Simple Projects
Project delivery that starts at the signed scope of work.
For PMs at MSPs, agencies, AEC firms, and consulting practices delivering against signed SOWs.
The plan and the contract drift apart in week two. Simple Projects reads the signed statement of work, drafts phases, tasks, dependencies, and the exclusions the document rules out, and asks a delivery lead to approve it before anything is committed. Tasks the model writes carry a confidence score and the sentence from the SOW that justifies them, so a lead accepts or rejects on evidence.
How to get Simple Projects
Available on requestRunning in production and taking new customers. There is no self-serve trial or purchase on this one yet, so access is set up with us directly.
Where this stands
- Availability
- Available on request
What it costs
$299 per month
Growth plan
Flat monthly price, not per seat.
Published pricing. There is no self-serve checkout for this product yet.
Routes that are not open
- Direct trial with Simple Intelligence Not open yet
- Direct subscription from Simple Intelligence Not open yet
- Microsoft Marketplace Not published yet
- Sponsored subscription for member companies Not open yet
The plan and the contract stop agreeing in week two.
Somebody read the statement of work, opened a project tool, and retyped an approximation of it. The exclusions did not make the trip, because there is no field for them. Six weeks later a client asks for something that the contract rules out, and the only person who knows that is the one who read the document in March. The argument that follows is not about the work. It is about what was agreed, which nobody can find.
Start the plan from the document that was signed.
Simple Projects reads the statement of work and proposes the project: phases, tasks, the dependencies between them, and the exclusions the document rules out. Each proposed task carries the sentence from the SOW that justifies it and a confidence value, so a delivery lead accepts or rejects on evidence rather than on faith in a model. The scope stays attached to the plan instead of sitting in a PDF nobody reopens.
Nothing is committed because a model suggested it.
The parse and the commit are two separate operations. A lead can read the whole proposed plan before a single row exists in the database, and the resource-leveling pass writes proposals for review rather than moving anyone. The point is not that the model is untrustworthy. It is that the person who signs the delivery is the person who should decide what the plan says.
What works today, and what does not.
Simple Projects has been running in production on Azure since May 2026, with its own pricing, legal, and support pages. What is unfinished is mostly integration surface: connectors whose vendor registrations are not done, and a Marketplace listing that is drafted but not published.
Working today
- Statement of work to a drafted plan with phases, tasks, dependencies, and the exclusions the document rules out, from PDF, DOCX, Markdown, or plain text
- A preview that parses the document without persisting anything, so the plan can be read before it is committed
- Board, list, calendar, and Gantt views, with milestones and acceptance criteria on the project
- Portfolio risk scoring, overdue detection, and a red-amber-green report across every active project
- A workload heatmap with an AI rebalance suggestion, and a portfolio-wide leveling pass that writes proposals for a human to accept
- Vendor organisations, vendor scorecards, and a tokenized read-only burn portal that shows hours against budget and never shows rates
- Assigned-only visibility, per-tenant isolation, Entra ID sign-in, and a Microsoft Teams tab with Teams single sign-on
- An MCP server with five tools, and a versioned JSON export so a tenant can pull projects and tasks into its own reporting
- Hours rolled up from Simple Time, and signed scope imported from Simple Growth
Not yet
- Migration import from Asana, Monday.com, Trello, ClickUp, and Notion is implemented in code, but each vendor needs a developer-app registration before its Connect button goes live. Until those credentials are in place the product shows that source as pending rather than pretending it is connectable.
- Import is one-way migration. Nothing is written back to the source tool, and there is no Jira connector.
- There is no Microsoft Planner, To Do, or Outlook task synchronization. The Microsoft surfaces are sign-in, the Teams tab, and product email sent through Microsoft Graph.
- Scope import reads a text layer. A scanned SOW with no text in it is rejected rather than guessed at, and extraction is capped at a 10 MB file.
- The Teams surface is a tab. There is no bot and no messaging extension.
- The Azure Marketplace offer is drafted in the repository but not published, so this cannot yet be bought through a Microsoft agreement or drawn against a Microsoft consumption commitment.
