Promethium enables enterprises to get accurate insights by “talking” to all their business data. Go beyond BI with our AI Insights Fabric. Gartner Cool Vendor.
Joined October 2018
- Tweets 885
- Following 1,145
- Followers 366
- Likes 357
443 Photos and videos
Jun 12
Better models won't save a weak context layer.
Swap in the most capable model on the market — if the agent can't reach your definitions and tribal knowledge, accuracy stalls anyway.
Context is the ceiling.
→ hubs.ly/Q04kb7r90
7
Jun 11
Context engineering, in five levels: schemas → relationships → business definitions → semantic layer → tribal knowledge
Accuracy climbs as you add. Most agents plateau at Level 2 and guess at the rest.
A CDO's playbook for building the whole stack:
→ hubs.ly/Q04k9Lwv0
10
Jun 10
"I don't think our leaders have this figured out — not in politics, not across corporate America."
Andrew Clyne on the K-shaped economy AI is accelerating: productivity gains are real, the worker transition plan isn't. That gap is the next opportunity.
→ hubs.ly/Q04k9SVG0
5
Jun 9
New Gartner research: context graphs are the essential infrastructure for agentic systems — capturing not just data, but the decision flows and institutional memory AI keeps missing.
Gartner's call: >50% of AI agent systems will use them by 2026.
→ hubs.ly/Q04k9Vfz0
7
Jun 8
Most analytics agents don't stall because the model isn't smart enough.
They stall because "how far we've scaled" and "what we designed" get treated as one decision.
Maturity expands. Architecture is a trade-off. Confuse them and production breaks.
→ hubs.ly/Q04kbyYY0
4
Jun 5
Your agents are only as good as the weakest pillar.
Three capabilities. Most enterprises solve for one. The Venn shows where it breaks.
→ hubs.ly/Q04jLl1s0
2
Jun 4
"What does it take to lead with empathy and have people actually follow you?"
Andrew Clyne — 30 years, 5 CDO roles — on the one skill no technical background gives you.
→ hubs.ly/Q04jLqK10
5
Jun 3
The agent runs. The answer renders. The number looks plausible. And nothing tells you whether to act on it.
That's the gap Mantra's Trust Harness is built to close.
→ hubs.ly/Q04jLcTj0
7
Jun 2
1 in 7 agentic AI programs reach sustained production.
Three studies. Three methodologies. One number.
What the 1 in 7 do differently →
hubs.ly/Q04jLg-K0
9
Jun 1
Where does your D&A team's AI budget flow — and is it the highest-ROI bet?
New data from 500 CDAOs says it isn't code generation but analytics.
→ hubs.ly/Q04jzS7G0
3
May 29
"AI-ready data means three things: the right data, the right context, and a clear read on user intent."
Miss one and the agent guesses.
That's what Mantra is built for.
→ hubs.ly/Q04gQRqX0
7
May 28
"My favorite employees now? It's Grok, Gamma, and Base 44."
— Andrew Clyne (CDO at Comcast, Visa, Expedia, CenturyLink) on the new generation of "team members" in his stack.
On The AI Data Fabric Show: hubs.ly/Q04h8djZ0
40
May 27
"The pilot worked. Six months later, it still only works on the fifteen questions it was built for."
Every operator has been here.
New playbook: 5 failure modes, context-readiness audit, reference architecture, launch-ready checklist.
→ hubs.ly/Q04gQQkq0
16
May 26
Where's your D&A team spending its AI budget?
For most: code generation pilots. Clean demo, easy ROI story.
Gartner CDAO Survey 2026: AI for analytics delivers 42% business-value lift. Code gen only 27%.
→ hubs.ly/Q04gQCSS0
24
May 22
Designing agentic analytics? Two decisions you can't conflate:
Maturity dimensions (domain scope, system scope, use-case complexity).
Design choices (autonomy, action scope, interaction mode) .
Most teams treat them as one. The pilot doesn't survive it.
→ hubs.ly/Q04gQJdR0
13
May 21
New BARC Spotlight from Kevin Petrie: the "agentic analytics fabric" lets analytics agents reach distributed data live and activate the metadata that gives them context.
47% of orgs have AI data in multiple physical locations. The fabric is built for that. hubs.ly/Q04gQGm30
1
2
24
May 20
At Gartner D&A EMEA last week, we got the same question on repeat:
Context graph vs. knowledge graph vs. data catalog vs. semantic layer — what's the difference, and why does AI need any of them?
We wrote the breakdown.
→ hubs.ly/Q04gQdv70
1
14
May 19
New on The AI Data Fabric Show: Prat Moghe with Andrew Clyne — serial CDO (Comcast, Visa, Expedia, CenturyLink).
His take: the CDO role has permanently shifted from protecting the bottom line to driving top-line growth.
Prat's takeaways: hubs.ly/Q04h7Rpc0
1
33
May 19
Watch the full episode:
- YouTube: youtu.be/OWKievu0XIE
- Spotify: open.spotify.com/episode/7A7…
- Apple Podcast: podcasts.apple.com/ar/podcas…
1
9
May 18
What's the highest-ROI area for CDAO investment?
New Gartner CDAO Agenda Survey 2026: analytics use cases deliver the highest ROI — and have the lowest enterprise penetration.
42% lift, vs. 27% for code generation, where most teams are spending today.
→ hubs.ly/Q04gQJcY0
32