AI Isn't a Side Project - It Is the Business Strategy
Generative AI isn't something you bolt onto the edges of your company. Accenture's Chief Responsible AI Officer Arnab Chakraborty put it plainly: "The data and AI strategy is not a separate strategy, it is the business strategy." If you're still treating AI as a pilot, you're already behind.
That mindset is what drove Accenture and Telstra to do something unusual. They formed a dedicated joint venture in January 2025 to move faster than big-company structures allow. The goal: compress a five-year plan into two.
From Projects to a New Operating Model
Chakraborty called the JV a "very bold move" and an "industry first." It let the teams bypass the drag of annual planning and build a new identity centered on data and AI, while staying tightly aligned to Telstra's core business.
The setup blends Telstra's domain with consulting speed and Silicon Valley-style experimentation. That combination matters more than any single tool.
Fix the Foundation Before the Flash
"Don't get carried away with the shiny AI," Chakraborty warned. Clean up the legacy mess first. Culture and people are the starting line, not the afterthought. You're not only adopting new tech - you're changing how the company works.
Dayle Stevens from Telstra was blunt about technical debt. The company had "80 different data platforms" built up over time. You won't get data quality until you simplify the ecosystem. Telstra is consolidating those 80 down to three, already at 27, with a plan to finish in 18 months.
Escape the Annual Budget Trap
Big enterprises set spending once a year, then the market shifts. "AI moves really fast," Stevens said. Telstra had to become more adaptive - able to pivot as new models, infrastructure, and risks show up. That's a funding and governance issue as much as a technology issue.
Adoption Starts With Empathy
Telstra rolled out a generative AI tool called "Ask Telstra" to frontline teams. They didn't sell it as a productivity play. They aimed it at the teams' biggest pain points. The result: "We literally got a standing ovation."
They also launched an "AI Academy" to make reskilling visible and credible. People will lean in when they see personal upside and real support.
Trust Makes or Breaks ROI
Chakraborty was clear: trust is the bottleneck. Companies have poured billions into infrastructure without a line of sight to returns because leaders, employees, and stakeholders don't fully trust the systems or the outcomes.
Trust by design means model oversight, data lineage, human-in-the-loop, clear policies, and education. For a useful reference, see the NIST AI Risk Management Framework (NIST AI RMF).
Leadership Has to "Walk the Talk"
This can't be delegated. "The CEO and the CEO's direct reports [have to be] fully behind this," Chakraborty said. Leaders need to pull the organization forward, not passively approve budgets and wait for updates.
A Practical Playbook for Executives
- Make AI the strategy: Tie AI outcomes directly to revenue, cost, risk, and customer metrics. No side bets, no vanity pilots.
- Create an operating vehicle: If your org can't move fast, build a JV, newco, or separate P&L with the mandate to ship.
- Break the one-year funding cycle: Use rolling quarters and stage gates. Fund outcomes, not projects.
- Consolidate data platforms: Pick a target state (e.g., 3 platforms), define a 12-18 month cutover, and stop net-new on legacy.
- Start with frontline pain points: Deploy tools that remove daily friction. Win hearts, then scale.
- Engineer trust: Data governance, model evaluation, human oversight, audit trails, and clear accountability. Publish trust metrics.
- Upskill at scale: Stand up an AI academy, certify roles, and attach learning to career progression and compensation.
- Measure real value: Track cycle times, cost to serve, NPS/CSAT, churn, revenue per employee, and incident rates. Report monthly.
- Blend talent: Pair domain experts with data, product, and engineering. Incentivize shared outcomes.
If You're Starting This Quarter
- Appoint an executive sponsor with budget authority.
- Pick two frontline use cases tied to revenue or cost. Ship within 90 days.
- Freeze legacy sprawl. Set the target data-platform count and publish the decommission plan.
- Stand up a trust checklist for every AI release: data source, model risk, human oversight, monitoring, fallback.
- Launch a visible reskilling path for managers and frontline teams. Make it part of performance reviews.
Resource: Executive Upskilling
If you're building an internal AI academy or mapping roles to capability paths, see curated programs by job role here: AI courses by job.
The point is simple: don't treat AI as a project. Treat it as the operating system of the business. Build trust, fix the foundation, and lead from the front.
Your membership also unlocks:
