued:
Any yes → Grade 2.
The client sees delivered infrastructure in the first week and four working things by Day 30.
Selection is fixed-choice so it can be written into an Order Form, and sized by Specialist effort.
| Tier | Count | Due | What it is |
|---|---|---|---|
| Tier 0 · Foundations | Up to 3 | Day 5 | Infrastructure spun up from the standard composition. No client data needed; half a day of DevOps for the whole set. The fastest visible thing REMAP does. |
| Tier 1 · Quick Wins | 2 | Day 14 | Small automations, under a day of Specialist time each, wired to something the client already runs (mail, calendar, CRM, PMS, forms). If the client has none of those, one slot converts to a fourth Foundation. |
| Tier 2 · Core Build | 1 | Day 24 | One system configured to the business. Two to three days of Specialist time. If they already have a CRM, the Core Build is hygiene automation on it, not a replacement. |
| Tier 3 · Scorecard | 1 | Day 30 | Their own numbers, live: a dashboard or brief on the metrics the owner named at discovery. One to two days. Always their data, never a template. It is also the retention lever: ROI stays visible every month after. |
About six to eight Specialist days in total, leaving room for the audit and admin setup. Every item is named as an outcome the client gets, not a tool.
Naming: "Kickoff Sprint" replaces "Banquet". Foundations · Quick Wins · Core Build · Scorecard replace the course metaphor.
| When | Step | Side |
|---|---|---|
| D−14 | Remote assessment | Sales |
| D−10 | Proposal page + discovery call | Sales |
| D−3 | Signed Order Form and funds received | Sales |
| Day 0 | Seat starts | Delivery |
| Days 1–5 | Foundations live | Delivery |
| Days 6–14 | Quick Wins delivered | Delivery |
| Days 15–24 | Core Build delivered | Delivery |
| Days 25–30 | Scorecard delivered | Delivery |
| Day 30 | Kickoff review | Delivery |
| Month 3 | Strategy Review | Delivery |
The proposal page is the assembly point. It is built by the client-page skill from three inputs: the intelligence assessment, the discovery transcript, and the Kickoff Sprint picks. Output: a gated proposal with Month 1 items selected, Months 2–3 named, and one line saying more was identified on the call. The gated page never indexes; the anonymised public industry page is the SEO asset. Don't merge them.
| Month | Focus | Plus |
|---|---|---|
| 1 | Kickoff Sprint: Foundations + 2 Quick Wins + 1 Core Build + 1 Scorecard, committed in the Order Form | — |
| 2 | Roadmap project (e.g. AI Executive Agent) | 2 Catalogue items chosen at the Day 30 review |
| 3 | Custom project from the audit's business layer · Strategy Review | 2 Catalogue items |
| 4 | Roadmap project; team vote on the vertical-filtered Catalogue feeds the pick | 2 Catalogue items |
| 5 | Custom or roadmap; team vote decides | 2 Catalogue items |
| 6 | Project + renewal: next six months sequenced · Strategy Review | 2 Catalogue items |
Two catalogue items plus one project a month gives the client a visible win every fortnight, keeps the Specialist from drowning in bespoke work, and the quarterly review re-sequences. The stickiness comes from the client picking the next thing themselves.
| Every month | At each Strategy Review |
|---|---|
| Catalogue re-presented; client picks two | Roadmap re-audited |
| One deployment piece published | Catalogue re-presented; team vote reviewed |
| BA captures what was built | Gaps surfaced; months ahead re-sequenced |
The team vote: runs against the vertical-filtered Catalogue beforehand. Three votes each. It is input, not a queue. No votes → the Architect recommends. The vote needs a browsable catalogue (outcome name, one-line description, who benefits, logo, link), not just a form; both write to the same table.
One list, two surfaces: remap.ai (public, AEO/GEO) and clients.remap.ai (selection and vote).
Verticals: RE-S = Real estate sales · RE-PM = Real estate property management · SMB = General SMB
Evidence grades: A independent, peer-reviewed or analyst source · B named vendor case study · C vendor benchmark, directional · – operational benefit, no headline number.
