Case studies · Agentic AI strategy
Case Studies
Real AI strategy journeys, from the first measurement to what changed, with dates, evidence and what didn't work. We publish a client's story only when they agree to be named and every step can be checked.
Client journeys
Our first client stories are in progress
None are published yet. We don't show anonymous percentages, client counts or before-and-after numbers that you can't check. Each client story will appear here once the client agrees to be named and the work has dated results, and each one will show:
01
The client's real name, with their written permission
02
A dated starting measurement, with the questions and AI platforms tested
03
The work we actually did, with dates
04
What changed afterward, with dates and evidence
05
What did not improve
06
The limits on crediting any change to our work
Internal experiment, not a client result
BackTier on itself
Before asking clients to trust our method, we run it on BackTier and Jason AI Wade first, and publish every dated answer, including the ones where AI gets us wrong.
Sep 6, 2026 · Starting point
Asked a Gemini model four questions about BackTier. It read "BackTier" as a generic technical term, attached the wrong business to the founder's name, and left BackTier out of the providers it recommended. Evidence →
Oct 4, 2026 · Five-model check
None of five AI models identified BackTier as a company; three invented confident wrong descriptions. Every answer is saved word for word. Evidence →
Oct 25, 2026 · Work done
Standardized the founder's name to Jason AI Wade across the site and its structured data, kept older forms only as recorded alternate names, and logged the change publicly. Evidence →
Oct 25, 2026 · Work done
Built and ran the free Agent Readiness Checker on backtier.com: all 12 checks good, and none of 15 AI crawlers and agents blocked from the homepage. Evidence →
Oct 26, 2026 · Latest measurement
Asked four AI models eight questions. 0 of 32 answers identified BackTier; in 6 of 8 answers about the founder, the model described the singer Jason Wade of Lifehouse instead. Evidence →
What has not improved yet
- No model we tested identifies BackTier or Jason AI Wade yet.
- The founder's name still resolves to a more famous person in most answers.
Limits: These are single answers from models without web search, a few weeks apart. Models change without notice, and nothing here shows that any one change we made caused, or will cause, a different answer.
Research, not client results
Field studies
How we would approach specific markets, written before any engagement. They describe a method, not a delivered outcome.
Start your own journey
Every journey starts with a conversation about your business and a dated measurement of where you stand.