Cash Forecast Tutorial

How to Build Your Cash Forecast Agent

Building the agent isn't a development project — it's four decisions. Here's exactly how it works.

Pick the platform you already use, add your open invoices and how those customers have paid, and connect our MCP. After that, the agent returns a week-by-week cash range, the chance of falling below your minimum, and the customers who move the tight week.

Four decisions

Step 1: Find an Agent Builder

Pick the platform you already use, or the one your team is standardizing on. Common choices:

You're not locked in — the same underlying intelligence connects to whichever builder your organization runs on.

Step 2: Add a Data Source

Every agent builder needs to know where your cash data lives. There are two ways to connect it:

CSV / Excel

Export the spreadsheet your accounting software already produces — open invoices with customer, amount and due date, plus how those invoices were paid — and upload it. Add an opening balance and the outflows you want included. If Tally, your ERP, or Excel can export a spreadsheet, you already have the core of what the agent needs.

MCP

Connect through a standard protocol built for secure, structured access. The agent queries a calculated cash forecast instead of a file that went stale the moment you exported it.

What Is MCP, and How Does It Work?

MCP (Model Context Protocol) is an open standard that lets an AI agent connect directly to a live source — instead of you exporting a file and hoping it's still accurate by the time someone reads it.

In plain terms:

  • Your agent builder (Copilot Studio, Bedrock, Gemini Enterprise, Agentforce, Cursor, etc.) acts as the client.
  • Our MCP acts as the server — it exposes the cash forecast (week-by-week P10 / P50 / P90, the chance of falling below your minimum, and the drivers of a tight week) in a structured, secure format the agent can query directly.
  • When your agent needs an answer — "which week are we most likely to fall below the minimum, and which customers drive it?" — it asks the MCP server directly, instead of working off a static file someone uploaded weeks ago.
  • Because it's a standard protocol, not a custom integration, it works the same way across every agent builder — connect once per platform, and every query after that pulls a current, calculated forecast.

The result: your agent isn't reading a due-date snapshot. It's asking for a calculated range every time — which is what keeps the tight week current and the access controlled.

Step 3: Choose MCP Where Possible

CSV upload works, but MCP is the preferable path — it's the secure, standardized way for an agent to access a calculated forecast, rather than a static file that goes stale the moment you export it.

Step 4: Connect Our MCP

This is the step that turns a generic agent builder into a Cash Forecast agent. Connecting our MCP is what makes the weeks behind every answer calculated — not a guess dressed up as a forecast, a range you can re-run.

Pick your platform below. Connecting our MCP is the same idea on each one — you add it once, then sign in with your Saifs AI account.

Steps to Enable MCP on AI Platforms

Open Claude and reveal the main menu

Launch Claude Desktop. In the top-left corner, click the three-dot menu to open the app's main menu.

Go to Settings

From the three-dot menu, hover over File. Click Settings (or press Ctrl + Comma).

Open the Developer config editor

In Settings, select the Developer tab. Click Edit Config at the bottom of the panel. A File Explorer window will open at the Claude config folder. Click to select claude_desktop_config.json.

Paste the Cash Forecast MCP

With the config file open, add this server — not the generic Saifs AI MCP. Name it saifs-cash-forecast. This one is Cash Forecast only. Then restart Claude.

Launch Cursor

Open the Cursor application on your desktop. On the home screen, choose Open project, Clone repo, or Connect via SSH to begin working on a project.

Settings → Tools & Integrations

Click the gear icon (top right) or go to File → Settings. In the left panel, click Tools & Integrations. Scroll to MCP Tools and click Add Custom MCP.

Add the Cash Forecast MCP

After clicking Add Custom MCP, Cursor opens the MCP config file. Add this server — not the generic Saifs AI MCP. Name it saifs-cash-forecast. This one is Cash Forecast only. Click Connect next to the new MCP server under MCP Tools to start the integration.

Complete authentication

When prompted, select Open so Cursor can process the URI and start the login flow. Sign in with your Saifs AI account, then click Open to finish the connection and confirm the toggle is ON.

Open ChatGPT Settings

From the ChatGPT home page, go to Settings.

Open Connectors

Click Connectors in the settings menu. Click Create in the browse connectors section.

Add the Cash Forecast MCP

Enter this server — not the generic Saifs AI MCP. Name it saifs-cash-forecast. This one is Cash Forecast only. Then click Save.