Only A-grade numbers go into client copy unqualified.
| ID | What the client gets | What it is | Stack | Verticals | Client-choosable |
|---|---|---|---|---|---|
| F1 | A CRM that's actually yours | Pipeline, contacts, companies, notes; exportable when they leave | Twenty | All | Yes |
| F2 | A company brain your team can search | Internal wiki, seeded from existing docs if chosen as the Core Build | Wiki.js + Postgres | All | Yes |
| F3 | Live dashboards on your own data | BI layer on a client subdomain; populated by the Scorecard | Metabase | All | Yes |
| F4 | Booking links that respect your calendar | Scheduling, round-robin, CRM write-back | Cal.com | All | Yes |
| F5 | Social posts scheduled from one place | Multi-channel scheduler with approval queue | Postiz | RE-S, SMB | Yes |
| F6 | One inbox for WhatsApp, web chat and email | Shared inbox and agent console | Chatwoot | All | Yes |
| F7 | Contracts signed without per-envelope fees | E-signature with audit trail | Documenso | All | Yes |
| F8 | Automation engine behind everything | Dedicated instance; REMAP workflow library licensed | n8n | All | Always included |
| F9 | Any AI model, your region, zero retention | Model gateway to Bedrock | LiteLLM | All | Always included |
| F10 | Your passwords and API keys locked down | Secrets manager | Infisical | All | Always included |
| F11 | Web and LinkedIn research on demand | Scraping and enrichment | Apify | RE-S, SMB | Yes |
| F12 | Private AI environment, your region | The composition itself; shown on every invoice as included value | Private AI hosting | All | Always included |
| ID | What the client gets | Who benefits | Verticals | Evidence | Days |
|---|---|---|---|---|---|
| Q1 | Every lead answered in under 5 minutes | Sales, agents, front desk | RE-S, SMB | A | 1 |
| Q2 | Meeting notes that write themselves into the CRM | Anyone in meetings | RE-S, SMB | C | 1 |
| Q3 | Never miss a follow-up | Sales, principals | All | – | 0.5 |
| Q4 | Appraisal follow-up on autopilot | Listing agents | RE-S | C | 1 |
| Q5 | Weekly numbers in your inbox | Owners, managers | All | – | 0.5 |
| Q6 | Your day, briefed before it starts | Leaders | SMB, RE-S | – | 0.5 |
| Q7 | Documents drafted from a template in seconds | Admin, sales | RE-S, SMB | A | 1 |
| Q8 | Rent reminders and arrears chasing, automated | Property managers, accounts | RE-PM | B | 1 |
| Q9 | Lease renewals started 90 days out | Property managers | RE-PM | B | 1 |
| Q10 | Tenants acknowledged instantly, updated automatically | Property managers | RE-PM | B | 1 |
| Q11 | Invoices chased without the awkward call | Finance, owner | SMB | C | 1 |
| Q12 | Receipts and expenses coded automatically | Everyone, finance | SMB | B | 1 |
| Q13 | Listing copy and socials drafted from the property data | Marketing coordinator, agents | RE-S | C | 1 |
| Q14 | Approved content posted on schedule | Marketing, owner | RE-S, SMB | B | 0.5 |
| Q15 | Website and forms watched 24/7 | Owner, marketing | All | – | 0.5 |
| Q16 | Leads tagged by source, automatically | Sales, marketing | RE-S, SMB | – | 0.5 |
| Q17 | Payment status at a glance | Owner, finance | SMB | – | 0.5 |
| Q18 | Interviews booked without the email tennis | Hiring managers | SMB | B | 1 |
| Q19 | Your suburb, every Monday (Real Estate industry starter) | Sales agents, principals | RE-S | – | 1 |
| ID | What the client gets | Who benefits | Verticals | Evidence | Days |
|---|---|---|---|---|---|
| C1 | A CRM configured to how you actually sell (or hygiene automation on their existing CRM) | Sales team | RE-S, SMB | C | 3 |
| C2 | Your company knowledge, searchable and current | Whole team | All | – | 2 |
| C3 | Bookings that land in the right calendar and the CRM | Sales, admin, front desk | All | A | 2 |
| C4 | Intake and onboarding without the back-and-forth | Admin, ops | All | B | 3 |
| C5 | Maintenance requests triaged and dispatched 24/7 | Property managers, maintenance | RE-PM | B | 3 |