- Self-serve signup is not the path this site offers. New workspaces are arranged with us while the product moves off the no-cost tier described in the questions below.
Verified against the product on 2026-08-20.
The projects this was built for.
Turning a signed SOW into a plan the same week
Upload the executed document, read the proposed phases and tasks with the quoted sentence behind each one, cut what does not belong, and commit. The exclusions the document rules out are captured as records on the project rather than as a paragraph somebody remembers.
Finding the slipping project before the status meeting
Risk scoring runs across the portfolio and flags projects drifting on scope, schedule, or ownership, with a short explanation of why each was flagged. The report is a standing view rather than a thing somebody assembles the night before.
Showing a subcontractor their burn without opening the workspace
A vendor organisation is attached to a project, its contacts see that project and nothing else, and a tokenized read-only portal shows hours consumed against the budget agreed with them. Rates and dollar amounts are not in that view by construction.
Reporting delivery margin without a spreadsheet in the middle
Hours logged in Simple Time roll up against the project they were worked on, and the whole tenant exports as a versioned JSON document that a reporting stack can read directly instead of scraping screens.
From a signed document to a plan someone approved.
Sign in with your work account
Simple Projects is a multi-tenant Entra ID application, so a customer signs in with the Microsoft 365 account they already have and the tenant id on the token scopes every later query. Google sign-in and an email link are supported for people outside a Microsoft tenant.
Bring the scope in
Upload the statement of work as PDF, DOCX, Markdown, or text, or import a signed scope straight from Simple Growth if the deal was closed there. Extraction runs on the server and rejects a file it cannot read rather than sending an empty document to the model.
Read the plan before it becomes a project
The parse step returns the proposed plan without persisting it: name, dates, phases, tasks, dependencies, exclusions, each task carrying the source sentence and a confidence value. Nothing exists in the database until somebody commits it.
Work it in the view that suits the project
Board, list, calendar, and Gantt are four views of the same tasks, with milestones and acceptance criteria attached to the project and vendor scope attached where subcontractors are involved.
Let the portfolio watch itself
Scheduled passes score delivery risk, surface overdue work, propose a flatter distribution of load across the team, and send burn digests to the people who asked for them. Every one of those produces a proposal or a notification, not a silent change.
What it does between signature and the final invoice.
Planning that cites the contract
The model is asked for a plan and for the justification of each line: which sentence in the SOW produced this task. Dependencies between tasks and the exclusions the document rules out are persisted in the same transaction as the project itself, so the scope boundary is a record you can query rather than a memory somebody has.
Risk scoring you can argue with
Projects are scored on drift in scope, schedule, and ownership, and each flag carries the reason it was raised. A delivery lead can disagree with a flag, which is the point. A number with no explanation behind it gets ignored by the second week.
Resource leveling that proposes and stops
The portfolio pass scans projected load per person per week, finds the overloaded weeks, and writes a proposed set of moves that flattens the curve. The arithmetic is deterministic and the model writes only the rationale, so a bad proposal is a code review problem rather than a mystery. Proposals are stored as proposals; nobody is reassigned by a scheduled job.
Vendors and clients get a narrow window
A vendor organisation attached to a project makes that project visible to its contacts and nothing else in the workspace. The external burn portal is reached through a token, is read only, and reports hours against budget without exposing rates or amounts.
An agent surface and an export, both scoped
The MCP server exposes five tools: parse a SOW, create a project from one, add a task, flag project risk, and list overdue work. Tenant scope is enforced inside the server rather than trusted from the caller, and calls are written to an audit table. The reporting export is a versioned JSON contract that applies the same tenant and visibility rules the screens do.
Where the work lives, and who can see it.
One database, tenant scoped in the query layer
Postgres through Prisma, every domain table carrying a tenant id, and the tenant filter composed into queries rather than left to each route to remember. Projects, tasks, and audit records are durable business records, so this is a persistent store.
Assigned-only visibility, with one deliberate exception
A member sees projects they own or are assigned tasks in. Administrators see the workspace. The one addition is vendor contacts, who see the projects their organisation is attached to, because otherwise a subcontractor invited to deliver work sees an empty list. The rule is applied server side in a shared query fragment, not by hiding rows in the interface.
