Home / AI & Automation / Build & Operate
Fixed-scope build · Fixed price · Operate retainer optional

A specific AI outcome. Built. Then kept working.
Fixed price, written acceptance tests, on Microsoft 365 and Azure. Vendor-neutral on AI.

Build & Operate is the third pillar — how you put a specific AI use case into your business and keep it running afterwards. You see the deliverable and the price before you commit. Optional managed operations keep the build secure, monitored, and updated as models change. No MSP relationship required.

See sample builds → Not sure yet? Start with the Sprint
EVISENT // BUILD SPEC LIVE · ROLLING OUT
PCG Automation
CLAUDE API + AZURE · IN-TENANT
Discovery & scoping~2 weeks
Substrate + 14 AUTO sections3 weeks
4 HYBRID sections + UAT1 week
Portability (2nd contract type)1 week
Acceptance + handoverwritten
Scope · 18 sections 14 AUTO + 4 HYBRID
Across ~120 PCG reports/mo · at full portfolio rollout
Projected outcome
$360K / year projected
up to 3,600 hrs/yr senior PM time · 3h→30m per report
"AI adoption is easy. Operations is where most programs stall. Build & Operate exists to make sure what we build doesn't end up in that pile."
Where the work lives

We pick the right AI for the job. We build it inside Microsoft 365 and Azure.

Every Build runs inside Microsoft 365 and Azure. That's where we have nine years of depth, and where the security stack we've already deployed for your business naturally extends. The AI tooling on top — Claude, GPT, Gemini, open-source — is chosen per project for fit, not for partner margin.

We don't deliver AWS-native or GCP-native infrastructure work. If your use case genuinely needs that, we'll say so and recommend a different partner. We earn margin on the engagement, not on the licence — that recommendation costs us nothing.

The Build environment by default

  • Anthropic Claude — our default model family for reasoning-heavy work where behaviour matters more than stack alignment.
  • Microsoft 365 — Copilot, Copilot Studio, Power Automate, SharePoint, Teams, Graph integration.
  • Azure infrastructure — Azure-hosted AI services, secure compute and managed data services, all in your tenant.
  • Microsoft Purview — governance and DLP for AI workflows. Sensitivity labels deployed where the use case warrants it.
  • Endpoint-flexible — Windows or macOS, your team's choice.
Sample build catalogue

Fixed-scope builds. Custom briefs welcome.

The patterns below cover most of what businesses ask for. Custom work is quoted on the same engineering discipline — fixed scope, fixed price, written acceptance test before commitment. Most custom Builds fit one of these patterns within ~20%.

Entry-level

Single-flow automation

Toolchain: Power Automate or Claude API, one trigger, one outcome.
The smallest fixed-scope Build we ship — one specific workflow, one trigger, one outcome. Examples: triage-and-route incoming form submissions, auto-summarise weekly PDFs into a Teams channel, draft client follow-up emails from CRM signals. Useful when you want a quick proof point before committing to a larger Build.
From $4,000 + GST 1–2 weeks
Productivity

Copilot deployment + governance

Toolchain: Microsoft 365 Copilot, Microsoft Purview, Intune.
Tenant-wide M365 Copilot rollout with Purview controls aligned to Essential Eight, training, change management, and adoption telemetry. The "do Copilot properly" build.
From $7,500 + GST 2–3 weeks
Reasoning + drafting

Anthropic Claude integration

Toolchain: Anthropic Claude API, Azure-hosted sandbox.
Claude integrated into your business workflows — long-document analysis, regulatory review, complex drafting. Our default pick for reasoning-heavy use cases where model behaviour matters more than stack alignment. Pilot rollout where the use case fits.
From $8,500 + GST 3–5 weeks
Custom AI agent

Microsoft Copilot Studio agent

Toolchain: Microsoft Copilot Studio, M365 connectors, SharePoint, Teams.
Custom agent that lives inside your Microsoft tenant — client onboarding, support triage, document review, knowledge base access. Picked when M365-native integration matters more than model choice; surfaces through Teams or Outlook.
From $9,500 + GST 3–6 weeks
Workflow automation

Power Automate workflow suite

Toolchain: Power Automate, M365 connectors, Claude or Copilot AI steps where useful.
Multi-step business workflow automation — approvals, document routing, data sync, notifications. AI steps where they add value, deterministic logic where they don't.
From $6,500 + GST 2–4 weeks
Discovery + monitoring

Shadow-AI discovery automation

Toolchain: Microsoft Defender, Sentinel, Purview, custom dashboarding.
Continuous, automated discovery of AI tool usage across your environment — beyond the one-shot Sprint audit. Alerts on new tools, new data flows, policy violations. Outputs a live dashboard for your governance team.
From $11,500 + GST 4 weeks
Knowledge + search

AI knowledge base / search

Toolchain: Claude or Azure-hosted models, SharePoint, semantic search.
Searchable AI-powered knowledge base over your company documents. Asks questions in plain English, gets answers with citations back to source. Lives on your tenant; your data never leaves Azure.
From $14,000 + GST 5–8 weeks

DON'T SEE YOUR USE CASE? IF IT FITS THE M365/AZURE INFRASTRUCTURE, WE'LL SCOPE IT TO THE SAME ENGINEERING DISCIPLINE.

