
.jpg?h=665&iar=0&w=1482&sc_lang=en&hash=259F979BC3E10533D78F931582D4186B)
Publication
Leading Work Transformation with AI: Lessons from the Work Innovation Tour 2026, Mumbai
Author: Rahul Khosla
Last Updated: 20 November, 2026
AI has made our people faster. It has not yet made our companies faster. That gap was the thread running through the Work Innovation Tour, which Altudo and Asana co-hosted in Mumbai.
The event was a full house and brought together more than 100 attendees, including over 12 CXOs, to explore what needs to change when AI becomes part of everyday work. From live demonstrations of Asana + AI to a customer panel featuring Alok Jaitley and Rohit Bhatti from Maruti Suzuki and Abhishek Kumar from Blue Star, the conversations offered practical perspectives on making human + AI collaboration work at an enterprise level.
There was one fundamental question on the table: how does Human + AI collaboration change the way work gets done?
The question points to the contradiction in enterprise AI today. AI is helping individuals work faster, yet many organisations are still struggling to translate those productivity gains into more effective ways of working across the business and translating that into business value.
Altudo works with enterprises to bridge this gap, helping them move beyond isolated AI experiments towards connected workflows and measurable business outcomes. Asana provides the platform for Agentic Work Management that makes human and AI collaboration seamless.
Lessons on transforming work with AI
Value comes from redesigning work, not from adding models to it. The nine lessons below, from presenters, customers and CXOs in the room, show what that takes in practice:
1. Redesign the work before adding AI
AI transformation is a work design problem, not a technology problem. Every company can rent the same models, so the advantage lies in how the work is designed. Dropping AI into today's ways of working will not deliver transformational returns; the work itself has to be reimagined. History makes the point. Factories saw little productivity gain from electricity for nearly four decades, until they redesigned the factory floor around it. India applied the same logic with UPI: it did not digitise the cheque; it redesigned the payment. Enterprises now need to rethink how work flows in the same way.
In practice, that starts with mapping how work actually moves. Remove unnecessary steps, clarify decision rights and identify where human judgement matters most. Only then decide what agents should handle. The goal is not a faster version of a broken workflow, but a better one.
A campaign management example was presented at the event, moving from a manual workflow to an agentified workflow in Asana:
Workflow in action without AI agents
Workflow in action with AI agents
Asana brings Humans and Agents together on a single canvas.
2. Choose the use case with a business outcome
Start with the business: does the use case grow revenue, protect revenue or reduce cost? If it does none of the three, it is a demo. Then test it. Is the work repetitive, valuable, complex and expensive? The more that apply, the stronger the case. Prioritise high value and low effort first. Baseline the KPI, check the data and name an owner before building. And aim past time saved. Hours saved are useful, but revenue and cost outcomes are what the CFO tracks.
Source: James Tidey, Asana
Align your business use cases with business objectives. Altudo’s Business Value Framework links initiatives to business objectives. An illustration was shown at the event:
The view below zooms out to the full picture, then zooms in on one scenario.
3. Create a context layer that becomes the ‘company brain’
Context is what turns an agent from something that simply responds into something that gets things done.
Roger Federer may be one of the greatest tennis players, but hand him a cricket bat on a Mumbai maidan, would he be as great? The context of tennis and cricket are different (although I admit, it’s hard to imagine RF not being elegant at everything he does!). Agents are no different. They need to understand the environment they're operating in, the goals they're working toward, who owns what, and the decisions that have shaped the business.
That's why shared company knowledge matters. A company's collective knowledge, from documented processes and institutional experience to data across CRMs, ERPs, HR systems, and other platforms, gives agents the context they need to make better decisions.
Asana's Work Graph brings this context together by creating a connected, real-time picture of who is doing what, when it's due, and how it contributes to broader business goals. Through integrations, it connects work across teams and systems, giving agents a clearer understanding of how the organisation operates.
The following illustration, shown at the event, described how Asana can pull context from underlying systems and power human and agentic work.
Agents need a harness. James Tidey from Asana pointed to five essentials that allow agents to work reliably as part of a workflow: context, memory, reflection, tools and governance with the right guardrails. When any of these is missing, agents are more likely to make mistakes, take wrong actions and lose people's trust. The level of risk should decide where humans stay in the loop. Workflows sit on a spectrum from human-led to agent-led. James made a good point - nobody is comfortable letting an agent sign off a $1M agreement alone.
James Tidey speaking about the Agentic Harness
4. Treat agents like teammates, not tools
Give AI agents a clear scope, provide the context they need, and let them build on what they learn. Keep humans in the loop with guardrails that let the team review, guide and take ownership of the result.