Prompts stay inside the Azure boundary, and that is enforced
Model calls go to Azure AI Foundry through a gateway that refuses to construct a client unless the endpoint is an approved Foundry host. Foundry publishes some models at both an Anthropic-served and an Azure-served version, so the pinned model id decides whether the Azure statement is true; a test fails the build if a pin leaves the approved list.
Runs on Azure, in the SIG subscription
Azure Container Apps behind Front Door, secrets in Key Vault reached through a managed identity, and imported vendor tokens encrypted at rest with a key held in the same vault. No AWS, no Google Cloud, and no third-party platform-as-a-service in the path.
What it connects to, and what it does not.
A name here is a system this product is built to work with, not a promise that the connection is live in your tenant. Where a connector carries a status label, that label is the current state. Anything that needs its own vendor registration, its own consent, or its own approval says so, and nothing on this page can be taken as available until it is confirmed for your tenant.
Microsoft
Other ways in
Migration import (one-way, and each needs its vendor registration finished)
Across the portfolio
For your own systems
Reading data from an external system is not a way to sign in. Customer identity stays with Microsoft Entra ID, and no data source on this list authenticates a person.
Where each of these claims comes from.
What it costs.
- Up to 25 projects
- Up to 25 members
- 100 AI planning and risk runs a month
- Email support
- Unlimited projects and members
- Unlimited AI planning and risk runs
- The Microsoft Teams tab
- Priority support
- Everything in Growth
- Your master agreement, redlines welcome
- Single-tenant data isolation as an option
- A named contact and an agreed response time
How access works right now.
- How access starts
- A conversation first. We scope the evaluation with you and provision the tenant ourselves.
- Evaluation length
- Agreed when the evaluation is scoped, up to 45 days for a standard evaluation. A longer guided pilot is approved case by case.
- Payment details
- None collected to begin an evaluation.
How ICM informs Simple Projects
Simple Projects reads a statement of work and keeps a plan honest against it. Everything the agent concludes arrives as a proposal with a rationale, and an administrator decides which ones become real.
Built, and verified against the product code
-
Context in
The plan, scoped to the calling tenant
Every agent tool resolves the tenant from the server-side session. A tenant identifier passed in as an argument is ignored, so a caller cannot widen its own reach by asking nicely.
Gated
-
Retained or discarded
The plan hash and the rule versions
A completeness review pins the hash of the plan revision it read and the versions of the checklist rules it ran, so a suggestion can be traced back to the exact plan and the exact rule that produced it rather than to "the model, at some point".
Gated
-
What the agent proposes
Deltas, proposed and never applied
Reschedule, reassign, flag a risk, unblock a chain, split a task: each is written as a proposal carrying its reasoning and the audit row it came from. Resource levelling plans record the overload before and after so the trade is visible before anyone takes it.
Gated
-
Human review a person decides
An administrator applies or rejects
Apply and reject are the only routes from proposal to change, both restricted to administrators, both matching only proposals nobody has acted on yet. A second click on the same proposal fails rather than running twice.
Gated
-
Permitted effect
The change and the record settle together
The domain mutation and the status flip run in one transaction. A partial apply cannot leave a proposal sitting at "proposed" while its change has already landed, or the reverse.
Gated
-
Evidence recorded
An audit row for the proposal and for the click
The tool call that produced a proposal writes an audit row, and applying it writes another naming who applied it and when. Those two rows are what makes "who decided this" answerable from the settings screen.
Gated
-
Failure and refusal
Rejections carry a reason, proposals expire
A proposal can be accepted, rejected with a stated reason, superseded, or left to expire. Reservations against a workspace's AI budget stop counting after their time limit, so a call that was abandoned halfway never permanently consumes a cap.
Gated
These labels describe each step of this workflow on its own. They are not the availability of the product, which is the status shown at the top of this page.
