2023: Everyone bought AI.
Almost no one changed the work.
For the last three years, most people used AI
↳ To draft faster.
↳ To summarize documents.
↳ To automate repeatable tasks.
They said people are still needed for decisions on the bubble.
Those that required seeing across systems.
Edge cases that needed judgment.
Across BCG, Deloitte, McKinsey and Anthropic, a pattern is emerging.
Automation is not enough.
Not anymore. Agentic AI changes the map.
In most companies, it takes a custom IT project to make 17 different systems talk.
The invoice is in one place, the contract in another, the rate card in a third and no one tracks the promises in email.
Someone on your team bridges between them.
This is why our AI ambitions only have a 5% success rate (MIT).
Every consultant says you need to clean up the cracks: Bad Data and Weak Governance.
Most companies don't take the time or money to do it.
Agentic AI is emerging to skip the foundations.
- It Finds Clues in the Mess
This new AI reads emails, PDFs, and screenshots like a human, focusing on what drives the decision.
- It Builds a Map
AI builds a giant web of clues, linking order numbers and dates together automatically and remembers past human choices.
- It Translates Secret Codes
Different companies use different words for the exact same thing. The AI acts like a universal translator based on the business' goals.
- It Audits its own work
Before taking any big action, it must trace its steps and prove its logic.
Most companies are wasting time automating tasks. And that leads to poor decisions.
Here is what to do instead:
✅ Feed AI the facts people look at, the rules it follows, what actions happen next, and how results are checked.
✅ Break the work into decisions, not tasks.
A vendor invoices $2.3M. The contract says $1.9M, but the rate card expired. An email extends terms 30 days.
This new AI works out which facts control, calculates the real amount.
It disputes the rest, and tracks the credit.
✅ Figure out the objective - not the task count.
Ex. Detection, drafting, and routing are not value.
✅ People build context. AI acts.
Once AI can see the facts, make the call, act, and prove why, the old workflow is gone.
♻️ Repost for the leader who want to understand what AI does.
🔔 Follow Betsy Tong for what AI can do now.