Global, Award-Winning, Technology Research, Advisory, Consulting & Education Firm | A @thofgr company | 📧 [email protected] | 💬 https://xitter.cf/t.co/WF57jmMx1Z
Global
Joined March 2013
- Tweets14.7K
- Following2
- Followers2.8K
- Likes788
Pinned Tweet
We sent this tweet 9 months ago but much has changed since. Here's an update on our YouTube channel, Greyhound Studios:
âś…11,000+ global IT Decision Maker subscribers (from 1,000 in Mar'22)
âś… 300,000+ views so far on the current videos (from 150,000 in Mar'22)
Thank you, all!
99 percent.
That is the roster coverage Fusion Claw's plan reaches in @Oracle's own hospital example, up from 70 percent. It fills 70 of 72 open shifts at a 5 percent relative cost increase, with no policy exceptions introduced.
In Oracle's shipment example, freight utilization rises from 61 percent to 88.2 percent, with 26 fewer moves and USD 58,000 in projected savings.
Oracle AI World opens tomorrow. If your business runs on Oracle Fusion, the question is no longer what Fusion Claw can do. It is how much authority to hand it, process by process, and who signs for it.
Greyhound Consilium, the B2B thought-leadership arm of The House Of Greyhound and a sister brand to Greyhound Research, has partnered with Oracle to turn that question into a practical guide: The Delegation Dial.
Fusion Claw is a dial, not a switch. Oracle's three settings, human in the loop, human in the lead and autonomous execution, change how often a person steps in, never whether controls apply. In Oracle's words, AI "can't grant itself authority to act."
Start every new process at human in the loop. Move it up only as the results prove it.
Before your first run, you need answers to five questions:
1/ Which outcome goes first, and what will it prove to the CFO?
2/ Is everything in place before the first test run?
3/ Who granted the authority, and where is it written down?
4/ What will each outcome cost in AI Units?
5/ Who signs for each decision, and who gives the second signature?
The guide gives you a tool for each, 13 in all.
So don't walk into AI World with a feature list. Walk in with a game plan: Read the guide, and use its scorecard to pick your first process.
Work through the five questions with its tools, and draft your roadmap on its 90-day plan.
Take that roadmap to your Oracle account manager, at AI World or the week after. Ask where Fusion already holds the controls you need, and what still has to be set up.
Book the four working sessions the guide closes with. The first needs only your shortlist and today's measured results. You come out with a dated plan.
Which process in your Fusion estate will Fusion Claw take on first, and who signs for it? If you cannot name both today, start there.
Full guide, with all 13 tools, in the first comment.
#OracleFusionClaw #OracleFusion #AgenticAI #EnterpriseAI #AIGovernance #CIO #OracleAIWorld #AIWorld
Read the full guide, with all 13 tools, on the Greyhound Research website:
greyhoundresearch.com/the-de…
Have a question about where Fusion Claw fits in your own processes? Ask Greyhound Brain, the AI-enabled research assistant on the Greyhound Research website.
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6014BHjdCI
116.
That is how many PERM certifications the six IT services firms on Washington's suspension list received combined in nine months.
Ten years earlier, the same six received 7,999 in a single year.
Two weeks ago, Washington announced it would stop processing PERM cases for these six firms and two software companies. The headlines have moved on. The clocks inside every affected worker's case have not.
Greyhound Research went back to the U.S. Department of Labor's (@USDOL) own files. The surprises keep coming.
The eight named employers received 1.8 percent of national PERM certifications in the nine months to June 2026, down from 9.9 percent in fiscal 2016.
Most of what remains sits with the two software companies. Of the eight's 1,898 certifications, @Microsoft received 1,600 and @Adobe 182. The six services firms received 116 combined.
The services firms were not alone in leaving PERM. @Amazon, @Google, @Meta, and @Apple had largely left too, and they are not on the list.
None of the eight was in the top ten by H-1B positions certified in the first half of fiscal 2026. The list does not follow an H-1B ranking.
The public files cannot explain why these eight were named. Named and unnamed employers sit side by side on each measure tested.
Of the ten figures officials reportedly cited, three are consistent with public counts. Two are not found in the public record, and four are not tested on it. The government may hold records that are not public.
As of October 20, Greyhound Research could not locate the notice, the legal authority, the entities covered, or the start date in public channels. Meanwhile, three of the eight clocks inside each worker's case can end a green card case or a stay outright.
For CIOs, the exposure is not the brand on the contract. It runs through named people in named services. Nothing in the officials' statements, as reported, indicates a product risk for software subscriptions.
The decision rule has four steps: map the business dependency, verify the sponsoring legal entity, obtain the supplier's case counts by stage, and fund a proportionate continuity response.
For suppliers, named or not, this is a pipeline test. Would their PERM pipelines and their recruitment and pay records stand scrutiny?
The full report runs to 15 sections, 4 exhibits, 22 tables, a buyer playbook and a supplier playbook, 13 open questions, and 183 sources.
Which of your critical services depends on one named person whose green card case sits with one of the eight sponsors? If you cannot answer that today, start there.
Full report in the first comment.
#PERM #GreenCard #H1B #CIO #ITServices #Outsourcing #Immigration
Read the full research report on the Greyhound Research website:
greyhoundresearch.com/washin…
Have a question about what the suspensions mean for your services or your suppliers? Ask Greyhound Brain, Greyhound Research's AI-enabled research assistant.
Greyhound Research retweeted
Microsoft says it now puts back more water than it takes out. Globally, that's true. In Quincy, Washington, home to its single largest draw, it isn't.
That gap is the story of AI infrastructure right now. The promises are made to towns. The proof is published for the planet.