This describes how the workflow is built. It is not a statement about availability, which is the status shown at the top of this page. ICM informs how context, review and evidence are designed here. It does not replace this product's database, its tenant isolation, its Entra identity, its audit tables, or its authorisation checks.
Methodology note
ICM was developed by Jake Van Clief and David McDermott. Simple Intelligence Group did not originate the methodology.
Simple Intelligence Group applies relevant ICM principles to context, provenance, traceability, human review, and workflow design.
Participation does not imply certification, endorsement, partnership, or an official ICM Architect designation.
What delivery leads ask first.
Does it sync with Microsoft Planner, or with Asana?
No. There is no Planner, To Do, or Outlook task synchronization, and nothing is written back to Asana or any other source tool. What exists is one-way migration import from Asana, Monday.com, Trello, ClickUp, and Notion, and each of those needs its vendor developer-app registration completed before it can be connected. The Microsoft surfaces that do work are Entra sign-in, the Teams tab, and product email through Graph.
Can the AI create a project without a person seeing it first?
In the product interface, no: the parse step shows you the whole plan before anything is persisted. Through the agent surface it is possible, because an agent can call the commit tool directly, and that call is written to the audit table with what it was given and what it produced. If you want that path closed for your workspace, say so during setup rather than assuming it is off.
What can it read from a statement of work?
PDF, DOCX, Markdown, and plain text, up to a 10 MB file, with extraction capped so that an enormous document is chunked rather than silently truncated. A scanned page with no text layer is rejected with an explanation instead of being guessed at.
Your product site still lists a plan that costs nothing. Which is right?
Both, for now. A workspace created before this change keeps the single-user plan at no cost that it signed up on, and nothing here removes it. Simple Intelligence no longer makes an open-ended no-cost plan the default offer for new customers, so this site arranges new workspaces through a conversation instead of pointing at that plan. If you are already on it, you are unaffected.
Can we buy it through our Microsoft agreement?
Not yet. The Azure Marketplace offer is drafted but not published, so there is no transactable listing and no Microsoft consumption commitment eligibility to claim. Purchase today is direct. If procurement through Microsoft is a condition rather than a preference, raise it early.
Who can see what, once several firms are working on one project?
A member sees what they own or are assigned to, administrators see the workspace, and contacts at a vendor organisation see the projects that organisation is attached to. The external burn portal is read only, reached through a token, and reports hours against budget without rates or amounts.
What happens to our data if we leave?
Projects and tasks export as a versioned JSON document scoped to your tenant, which is a stable contract rather than a screen scrape. Ask before you start if a particular export shape is a condition of buying, so it is settled while it is cheap to settle.
Nine questions worth answering before you replace the plan you have.
A short checklist for delivery leads. It runs in the page, stores nothing, and prints if you want to take it into a meeting.
Before you replace the plan you have.
None of these need a product to answer. If most of them are uncomfortable, the problem is the seam between the contract and the plan, and that is worth fixing whoever you buy from.
Nothing here is stored or sent. Tick boxes to work through it, or print the page and take it into the meeting.
What it connects to across the portfolio.
Simple Projects is the delivery layer of SimpleOS. A signed deal in Simple Growth becomes the plan here, the hours worked against that plan are captured in Simple Time, and the invoice is built from what the plan says was delivered.
How to start
-
Available on request Open
Running in production and taking new customers. There is no self-serve trial or purchase on this one yet, so access is set up with us directly.
Request access -
Direct trial with Simple Intelligence Not open yet
This product has no self-serve trial yet. Ask us and a person will set access up.
-
Direct subscription from Simple Intelligence Not open yet
There is no direct checkout for this product yet. Where a trial exists it converts to a paid subscription inside the product.
-
Microsoft Marketplace Not published yet
There is no transactable listing for this product yet, so it cannot be bought against a Microsoft agreement today.
-
Sponsored subscription for member companies Not open yet
The member sponsorship program is not open for this product yet.
Questions about evaluation length, extensions, or sponsored access are answered on the get started page.
Bring one signed SOW and see what it produces.
One real document tells you more than a demo does. If the plan it drafts is not worth correcting, that is a short conversation and no harm done.