Skip to content
PLM Pundits
PerspectiveDraft

Fix the thread before the tools

Most operations don't have a technology problem. They have a thread problem. Why disconnected product data, process and people undermine every new platform, and where to start instead.

Ankush Manhotra22 October 20262 min read

Draft: awaiting partner review before publishing. Not indexed by search engines.

When operations struggle, the instinct is to buy something: a new PLM, a new ERP module, an AI pilot. Sometimes that is the right answer. More often the root cause sits underneath the tools, in the thread that should connect product data, process and people from first sketch to final shipment.

What a broken thread looks like

The symptoms differ by industry, but the pattern is the same:

  • Fragmented product truth. Specifications, BOMs and formulations sit in spreadsheets, email and legacy systems. Every team trusts a different version.
  • Process debt. Years of workarounds and customisation slow every launch, change and audit.
  • Automation without architecture. AI pilots stall because the data, processes and systems underneath were never designed to connect.

In a shipyard it shows up as drawings reaching the yard out of date. In apparel it is factories working from old tech packs. In pharma it is days spent assembling evidence for an inspection. The tools are different. The thread problem is the same.

Why new tools rarely fix it alone

A new platform inherits whatever it is given. Migrate inconsistent data and you move the inconsistency. Configure a system to undocumented processes and you encode today's workarounds. Add an AI agent on top and it automates the confusion faster.

Where to start instead

  1. Baseline the thread. Assess process, data and technology maturity, and map how a product, change or order actually flows. Measure where time and quality are lost.
  2. Engineer the processes. Define target processes from L1 value chain to L4 work instructions, agreed by business and IT.
  3. Choose tools that fit. Select platforms against those processes, on fit, usability and total cost of ownership, not on the demo.
  4. Then automate. Put agents to work on the clean, connected thread, where they have the data and rules they need.

The good news is that fixing the thread is not a multi-year detour. A focused diagnostic shows where the thread breaks and what it is worth to repair, and gives every later technology decision a factual foundation.

Related

Keep reading.

Perspective · FMCGDraft

Recipe to retail shelf: where FMCG margin leaks, and how to stop it

Food and snack brands rarely lose margin in the kitchen. They lose it in the gaps between recipe sheets, the plant, distributors and finance. Four leaks to look for, and why closing them starts with one data model.

25 October 2026 · 3 min read

Perspective · AutomotiveDraft

Process engineering from L1 to L4, explained

What the four levels of process detail mean, why implementations stall without L3 and L4, and how the same model guides PLM configuration and AI agent design.

15 October 2026 · 2 min read

Start a conversation

Discuss this with a partner.

Tell us how this applies to your operations and a partner will come back with a working view of what to do first.