| C6 | Client reports that write themselves | Account managers, advisors | SMB | A | 3 |
| C7 | Invoices processed without typing | AP, bookkeeper | SMB | C | 3 |
| C8 | Tier-1 support handled, the rest routed | Support, ops | SMB | A | 3 |
| ID | What the client gets | Who benefits | Verticals | Days |
|---|---|---|---|---|
| S1 | Your top five numbers, live | Owner, leadership | All | 1.5 |
| S2 | Appraisal-to-listing conversion, by agent and source | Principals, sales managers | RE-S | 1.5 |
| S3 | Doors per manager and where the time goes | Agency principal, PM head | RE-PM | 1.5 |
| S4 | Ask your documents anything | Whole team | All | 1.5 |
| S5 | Pipeline forecast you can defend | Sales manager, owner | SMB | 2 |
| S6 | Where the money leaks | Owner, finance | SMB | 1.5 |
| Project | What the client gets | Verticals |
|---|---|---|
| AI Executive Agent | The "where should I focus" brief across mail, calendar, CRM, tasks and memory | All |
| Client WhatsApp Agent | Inbound handling, qualification, booking, handoff | All |
| Engagement Agent | Outbound nurture and re-engagement from CRM triggers | RE-S, SMB |
| Onboarding / Intake Agent | New-client or new-hire sequence with document collection and provisioning | All |
| Speed-to-lead full stack | Unified inbox, AI first response, qualification, routing, CRM logging, all channels (Q1 grown into the front door) | RE-S, SMB |
| Tenant Lifecycle Agent | Q8 + Q9 + Q10 + C5 unified: maintenance, arrears, renewals and comms in one agent | RE-PM |
| Custom project | Discovered in the audit's business layer (process automations, AEO/GEO content, air-gap deployment, Open WebUI setup) | All |
Deliberately excluded: an inbox triage tool. Copilot and Google already do this; REMAP configures the client's existing tool rather than competing with it.
Every item is an instance of one of these. Building each once is the margin.
| Playbook | Items |
|---|---|
| Speed-to-lead / instant response | Q1, C8, Client WhatsApp Agent, Speed-to-lead full stack |
| Scheduling / dispatch / routing | C3, C5, Q18 |
| Communication + reminders | Q3, Q4, Q8, Q9, Q10, Q11, Engagement Agent |
| Reporting / documentation drafting | Q2, Q5, Q7, C6, every Scorecard item |
The AI Audit becomes a standardised report with a custom layer. About half the roadmap is pre-built from the Catalogue; the business-specific layer is the other 30–50%.
| intelligence.remap.ai | clients.remap.ai | |
|---|---|---|
| Answers | What should we do, and why | What is being done |
| Contains | Assessment and both audit layers; Catalogue browser and team vote (beside it for v1); proposed six-month sequence | Delivery pathway mirrored from Jira; status against the Roadmap record; links to dashboard and stack |
| Users | Prospects first, then clients | Clients only |
| Owner | Marketing | Delivery & Client Operations |
Shared identity through one auth provider. A single Roadmap record that both apps read and write: Intelligence creates proposed items; acceptance converts an item to a Jira epic; clients.remap.ai shows status against it.
Gating: the proposal page shows only the Month 1 picks, a named Month 2 and Month 3 project, and one line saying more was identified. The full Catalogue and up to ten custom items appear only after discovery, inside the signed-in report.
Content is not a separate marketing activity beside delivery. It is the mechanism by which delivery grows the library, and the library grows the funnel.
Scorecard measure: Catalogue items self-selected by clients through the portal each month. The cleanest leading indicator of retention the model produces.
Second entry point: open-source tool review. Technology evaluates a candidate; R&D runs the two-step (MVP fast, then deploy internally properly); if it survives internal use, it enters the Catalogue as a Foundation or Core Build.
The BA capture template is the cheapest item in the whole plan and the highest leverage. Without it, the library never fills.
