User
Write something
Pinned
Start here: what this is, and how to use me
I'm Ken Corigliano. Think of this group as a technology and AI officer you can call on when a decision is live, without hiring one full time. Where the judgment comes from: 3x CTO; 1xChief AI Officer; Deputy Chief of Innovation at the Defense Intelligence Agency, Air Force lead on the White House AI Futures Working Group, more than 1,000 venture offerings diligenced for In-Q-Tel and DIU, retired Air Force Lt Col. Today I run technology and AI for operating companies. I have no vendor to favor and no budget to defend. If you already have a CTO or a data chief, good. Use me as the outside view they cannot give you. Three things to do today. First, post in Ask: the one technology or AI decision in front of you right now, and what it costs you if you get it wrong. Second, open the Classroom and take lesson 1. It takes ten minutes. Third, tell me your industry and the size of your team, so my answers fit your situation. I read every post. You will get a straight answer, including when the answer is "do nothing yet."
0
0
Watch my journey, Part 4: one engine, every signal, an AI magnet
Part 4 of the NuCalm fractional CTO rebuild I'm leading: the analytics behind it. Sources for every claim are in the comments. 20 different AI systems now read nucalm.com. About 1,500 AI page reads a day in early November, up 60 percent since October, and 3 times as many AI assistants fetching our pages for a single person. One engine, every signal, built to be an AI magnet. Part 3, the team system, is the post just above this one. I'm Ken Corigliano, fractional CTO at NuCalm and an on demand AI and technology officer for a handful of other companies.
Watch my journey, Part 4: one engine, every signal, an AI magnet
Watch my journey, Part 3: the team system
Update #3 on the NuCalm fractional CTO rebuild I'm leading. For a long time the pattern was: one person writes every spec, every team sends it back for changes, and ideas die in the gap between a good instinct and a finished thing nobody has to re-explain. This is how we rebuilt that into a team system. Idea to impact, no send-backs. Part 4, the analytics behind it, is up on YouTube now too. I'm Ken Corigliano, fractional CTO at NuCalm and an on demand AI and technology officer for a handful of other companies.
0
0
Watch my journey, Part 3: the team system
Your vendors have vendors: where your data actually goes
Three threads on LinkedIn, one problem: where your data goes once a vendor has it. The lock-in The AI add-ins bundled into the tools you already pay for are holding your data hostage. Many train on your data. Some don't. The ones that don't are still storing your AI: the prompts, the workflows, everything it has learned about how you run the business. That's what makes you hesitate to switch vendors, and it never appears on a P&L. Full reply: https://www.linkedin.com/feed/update/urn:li:groupPost:2046019-7520833773517778944/?dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287523763835007041537%2Curn%3Ali%3AgroupPost%3A2046019-7520833773517778944%29 The threat Microsoft says the attackers took control of ConnectWise's cloud apps, then signed in to customer environments using the trust those apps already had. That's one vendor. Vendors have vendors, who have vendors. This isn't a primary, secondary, or tertiary concern. Your data sits in dozens, if not hundreds, of joints and connections: at rest, moving, being "anonymously" analyzed. "We don't train on your data" doesn't mean zero attack vector. We used to target the devices that anonymize, cross domains, or sanitize. If they exist, sensitive data exists, and where they exist is where the bounty is. Full reply: https://www.linkedin.com/feed/update/urn:li:groupPost:10307956-7519021937407225857/?dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287523763526000271360%2Curn%3Ali%3AgroupPost%3A10307956-7519021937407225857%29 The controls List every place an AI tool can post, pay, or send something without a person approving it. That list is your control gap. Then extend the list to your vendors' tools, and to their vendors.
0
0
Case file #3: the other side of talent skimming. 93% of applicants don't fit, and half get flagged.
On LinkedIn this week I wrote about talent skimming: job posts with a headline salary and a hidden client, built to collect senior resumes rather than fill a seat. Several of you sent me the same question back: what does the employer side look like? A new report answers part of it. hackajob, which sells an AI screening tool, published "Who's Really Applying?" in September. The numbers come from its own customers, so read them as one vendor's view, not a census. What it found. Of 175,000 applications from job ads, 93 percent did not fit the role. Only 19 percent of those were a skills mismatch. The rest were salary (28 percent), location (23 percent), visa (15 percent), and experience (10 percent). Of 70,000 US candidates open to remote work, 46 percent were flagged as a possible risk before a human ever spoke to them. Every application sent through an AI auto-apply agent was flagged. The outside numbers are harder to argue with. In a GetReal Security survey of 668 leaders, 41 percent said they had already hired and onboarded a fake candidate. Gartner projects one in four candidate profiles will be fake by 2028. In February a federal court sentenced an identity-rental operator for placing North Korean IT workers at 40 US companies. So here is the loop. Recruiters get hundreds of applications per role and use AI to filter. Candidates apply to more jobs and hear nothing, so they automate. Employers trust applicants less, skimmers post fake roles to harvest the resumes that are flying around, and the real candidate and the real employer both lose. What I would do if I ran hiring at a company with 20 to 200 people and no technology officer. 1. Verify identity and location before you book the first interview, not after the offer. A fake remote hire is a payroll loss and a data loss. 2. Stop using the resume as the first filter. Three targeted questions about the actual work sort candidates better than a keyword scan, and they are much harder to automate. 3. Write down your AI use policy, and say in the posting that you check for misuse. The one-page template is lesson 3 in the Classroom. 4. Name your company in the posting. Hidden-client posts train good candidates to keep their resume. If you use an agency, make them say the search is retained. 5. Track how many applicants from each source get flagged and how many drop out at verification. If a job board sends you volume that is 67 percent flagged, as the report found, it is costing you more than it is bringing.
0
0
1-11 of 11
powered by
Your Chief AI Officer
skool.com/your-chief-ai-officer-3918
Get support from A Founding Member of the US Presidential AI Working Group.
Converting AI "thoughts," "theories," "headlines," "clickbait" to $.
Build your own community
Bring people together around your passion and get paid.
Powered by