
A global innovation firm focused on building best in class digital products across platforms. #RightTheFuture
Toronto, Ontario
Joined August 2015
- Tweets9.2K
- Following1.2K
- Followers2.8K
- Likes4.5K
Most AI work in automotive gets built before anyone agrees what finished means.
Only about a third of initiatives ever scale past pilot.
Write the test first, in numbers the business already tracks:
→ time to market
→ warranty cost
→ recall fix time
→ dealer revenue
Then fund what passes.
#AgenticAI #Automotive #AIinProduction #ProductEngineering
The AI-Native mandate has already come down from boards and investors, and most leadership teams agree with it. What nobody has answered is who does the building.
The people who would do it are already running the business, and none of them has done this before. A strategy deck doesn’t close that gap, and neither does a pilot that never leaves the demo. Someone has to be accountable for the work reaching production.
That’s the job we take. Bring us the problem.
hubs.la/Q04FlGW00
#AINative #EnterpriseAI
1
1
34
New AI tools. The same old ways of working.
The licences landed on top of the same systems and the same approval chains, so the productivity never reaches the P&L. That’s AI theatre.
Pick one real problem. Solve it properly. Put it in production. Then do it again, until the business has been redesigned around AI.
Bring us the problem. Let’s build the plan.
#AINative #EnterpriseAI #AITransformation #ProductEngineering
A demo runs on clean data and a scripted path. The live operation runs on fragmented systems, one delayed aircraft putting crew hours and connections at risk, and security rules that cannot be switched off for the test. 🧵
Score it on outcomes the business already tracks. Then set a production gate: fund the scale only when the parallel run beats the live operation on recovery, revenue and service cost.
1
3
Agents that stall on the way to production usually stall on the data underneath them, not the model. A parallel run surfaces that in weeks, before the budget is committed.
#AgenticAI #TravelTech #Aviation #AIinProduction
6
AI pays off when it adds capacity without adding risk. In most pilots a person is still copying output between systems and checking it by hand.
Three layers change that: a harness, a loop, and a graph.
The harness is where it starts. Telling an AI to stay out of payroll data is not a control. Removing its access is.
On our own platform, case study creation dropped from days to about fifteen minutes.
Our newest piece is at hubs.la/Q04Dtl_L0
#EnterpriseAI #AINativeEnterprise #TribalScale #AIAgents
43
Passed the test. Failed the customer. 🧵
50% of 157 organizations in a self-selected VentureBeat survey shipped AI last year that passed its tests, then failed a customer. Only 5% fully trust automated evaluation today.
1
30
Grade AI like an essay instead:
1. Owners define right, using real past cases.
2. Grade every change. If it drops, it does not ship.
3. Score live work after launch.
4. Every live miss joins the answer key.
5. Report the grade like revenue, at the exec table.
1
5
If you cannot show the grade, you do not know whether your AI works.
Bring us your hardest AI problem: tribalscale.com
#AINative #EnterpriseAI #AIEvals
6
Most companies buy AI the way they buy software, then count seats.
Capacity is the number that matters: how much work gets delivered, and who is accountable when it does.
People keep the judgment. Agents take bounded work, with a clear point where it hands back.
Read our newest piece: hubs.la/Q04D97MK0
#AIAgents #AINative #EnterpriseAI #FutureOfWork
“The AI story will be decided in the S&P 493, not the Mag 7.” 🧵
Apollo’s chief economist Torsten Slok split the S&P 500 into the seven companies everyone quotes and the other 493.
Slok is right that business runs on judgment and accountability. That is an argument for redesigning the work, not for waiting.
Pick one workflow that touches revenue or cost. Redesign it end to end, and decide where a person owns the judgment.
1
4
That is the work we do with clients: one workflow at a time, into production, and judged on whether the number moved.
#AI #EnterpriseAI #AITransformation #Leadership
7