Nothing here changes seat pricing or the six-month standard engagement. It adds structure around what the seat delivers.
| Element | Position |
|---|---|
| Interface | Client contracts directly with Anthropic or Microsoft. REMAP never bills for the interface. |
| Private AI environment | Included while a seat is active. Shown on every invoice as a visible line, then discounted in full as an included service, so the value is seen monthly and there is no surprise at exit. |
| Intercompany | Services entity pays the SaaS entity a monthly rate per active environment. |
| Pass-through | New tools, heavier model tiers or large data builds after Month 1 are one-line additions at management margin, agreed before build. |
| Exit | Retain (pay the environment fee) · Transfer (one-time migration to client AWS) · Export (workflows as JSON, memory as database export) · Decommission. The library stays licensed; the configured instance is theirs. |
| Entry products | AI Audit and the two-day Build Event, both at approximately US$2,500 with regional pricing. Both earn a discount against the first AIS month and include the AI Audit in Month 1. |
| Function or role | Owns |
|---|---|
| AI Architecture & Technology | Reference architecture, Docker composition, workflow library, Catalogue content and versioning, open-source tool review, security and residency standards, recommendation rules audit |
| Delivery & Client Operations | Deployment runbook, spin-up, Kickoff Sprint delivery, clients.remap.ai, exit execution |
| Business Analyst (Delivery) | Capturing what Specialists build and feeding the library; intelligence.remap.ai data and rules table upkeep |
| AI Architect | Month 3 and quarterly Strategy Reviews |
| AI Specialist in seat | Day-to-day operation, Catalogue delivery, skill provisioning, monthly re-presentation of the Catalogue |
| Marketing & Demand Generation | intelligence.remap.ai as top-of-funnel, recommendation rules content, public Catalogue, monthly deployment content |
| Revenue & Partnerships | Catalogue pricing, pass-through approvals, entry product discounts |
| Finance & Legal | Intercompany rate, invoice presentation, exit fee schedule, Order Form and engagement agreement amendments |
| # | Workstream | Owner | First deliverables |
|---|---|---|---|
| 1 | Standard composition and compliance | Technology | Parameterised Docker composition (n8n, LiteLLM, Infisical, Wiki.js + Postgres, Metabase; region as a parameter) · validate non-Anthropic routing on the Grade 2 client · MCP surface · Catalogue v1 Foundations confirmed against what spins up today |
| 2 | Runbooks and the client portal | Delivery & Client Ops | Day 0–5 deployment runbook · one runbook per Kickoff item and roadmap project · Roadmap record schema and item-to-Jira-epic workflow · clients.remap.ai pathway view · BA capture template |
| 3 | Assessment, Catalogue publishing and content | Marketing | Vertical inference for two verticals, rules-table recommendation, gated proposal page, signed-in report · Catalogue page and vote · public Catalogue on remap.ai; monthly deployment piece per client |
| 4 | Commercial paper | Finance & Legal with Revenue | Private AI line-item value and intercompany rate · exit fee schedule · Order Form language (Kickoff commitment, Months 2–6 sequence, AIS admin seat) · entry product discount mechanics |
The first six unblock the build; the rest can run in parallel.
| # | Decision | Owner |
|---|---|---|
| 1 | Catalogue v1 items approved (anything you wouldn't put in front of a client this quarter comes out) | Leadership |
| 2 | Two launch verticals confirmed: Real Estate (sales + PM) and General SMB | Leadership |
| 3 | Kickoff Sprint counts: as drafted, or trim Quick Wins to 1 for a strict cap of four | Leadership |
| 4 | Q19 suburb feed approved as the Real Estate industry starter (factual fields only; council register as source) | Leadership |
| 5 | Where the Roadmap record lives (Postgres table or otherwise) | Leadership |
| 6 | Team vote surface: Catalogue page plus a form writing to the same table, beside intelligence.remap.ai for v1 | Leadership |
| 7 | Private AI line-item value per month and the intercompany rate | Finance & Legal |
| 8 | Exit transfer fee and retained-hosting fee | Finance & Legal |
| 9 | Whether the AIS admin seat goes in the Order Form or the engagement agreement | Finance & Legal with Revenue |
| 10 | Owners and dates for each workstream's first deliverable | Leadership |
When the BOS review closes, this supplement becomes:
No new seats, no new meetings.
REMAP Business Operating System · Supplement: AIS Value Model · Draft v0.2 · Role titles only · Blended in after the current BOS review