A single, secure AI that works across Innova's real systems and shows up where each team already works — rolled out one proven step at a time, with no lock-in to any AI vendor.
The proposal
Stop renting scattered chatbots. Stand up one governed layer — pointed at our biggest cost: work-hours.
A dedicated Innova instance — our own secure environment — that reads our systems together, obeys each user's permissions, and appears inside the tools we already use. Start with two teams, prove it, then scale.
One layer, not four apps
Replaces the sprawl of personal AI subscriptions across four providers with one governed layer.
Secure & governed
Company sign-on, enterprise data terms — nothing trains on our data — per-user permissions, full audit trail.
Cross-system answers
Reads Salesforce, NetSuite, Drive and more together — something an off-the-shelf chatbot cannot do.
Aimed at payroll
The foundation for automations that absorb repeatable work-hours — the biggest line we pay for.
Six seats, two teams, a visible first build — and every later stage stands on the same backbone.
This page was itself researched, written, and built by AI under one person's review — the exact working model this brief proposes.
01
The cost that actually matters
Not software — payroll. Every task at Innova still runs on people's hours. And AI is already in the building, just not working for the company.
4
AI providers, paid for personally
separate accounts, separate bills, no shared standards
0
Oversight or audit trail
no company sign-on, no permissions — chats and data sit in personal accounts
1 at a time
What today's tools can see
a cross-system answer — Salesforce + NetSuite + Drive — is still stitched together by hand
Hours
The real spend
every new workload means more hours or another hire
And the know-how that makes those hours productive lives in people's heads and personal chat histories. When someone leaves, it leaves with them.
Software spend is a rounding error. Work-hours are the bill.
02
What we get from one layer
Six wins, one decision — everything below comes from standing up a single governed AI layer.
Consolidation
One governed layer replaces the sprawl of personal AI subscriptions across four providers.
Secure & governed
Company sign-on, enterprise data terms — no training on our data — per-user permissions, full audit trail. Data stops leaving with people.
Cross-system answers
Reads Salesforce, NetSuite, Drive and more together — something an off-the-shelf chatbot cannot do.
Knowledge that stays
Captures institutional know-how so it survives staff changes.
No lock-in
Provider-agnostic — we can swap the AI model behind it at any time.
Efficiency
The foundation for automations that let the same team do more, faster, and more consistently.
"Why can't an off-the-shelf chatbot do this?"
Off-the-shelf tools see one thing at a time and answer to nobody. The layer reads our systems together and answers under our rules — sign-on, permissions, audit trail.
03
Where it lives
A dedicated Innova instance — our own secure environment, with a private database, private hosting, and our own address. One governed brain, reached through many front doors.
Today — scattered chatbots
A person with a question
↓
A personal chatbot
sees nothing of ours · answers get copied back in by hand
↓
Our systems, untouched
Salesforce, NetSuite, Drive — each one checked separately, by a person
The instance — one governed brain
The same person, in Slack or their AI seat
↓
Innova AI — acting as that user
sees only what that person is already allowed to see
↓
Salesforce · NetSuite · Drive — together
one answer drawn from all of them, logged and auditable
Many front doors, one brain — it appears inside the tools we already use, not as another app to log into.
The governance rule that makes it safe
The instance runs as each individual user. It can never see more than the person asking — permissions are enforced by design, not by policy documents.
04
The plan — start small, prove it, then scale
No big-bang rollout. Each stage has to earn the next one.
Stage
Who
What it proves
Stage 1 · now
Design + Web Dev — about 6 seats on Claude Team
A visible, shippable result — under all our policies from day one
Stage 2
More teams, more connected systems
Proven workflows turn into automations
Ultimate
The whole company
One AI interface, in every department — same foundation, nothing rebuilt
The policies apply from the first day, not the last: company security, user profiles & permissions, and the AI acting only as each user. Design and Web Dev go first because they are unblocked today — and because their work is the kind directors can see.
05
The first thing we build
Launch a new product's website through chat — cross-functional, governed, and it ships something directors can see.
One request, end to end
A new product is ready to launch
↓
Ask in chat
one request, in plain language, in Slack
↓
Design imagery
product visuals drafted in our brand, with the Design team
↓
Site assembled
pages built and filled with the approved content
↓
Routed to a person to review & publish
a human approves the final step — always
Not a demo we stage — the first deliverable of Stage 1, started the day it's approved.
06
The vision — a global AI interface for Innova
Same foundation as Stage 1 — the backbone we stand up now carries straight through.
Connected to everything
Through standard AI connectors it can work with any AI platform — and with any part of our tech stack that can accept it.
Knowledge in every department
Company-wide knowledge bases, so every team's know-how is captured, searchable, and stays with Innova.
Do more with less
Automations absorb repeatable work — we reduce headcount where the work is absorbed, and raise the quality and consistency of results.
We don't buy this future separately — Stage 1 is its first brick.
07
The money
Seat subscriptions — small and flat. The payoff is measured against payroll, not against software spend.
$180/mo
All of Stage 1
about six seats on Claude Team, $25–30 per user per month
~$2.2k/yr
A full year of Stage 1
a fraction of a single salary
Usage-based
The instance itself
the cross-system layer is billed by use — a separate, small line
Item
Plan
Indicative cost
Design + Web Dev seats (~6 users)
Claude Team
~$25–30 / user / mo
Stage 1 monthly
≈ 6 seats
≈ $180 / mo (~$2.2k / yr)
Org-wide, later
Claude Enterprise — adds company sign-on & audit
custom · seat minimum
The instance (cross-system layer)
AI provider usage
billed by use · separate line
If the automations absorb even a small share of one role's repeatable work, the layer pays for itself many times over — and Stage 1 is how we find out, cheaply.
The ask: approve Stage 1 — six Claude Team seats for Design + Web Dev.
Governed from day one, no lock-in to any vendor — and the first build starts immediately.
08
Appendix — where the numbers come from
Sources, so any number can be checked.
The numbers
Seat pricing (~$25–30 / user / mo, Claude Team): published plan pricing, July 2026.
Stage 1 total (≈6 seats → ≈$180/mo, ~$2.2k/yr): seat count = the current Design + Web Dev headcount.
Org-wide (Claude Enterprise — company sign-on, user provisioning, audit): custom pricing with a seat minimum — priced only when we get there.
The instance (cross-system layer): AI provider usage pricing — billed by use, kept as its own line so the cost stays visible.
The working model (AI builds, a person reviews): already live at Innova — the brochure viewer runs in production on this model, and this page was written and built the same way.
Innova AI — director briefing, July 2026. Written and built with AI under human review — the same governed working model it proposes.
AI · Director briefing