My team at Greyhound Research (@Greyhound_R) and I went through @Microsoft's own disclosures, site by site. Of the 29 data center locations Microsoft lists, seven show any replenishment. Eleven carry a promise dated FY30. Eleven have no plan disclosed.
Those 29 sites cover only about a third of the water Microsoft reports withdrawing. Where the rest is drawn, site by site, is not published. And in the year the milestone landed, withdrawals rose 14.5 percent.
Credit where it's earned. Phoenix, San Antonio, Madrid, and Querétaro show more water put back than the site withdraws. Microsoft has ended non-disclosure agreements with local governments. And it took the harder route to its renewable numbers.
But a global total cannot tell a resident whether their own watershed is better off. For a CIO signing the next capacity deal, the question is no longer whether the provider is green on paper. It is which of your regions run on capacity that sits outside the published record.
The full report, with ten live exhibits and ten questions for Microsoft, is in the thread below.
#GreyhoundStandpoint #DataCenters #AIInfrastructure #Microsoft #Sustainability
Greyhound Research retweeted
“Enterprise #AI autonomy must be built on clear boundaries of authority, traceability and accountability. With #OracleFusionClaw, Oracle introduces a governed runtime that enables #FusionAgenticApplications to plan and execute work within the system of record, with an auditable record of outcomes,” says Sanchit Vir Gogia, chief analyst, Greyhound Research.
social.ora.cl/6015BHyZYP
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6010BH3Mei
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6017BHOcE9
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6014BHPzap
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6018BHKOKy
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6019BHKTCJ
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6019BHKSst
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6011BHKQiP
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6010BHz3nn
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6014BHJ88Q
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6019BHyFNd
Greyhound Research retweeted
In his latest report, @s_v_g, chief analyst, @Greyhound_R, examines #OracleFusionClaw through four critical requirements for enterprise AI autonomy: establishing agent identity, reading execution logs, bounding authority and reversing actions.
social.ora.cl/6016BHywOu
Oracle Fusion Claw is @Oracle's bid to make enterprise AI stop suggesting and start doing.
So, what is it? Announced on September 29, it is a governed runtime that lets Fusion's agentic applications complete complex work. A frontier model plans, and deterministic computation, running where Fusion is hosted, executes.
Customers set the autonomy, from quick assistance to full auto, and every outcome ends with a receipt. Oracle says 25 Claw-powered applications are rolling out, cohort by cohort, with Release 26D.
How did Oracle get here? In July, Oracle launched tools to build and run agentic applications inside Fusion. Greyhound Research then set four tests for any AI that acts: name its identity, read its log, bound its authority, and reverse its actions. Oracle met one, and no vendor reviewed met all four.
Does Fusion Claw pass? @Greyhound_R reran them against Oracle's documents and written answers, and added a fifth: the regulators.
1/ Authority: stronger than in July. Oracle's documentation says the AI "can't grant itself authority to act."
2/ Log: the receipt is the richest per-run record of any system-of-record vendor reviewed. Export and retention terms are not yet documented.
3/ Identity: Claw acts within its agents' role-based access, on the running user's token. Its own identity is the next step.
4/ Reversal: correction is handled in Fusion itself, as with every vendor reviewed.
5/ Regulators: across seven jurisdictions, the receipt meets the rules on record-keeping. Intervention falls to the controls and people around it.
What will the CFO and CISO ask? What it costs, and where it thinks. Planning on a frontier model is metered at 20x for a Claw action, while deterministic execution uses no units, Oracle confirms. Units per outcome are unpublished.
Reasoning runs where the model provider offers the model: for @OpenAI, Europe for Mumbai and Singapore, the United States for Sydney.
Greyhound Standpoint
At Greyhound Research, we believe Fusion Claw is new machinery, not new vocabulary. It makes the delegation of authority every enterprise already holds executable, with a receipt as evidence it was obeyed.
Oracle has built the part that obeys. The part that delegates is the buyer's, set in controls Fusion already holds.
So where should a CIO start? Set the boundaries and the unit cap first, with a named owner for each authority delegated. Then put the region of reasoning in writing, run research mode on one outcome class the controller can already reverse, read its receipt against the audit tables, and measure units per outcome. Full auto follows, one class at a time.
The full research report has the evidence, the Greyhound CXO Runbook, and ten questions only Oracle can settle.
When AI posts to your ledger on full auto, whose signature is on it?
Read it here: greyhoundresearch.com/oracle…
#OracleFusionClaw #AgenticAI #EnterpriseAI
Greyhound Research retweeted
Two weeks of coverage on @ITCInfotech acquiring @HappiestMinds Technologies, and almost none of it asked the question a customer actually cares about.
Whose delivery model survives?
Not whose logo goes on the door. Not the listing, and not the timetable. Whose engineering estate, whose security posture and whose account culture becomes the standard, and whose quietly gets rewritten.
At Greyhound Research, we believe the real story is not the transaction. The real story is that the middle of the Indian IT services market is being redrawn, and the mid-tier is where most enterprise buyers keep their leverage. Both firms bring something real to it.
ITC Infotech carries domain built inside a large diversified enterprise. Happiest Minds carries a born-digital engineering bench and a product instinct that larger firms have spent years trying to buy.
On paper, a serious combination. In practice, capability adds up on day one and engineering estates do not.
Every CIO reading this has been told at some point that an acquisition would not affect their account. Then the delivery lead changed. Then the offshore ratio moved. Then the certification you relied on was in transition.
@Greyhound_R recommends that buyers ask for the integration plan in writing rather than the synergy narrative, and that they ask this quarter rather than next year.
This research note was written for buyers, not for the feed. If ITC Infotech or Happiest Minds sits on your vendor list today, it is a must-read.
Link to the report in the comments.
#GreyhoundStandpoint #ITServices #CIO #IndianIT #Acquisition