How the build phase works

Five steps. No endless changes. Acceptance against written criteria.

Every Build follows the same five-step delivery model. Each step has a defined gate to the next — you always know what's been done, what's next, and how to push back if the work drifts from what you signed.

1

Scoping session

Defines deliverable, integration points, acceptance test.

2

SoW

Single-page Statement of Work. Scope, price, timeline, acceptance criteria, vendor-neutral architecture statement, and the 6-month support retainer priced to the build's complexity.

3

Build sprints

Typically 2-week iterations. End-of-sprint demo. You see progress every fortnight; no end-of-project surprises.

4

Acceptance

You sign off against the criteria written in the SoW. No moving goalposts. No "endless changes" trap.

5

Handover

Documentation, source artefacts, prompts, configs handed to you. Optional handoff into Operate retainer.

The Operate retainer · from $200/mo per build, priced to complexity · 6-month minimum on new builds · 10% off with 12-month commit

Most AI builds stall in operations. Yours don't have to.

AI changes monthly. Vendors update models, deprecate APIs, ship features that quietly change behaviour. Without operations, your $14,000 Build slowly degrades into a fragile, unsupported piece of your tech stack — which is exactly where most internal AI projects end up.

The Operate retainer is the practice that keeps Builds working. Monitoring, security posture, vendor change management, quarterly optimisation, and full documentation maintained so you can always take the work elsewhere.

Monitoring

Telemetry on adoption, cost, errors, drift. Alerts on anomalies.

Security posture

Build stays aligned to your Bundle or industry standard. Purview controls maintained.

Vendor change management

Claude model releases, Copilot updates, API breaks — we test, patch, document.

Quarterly review

Usage data analysed, cost-vs-value reviewed, refinements proposed.

Continuity

Every Build carries a 6-month support retainer, priced to its complexity — automations need tuning, monitoring and small adjustments as they bed in. After that you can continue on retainer, or take everything to another partner with full documentation and source artefacts. The point isn't to make you stay; it's that you choose to.

Included with every Operate retainer · scaled to your build

The AI Program Dashboard. In your tenant. Board-ready by default.

Every Build under an Operate retainer gets a live dashboard inside your own intranet, scaled to the size of the program. A single automation gets a clean, simple view; a full portfolio gets cost, hours saved, ROI multiplier, per-automation usage, and the residual-risk register your auditor would actually accept. Either way, the question "is the AI paying off?" has the same answer at the exec table that it does in the engineering log.

AI Program Dashboard — portfolio view with monthly spend, hours saved, ROI multiplier, active users and automation cards. Open the live demo →
$
Cost & ROI
Every dollar of AI spend reconciled against the hours it returned. ROI multiplier on the front page.
👥
Adoption
Drill into any automation to see active users and run counts. Drill into any user to see their patterns.
⏱
Trends
Month-on-month spend, hours saved, ROI. The picture your CFO wants before the next budget cycle.
🛡
Risk register
Named risks, residual ratings, owners, last review dates, controls. Maintained continuously — not assembled the week before the board paper.
Open the live demo →

The demo is a sanitised version of the dashboard built for our flagship construction-PM client (see the case study). All figures and names are fictitious. Your version lives in your own SharePoint and pulls from your own data.

The portability checklist

Every Build is architected against a vendor-neutral checklist.

Most AI engagements quietly create lock-in: bespoke prompts in vendor-locked tooling, data pipelines that don't export cleanly, governance assuming one vendor's posture. We architect against this by default — the checklist below sits inside every SoW so the client knows what they're getting.

See the full /how-we-work commitment →

Portability checklist (per Build)

  • Your automations run in your tenant, on infrastructure you control. You get a complete user guide for the people who use it, and full build documentation at handover. The reusable engineering patterns behind it stay with us — that's the IP we earn from doing this hundreds of times — but the running system, the data, the documentation, and the option to leave are always yours.
  • Data flows documented end-to-end, with substitute-vendor paths noted where realistic.
  • Full documentation handover at the end of every engagement; source artefacts and prompts vest in you on creation.
  • Where vendor lock-in is unavoidable (e.g. deep Microsoft Graph integration), it's named explicitly in the SoW. You make the choice with eyes open.
  • We have the right to recommend Claude over Copilot, or the reverse — no margin structure tilts the answer.
  • Source artefacts (prompts, flows, agent configs) vest in you on creation — not in Evisent. We retain no licence beyond the engagement term.
Pricing summary

Two components. Both transparent.

