Seco Tools (Sandvik Group)

The organisation

Seco Tools is a global manufacturer of precision cutting tools and tooling systems for the metalworking industry, and part of Sandvik, the Swedish high-tech engineering group serving manufacturing, mining and infrastructure customers worldwide. Seco Tools operates across dozens of countries with production, sales and engineering teams spanning Europe, the Americas and Asia, supplying components and expertise into demanding industrial supply chains that depend on precision, traceability and uptime.

The challenge

A legacy spreadsheet-based collaboration platform used for the company's project lifecycle no longer met security controls for working with external partners and had to be retired against a hard decommissioning date. Behind that single tool sat twelve years of project history, more than two thousand users across the UK, Germany and the United States, and four competing, non-authoritative requirement documents. Conventional delivery approaches pointed to a twelve-to-eighteen-month programme with a much larger team, a timeline the fixed decommissioning date could not absorb.

The solution

We committed to delivering nine independently owned platforms plus a three-wave data migration stream in four months with eight consultants, changing the engineering method to make that possible. Requirement owners were reconciled into one signed-off catalogue, and the Dataverse schema was locked as a binding contract between the build and migration teams. Sprint 0 ran a single loop of prototype, validate and convert: working applications were generated with AI assistance in hours, clicked through and corrected by requirement owners in the room, then converted into production Power Apps Code Apps under source control. Seven fortnightly sprints followed against the fixed gate date rather than the backlog, with specification documents written to be machine-readable so AI assisted agents could scaffold consistently against the locked schema, freeing consultants to focus on integration, security, performance and business logic. Each platform had a named internal owner paired one-to-one with a delivery stream lead. In parallel, twelve years of project history was profiled and split hot from cold, with active data loaded into Dataverse and historical data curated in Microsoft Fabric across three staged regional migration waves. Velocity was managed as an explicit discipline rather than treated as an outcome. Nine platforms across eight consultants means no platform ever gets a full team, so every stream was owned end to end by one engineer paired with a named business counterpart, and the shared spine, the locked schema, the common role and permission model, the status transition engine and the audit trail, was built once and reused rather than rebuilt nine times. Scope was cut against the gate date weekly rather than quarterly: each fortnightly sprint closed with a working, demonstrable increment on every platform and an explicit decision on what would ship at the gate and what would follow it. Progress was tracked platform by platform in the open, so each requirement owner could see the state of their own stream at any time instead of waiting for a programme report, which is what kept nine parallel conversations from collapsing into a single bottleneck. Migration rehearsal ran alongside build from the first sprint rather than at the end, so the final delta load was a repeat of a rehearsed procedure instead of a first attempt.

How we ran it

1. 01

   Discover

   Reconciled four competing requirement documents into a single signed-off catalogue and locked the Dataverse schema as a joint contract between build and migration teams.

2. 02

   Design

   Ran a Sprint 0 prototype-validate-convert loop, generating working applications with AI assistance and validating them live with each platform's named internal owner.

3. 03

   Develop

   Delivered seven fortnightly sprints against the fixed gate date, with machine-readable specifications letting AI assisted agents scaffold consistently while consultants focused on integration and security.

4. 04

   Deliver

   Migrated twelve years of history in three staged regional waves, rehearsing dry runs and reconciliation from the first sprint so the final delta load repeated a proven procedure rather than attempting one.

5. 05

   Hold the pace

   Ran velocity as a discipline: one engineer owning each platform end to end with a named business counterpart, a shared spine built once and reused across all nine, weekly scope decisions against the gate date, and per-platform progress visible to requirement owners rather than reported upward.

The result

Nine platforms and a parallel data migration stream were delivered by eight consultants inside the four-month window against a fixed decommissioning date, a delivery ratio the client's own teams had not previously seen achieved. That is close to one production platform per consultant, delivered while the same team carried a three-wave migration of twelve years of history. External collaboration moved inside the corporate security model through Entra External ID with role-scoped permissions, and twelve years of project history was preserved as a curated, read-only Fabric estate rather than dead weight in the operational database. The client was left with a reusable delivery model: small senior teams, AI assisted engineering, a locked schema and a gate date everyone builds towards.