Save and start asking

Click Save. You can now ask which week cash gets tight, and which customers move it, in ChatGPT.

Install an MCP-compatible client

Open the VS Code marketplace and install an MCP-compatible client.

Open the command palette

Open the command palette, type MCP: add, and select MCP: Add Server to start the wizard. Search MCP: Add Server and choose HTTP (HTTP or Server-Sent Events) to connect to a remote MCP server.

Enter the Cash Forecast MCP URL

Input this server URL — not the generic Saifs AI MCP. Name it saifs-cash-forecast. This one is Cash Forecast only. Press Enter to confirm. When prompted, choose Global (all workspaces) or Workspace (this workspace only).

Allow authentication

Allow the server to authenticate so VS Code can talk to the Cash Forecast MCP. The MCP server should show as Running. That confirms the Cash Forecast agent is connected in VS Code.

Walkthrough

What Our AI Agent Can Actually Do — A Walkthrough

Once connected, our AI Agent isn't answering from a static due-date sheet — it's querying a calculated forecast every time. Here's what that looks like in practice.

Say you're the treasury manager at a distributor with a few hundred open invoices. You export what you already have: open invoices (customer, amount, due date), how those customers paid before, the opening balance, and the outflows you know — payables, payroll, and the statutory dates. Every time you send a new batch, the MCP works through it:

  1. Each customer's delay is read from history. Paid date minus due date. Invoices still open stay in the curve, because they have already been late at least this long.
  2. A short history borrows the segment. Customers with only a few paid invoices use their segment's curve, weighted by how much history they have. Split payments stay as a chance of paying in two parts.
  3. Receipts are spread across the weeks. An overdue invoice is not dumped into this week. It uses the chance it lands in each remaining week, given how late it already is. Known outflows sit on the dates you supplied.
  4. You ask our AI Agent a plain question. In Cursor, Copilot Studio, or wherever the agent lives, you type: "Forecast cash for the next 13 weeks." It queries the MCP and answers: "Closing cash is tightest in week 7. The chance of falling below your minimum is highest that week. These customers move it."
  5. Our AI Agent explains itself when asked. You follow up: "Why is week 7 the tight one?" It doesn't deflect. It names the receipts that are still likely late and the outflows that fall in that week — payroll, a statutory date, a supplier bill.
  6. You try a what-if, and your team decides. Ask: "What if this customer pays on time?" The comparison uses the same paths, so you see the difference that one change makes. The agent never tells you to borrow, and it never computes interest.

For India, you can include the statutory dates you already pay — GST on the 20th, TDS on the 7th, PF and ESI on the 15th, and advance tax — as items you supply. The agent places them on those dates. It does not compute interest, and it never connects to your bank.

That's the shift: from every invoice sitting on its due date, to a 13-week range — so you know which week needs a decision.

"Forecast cash for the next 13 weeks"

"Which week are we most likely to fall below the minimum?"

"What if this customer pays on time?"

Why Connect Our MCP

A range you can re-run

Same input, same weeks, every time. The forecast is calculated by published methods, not generated as a plausible-sounding guess.

Privacy

Full SOC 2 compliance. You send exports, the request is not stored, and the agent never connects to your bank. No data leaves your premises.

Less token usage

Purpose-built models time the invoices instead of an LLM re-reading your entire export on every query — cheap enough to run whenever the book changes.

Battle-tested

Kaplan-Meier on payment delay (1958), Monte Carlo paths (Metropolis-Ulam, 1949), and a Trigg tracking signal (1964) against the previous forecast. Not a new algorithm hoping it works on your book.

Discover Our MCP on Leading AI Platforms

Microsoft

Microsoft

Forecast cash week by week from how customers pay.

In Review
Amazon

Amazon

Forecast cash week by week from how customers pay.

Coming Soon
Google

Google

Forecast cash week by week from how customers pay.

Coming Soon
Github

Github

Forecast cash week by week from how customers pay.

Coming Soon
Salesforce

Salesforce

Forecast cash week by week from how customers pay.

Coming Soon
Glama

Glama

Forecast cash week by week from how customers pay.

Coming Soon
AI Text
  • Coming Soon
Download the app
playstore-button applestore-button windows-button
Company
Join Us