We applied this ourselves: an agent set up this way helped build parts of the talk I gave itself. Its contributions were reviewed in the same way as the rest of the team's work.
5. Governance is how you go faster, and it covers people too
A roundtable discussion raised an important point: governance is not only for AI models. People need clear decision rights too. For example, how should a junior employee review an AI-generated recommendation that appears to require expert-level judgement?
Guardrails enable speed rather than block it. Clear permissions, review steps and escalation paths let teams delegate more work to agents with confidence.
6. Change management may be the biggest priority
Employees want to know how AI will affect their roles, creativity and careers, and that conversation cannot wait until rollout. People need opportunities to shape the change, learn new skills and understand where their judgement remains essential. Change management is not an afterthought; it should be at the heart of how companies implement AI.
One attendee asked whether AI can help people become more human. It’s an interesting question - the time AI gives back has to be deliberately redirected to work that needs human judgement, such as customer relationships, critical decisions and creative problem-solving. Without a plan for that time, the gains disappear into more of the same work.
7. Design the technology stack for humans and agents together
Most enterprise stacks were built for people viewing information and navigating content. Agents change who uses the stack. They read, decide and act across it. So the question is no longer which AI tool to buy. It is how people and agents engage with the whole stack, and how the layers fit together:
- Systems of work. Where work is planned, assigned, approved and tracked. This is where humans and agents meet on the same task, with an owner, a deadline and a visible trail.
- Agents and models. The reasoning and the digital teammates that do the work. Models will keep changing, so treat them as swappable. Keep scope, memory and permissions outside them.
- Systems of data and insight. The records, metrics and meaning. Agents need more than storage. They need shared definitions, relationships and history, so “approved” or “revenue” means the same thing to every agent.
The questions every leadership team should be asking:
- Which systems can agents safely read and act on today?
- Where does shared context live, and who owns it?
- Could you swap a model without rebuilding the workflow?
- Can you see what each agent did, and what it cost?
Asana is the operating system for human-agent teams, and it sits in the systems-of-work layer. Built on its Work Graph®, a live map of every person, agent, task, project, goal and dependency, it lets people and AI agents run critical workflows from the same plan, with the same context and under the same governance. Through AI Studio, AI Teammates and StackAI, those workflows can reach beyond Asana into CRMs, ERPs and other enterprise systems. The view below shows how the pieces fit together.
8. Build a programme, rather than scattered pilots
Altudo's WorkOS blueprint is a simple framework that 400+ of our customers use to transform how they work. The companies that get past scattered pilots treat this as a programme with five parts:
- Inspire people. Run art-of-the-possible workshops where teams reimagine how they work and are inspired by what is happening around the world.
- Select use cases against company objectives. Every use case traces back to a business goal, a KPI and an owner.
- Prove value early. Take one department, put a few limited use cases live, measure against a baseline, and let results convert the sceptics.
- Build the foundation. Change management so people know what to do with the time they get back. Governance so every agent knows what it decides alone. A context layer for agents to work from.
- Scale department by department. Crawl, walk, run. Crawl with one workflow in one department. Walk by extending in waves. Run when the context layer, governance and measurement are shared across the enterprise.
Maruti Suzuki's transformation journey with Asana
Maruti Suzuki produced 2.3 million vehicles in the preceding fiscal year, and every model must clear homologation, the certification process that confirms safety and environmental standards are met before it reaches the market.
The homologation process at Maruti Suzuki involves 10,000 tasks yearly and over 100 projects, run by teams across three cities.
Previously, the homologation process relied on tools such as Excel, with no single source of truth and no visibility into team capacity. Here is how they approached it:
- They chose the tool through a structured proof of concept, not demos, and set up a dedicated digitalisation team for homologation.
- They worked on three pillars together: people, process and tool. Asana champions sit in every team to drive adoption, and processes were re-engineered task by task.
- The live use case is the WLTP programme: 68 applications, 1,000+ tests and 500+ vehicles over an 18-month window. Milestones are linked, so a delay cascades visibly through dependent work. Leadership gets project status and a capacity view on one dashboard.
The following slides show how approach change management and move from an Excel-based operating system to Asana.
Source: MSIL at Work Innovation Tour
The lesson for any enterprise: the tool proved the proof of concept, but people and process delivered the value.
Ready to transform how you work with AI?
The next step does not have to be a company-wide AI rollout. It can begin with one honest assessment: where does work slow down today, and which of those delays could be removed by redesigning the process?
Choose a workflow that matters to the business. Define the outcome, involve the people who own the work, select the platform that can power human and agent collaboration, and establish guardrails. Test it, measure the results and use those lessons to guide the next phase.
That is how human + AI collaboration delivers measurable results: people remain accountable for decisions, while AI helps reduce coordination overhead and move work forward.
