The Margin Roundtable · Tuesday 19 January 2027, London · finance, procurement and commercial leaders, one room, no slides Ask for a seat →

Find where your margin is won and lost.

Then know if the next deal makes money before you commit.

Most clients start with spend analytics on the data they already have. Spendkey then gets back what suppliers and customers owe you, and gives every new purchase order, bid and contract a verdict before it is signed. For finance, procurement and commercial teams, and for private equity across a portfolio.

Start a four-week sprint

We find where margin is lost, get back what you are owed, and protect every new deal.

Three layers on the same data. Start with the one you need, and add the next when you are ready.

Under all three layers
  • AI agentsfind, draft and chase
  • Cassiananswers in plain English
  • Clean master dataone view of every supplier and customer
  • Audit trailof every decision
How the platform works

Finance, procurement and commercial teams at these companies already work this way.

Procurement and Supply Chain Awards 2024Start-Up Award winner, AI in procurementProcurement and Supply Chain Awards, 2024
Procurement and Supply Chain Awards 2025Runner-up, AI in procurementProcurement and Supply Chain Awards, 2025
The Hackett GroupUpper quartile for customer valueThe Hackett Group, 2026
Everest GroupInnovation Watch: InfluencerEverest Group, 2026

Why not just do this in ChatGPT or Claude?

Because a margin decision has to be right, has to be checked, and has to happen inside your approvals.

Tested code does every sum.

The language model plans the work and explains the answer. Tested code does the arithmetic, and any figure that cannot be traced to your own data is removed before you see it.

Six years of checked answers.

The agents follow a playbook our analysts built by hand since 2020: which category a line belongs in, which parent company a supplier belongs to, and what a correct contract test looks like.

It works inside your approvals.

Your rules decide which deals go straight through and which reach a named person, and every decision is kept with its effect on EBITDA.

Your own Copilot, ChatGPT or Claude can use the same cleaned data through our MCP server.How the AI works

ISO 27001SOC 1 and SOC 2 Type 2GDPRHosted on Microsoft AzureSecurity details for your IT team

Start with a four-week sprint on your own data.

Share one extract. Within four weeks you see your spend, your savings and the money you are owed, ranked, with the evidence behind every figure.

Start a four-week sprint

Next Margin RoundtableThursday 28 January 2027, breakfast, central LondonAsk for a seat →

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.