What Actually Works: Agentic AI Across Every Function

Most organisations are still running pilots. We have moved past that.

At Wall Street English, we are deploying agentic AI across every function in the business. Engineering. Product Management. Educational Content. Marketing. HR. Finance. All of it.

And it is working. Not in a promising proof of concept way. In a generating actual results way.

The Approach: Confident, Not Cautious

There is a temptation in large organisations to start small with AI. Run a pilot. Learn. Then maybe, slowly, scale.

We chose a different path: confident deployment, not cautious experimentation.

The difference is not reckless — it is strategic. We decided that AI capability is now a competitive differentiator, and waiting for perfect understanding means falling behind. So we moved fast, learned in production, and adjusted as we went.

Here is what we have learned about what actually works.

Engineering: Quality Over Quantity

In engineering, AI is not primarily about writing more code. It is about writing better code.

Our teams use AI assistants to:

• Review pull requests — catching issues before they reach production
• Generate tests — increasing coverage without slowing delivery
• Explain codebases — onboarding new engineers faster
• Automate documentation — keeping knowledge bases current

The result: fewer bugs, faster onboarding, more time for engineers to focus on architecture and solving hard problems.

Product Management: Synthesis at Scale

Product managers are drowning in information. User research, market data, competitive analysis, stakeholder feedback — the volume is overwhelming.

AI helps our PMs:

• Synthesise user research from multiple sources into actionable themes
• Draft product requirements with clear success criteria
• Prepare for stakeholder meetings with AI-generated briefing documents
• Track and analyse metrics across multiple products

The result: PMs spend less time on synthesis and more time on judgment.

Educational Content: The Human Touch, Amplified

This one matters. We are a language learning company — Wall Street English. The product is education — human education, facilitated by technology.

AI is not replacing our educators. It is amplifying them:

• Lesson planning assistance — helping lesson designers prepare faster
• Content personalisation — adapting materials to individual learning paths
• Feedback generation — giving students instant input on classes and practice work
• Quality assurance — ensuring content meets pedagogical standards

The result: better learning outcomes, more time for teachers to teach, more personalisation for students.

Marketing: From Campaign to Conversation

Marketing is transforming:

• Content generation — drafts for social, email, and web at scale
• Personalisation — tailored messaging for different segments
• Analytics — deeper insights from campaign data
• Customer journey mapping — AI-powered touchpoint optimisation

The result (early results): more relevant content, faster campaign execution, better measurement.

HR: People Operations Reimagined

HR often gets overlooked in AI transformations. We are changing that:

• Recruitment screening — AI-assisted candidate shortlisting (in progress)
• Onboarding automation — personalised welcome journeys for new hires (in progress)
• Policy QandA — instant answers to common HR questions (live)
• Performance analytics — deeper insights into team health (exploring)

The potential: HR focuses on strategy and people, not administration.

Finance: From Reporting to Insight

Finance teams are typically buried in reporting. AI changes that:

• Automated reporting — real-time dashboards instead of manual updates
• Forecasting assistance — AI-augmented financial modelling
• Anomaly detection — catching irregularities faster
• Document processing — automating invoice and receipt handling

The potential: finance teams become strategic advisors, not report generators.

What Made the Difference

All of this works because we did three things differently:

1. Function-specific deployment — We did not roll out generic AI tools. We identified specific use cases for each function and deployed accordingly.
2. Leadership commitment — The leadership team, myself included, used the tools publicly. We normalised AI use by demonstrating it.
3. Training with context — We did not just train people on how to use AI. We trained them on when to use it and when not to.

Actionable Takeaways

1. Deploy function-specific, not generic — One AI tool does not fit all functions. Customise for each.
2. Lead by example — Leaders must use AI visibly to normalise it
3. Train on judgment, not just tools — Knowing how to use AI is easy. Knowing when to use it is the skill.