Build is one-off, Operate is monthly. You can take just the Build (most do at first), or pair with Operate if you want the build kept current. Pricing is in every SoW; nothing hidden until invoice. Not sure it stacks up? Run your numbers — 30-second ROI calculator →

BUILD · ONE-OFF
From $4,000
+ GST · 1–8 weeks · scope-dependent · fixed-price per SoW
  • Scoping session
  • Single-page Statement of Work
  • 2-week build sprints with demos
  • Acceptance against written criteria
  • Full documentation + source artefacts handover
  • Vendor-neutral architecture (checklist in SoW)
  • 6-month support retainer (priced to complexity)
Book a scoping session →
OPERATE · MONTHLY PER BUILD
From $200/mo
+ GST · per build · monthly · cancel with 30 days' notice
  • Monitoring + telemetry alerts
  • Security posture maintenance
  • Vendor change management
  • Quarterly optimisation review
  • Full documentation kept current
  • Take-elsewhere optionality preserved
  • Management Dashboard included — governance, ROI, board-ready reporting
Discuss Operate retainer →
Who Build & Operate is for

Five typical buyers we work with.

Post-Sprint buyers

Did the AI Readiness Sprint, ready to act on something that surfaced in the board summary. Most common entry path.

Governance-minded buyers

Want board-level oversight of AI from day one. Every Build & Operate engagement under ongoing support includes the Management Dashboard — governance, ROI and quarterly board-ready reporting built in.

Vendor-neutral seekers

IT/Ops leaders explicitly wanting a partner who'll honestly say "Copilot isn't the right answer here" or "Claude is the better fit". Most channel-aligned partners can't say that.

Referrals with a use case

Sent by an accountant, consultant, or solicitor whose client has surfaced a specific AI need. Scoping session converts the referral.

Existing AI take-overs

Have an internal AI build that's struggling — built by someone who's left, vendor that's gone quiet, or simply drifting without maintenance. Operate-only engagements possible after a take-over audit.

Existing Evisent Managed IT clients

Want a build delivered by our AI & Automation team. Sold as a separate engagement — coordinated with the MSP team so it sits cleanly on top of the security stack we already run.

Common questions

What people ask before they book.

Most custom Builds fit one of the six patterns within ~20%. If yours genuinely doesn't, the scoping session defines the deliverable and writes the SoW from scratch. The engineering discipline is the same — fixed scope, fixed price, written acceptance test before commitment. We don't take blank-cheque engagements.

Yes. Source artefacts (prompts, flows, agent configurations, integration code) vest in you on creation — they live in your repository or tenant, not in ours. Documentation is handed over at the end of the build. Builds carry a 6-month support retainer to secure a successful outcome — after that, full documentation makes a move clean.

For the first 6 months, yes — every new Build carries a support retainer, priced to the build's complexity, because automations need tuning, monitoring and small adjustments while they bed in, and we won't ship something that quietly decays. After the 6-month minimum it's entirely your call: continue on retainer (most clients do), or take it fully in-house with complete documentation and source artefacts.

Not natively. Our delivery stack is Microsoft 365 and Azure — that's where nine years of depth lives. If your use case genuinely needs AWS-native or GCP-native infrastructure, we'll say so and recommend a different partner. We earn margin on the engagement, not on the licence, so that recommendation costs us nothing. Where the use case is cloud-agnostic, we'll build on Azure.

The smallest fixed-scope Build is the single-flow automation at from $4,000 — one trigger, one outcome, one written acceptance test. Below that scope we usually steer you to the AI Readiness Sprint ($4,950) instead, which produces a board-ready summary plus an Acceptable Use Policy and AI tooling and Purview review. For micro-scope one-off work we sometimes refer to specialist consultants we trust.

Three differences. First, we have no margin structure that biases AI vendor choice — so when Claude is the right answer instead of Copilot, we say so. Second, every Build inherits our security DNA — Purview controls, Essential Eight alignment, framework mapping for your industry regulator. Third, the engineering discipline — fixed scope, fixed price, written acceptance test, defined maintenance period — is unusual in the Microsoft Partner ecosystem, which tends toward time-and-materials engagements.

From idea to working system

Built. Tested. Then kept running.

Fixed-scope, fixed-price builds with written acceptance tests. Optional managed operations keep the build secure as models and vendors change. You see the deliverable and the price before you commit.

Book a build scoping call → Not sure yet? Start with the Sprint
★★★★★50 Google Reviews