How AI Will Transform Product Development and Speed to Market
The race to market is tighter than ever. Yet many product teams are stuck with outdated systems and siloed data that make serious AI work impossible. Reid Swanson, VP of sales, new markets at Centric Software, flags a "shadow stack" of legacy tools and off-system workarounds that drag timelines, cut margins, and derail AI. The fix starts with clean, structured, reliable data.
The blocker: data silos and the shadow stack
Plenty of companies have PLM or related platforms, yet teams still rely on spreadsheets, SharePoint folders, and local files to keep work moving. Those off-system flows erode transparency, cause version drift, and create data-quality issues. The result: longer cycles, higher costs, and AI that can't be trusted because inputs are messy.
AI needs consistent identifiers, structured attributes, and a single source of truth. Without that, you're feeding models incomplete or conflicting signals and getting noisy recommendations back.
Build the data base your AI needs
- Map the flow end-to-end: Document every handoff, tool, export, macro, and email thread. Quantify where delays and rework happen.
- Define a single product data model: SKUs, variants, attributes, components, suppliers, costs, and specs with clear IDs and naming rules.
- Clean and structure: Extract from PDFs, normalize attributes, de-duplicate records, reconcile BOMs, and archive stale items.
- Integrate: Use APIs and connectors for bi-directional sync and event-based updates. Eliminate manual rekeying.
- Govern: Assign data stewards, set SLAs, enforce audit trails and access controls, and version everything.
- Measure: Track defect rates, duplicate counts, time-to-truth, and coverage of the canonical model.
Where AI cuts weeks from launch
- Market intelligence at scale: Compare your assortment to competitors to spot gaps in pricing, discounting, and coverage. Insights arrive in hours instead of weeks, so you can adjust ranges, price ladders, and promotions faster.
- Design acceleration: Generate new styles, variations, and mark-ups in seconds. Produce photo-real images for marketing and early reviews across merchandising, sales, and development.
- From concept to components: Automatically create bills of materials and guide teams through each stage with clear next steps.
- Explainable decisions: See why the system suggested an option, then tune for cost, lead time, sustainability, or margin targets.
KPIs and views that keep teams aligned
- Design: Concept-to-approval days, adoption rate, sample iterations per style.
- Sourcing: Cost vs. target variance, supplier lead-time variability, PO confirmation cycle, on-time sample rate.
- Merchandising: Assortment coverage vs. plan (styles, price points, categories), competitor price/discount index.
- End-to-end: Plan → developed → adopted → prototyped → approved → sourced → sold, tracked in a single line of sight.
Use real-time dashboards, executive cockpits, and visual assortments so every team sees the same status and exceptions. Add threshold alerts and "aging" indicators to spotlight bottlenecks before they become delays.
The next 18-24 months
"AI capabilities are doubling every three to five months," says Reid Swanson. Multi-year forecasts age out quickly under that pace. As teams apply AI across marketing, merchandising, planning, product development, R&D, engineering, sourcing, and supply chain, launch cycles compress and learning loops speed up.
"It will be survival of the fittest." Those who adapt will ship the right product, at the right price, in the right place, with the right promotion-faster than their category can react.
A 30-60-90 day plan to get AI-ready
- 0-30 days: Audit your product data and tool stack. Pick 2-3 use cases (market intelligence, design generation, BOM automation). Set success metrics and baselines. Stand up data governance and ownership.
- 31-60 days: Clean and integrate priority data sets. Pilot with one category or region. Build real-time dashboards and exception alerts. Train the core team on prompts, review cycles, and decision rights.
- 61-90 days: Expand pilots, retire redundant tools, and productionize integrations. Add SSO and access controls. Lock in a monthly operating rhythm for model review, data quality checks, and KPI reporting.
Common traps to avoid
- Building models on top of messy data and hoping governance "catches up."
- Letting spreadsheets and email approvals persist alongside the platform.
- Running isolated pilots with no integration to core systems or KPIs.
- Skipping clear data ownership and SLAs for fixes.
- Underinvesting in training and change support, so teams revert to old habits.
- Chasing features instead of removing measured bottlenecks.
Upskill your team
The tech is moving fast, and your process has to keep up. If you need structured upskilling for product, sourcing, and merchandising teams, explore focused training paths here: AI courses by job.
Further reading
For broader context on productivity gains from generative AI, see this overview: McKinsey: The economic potential of generative AI.
Bottom line: if your data is noisy and scattered, AI will amplify the mess. Fix the base, pilot where you can prove cycle-time wins, and scale what works. That's how you speed launches and protect margin.
Your membership also unlocks:
