KalFlow · Editorial Reference
Sources Behind Keeping Up with KalFlow
Every external source cited across the KalFlow blog, mapped to the article and the claim it supports — with each one graded for how far it sits from the original research.
Compiled August 28, 2026 · Covers all 6 published posts on Keeping Up with KalFlow · 25 external sources
Bottom Line Up Front
- Three of six posts cite external sources. The other three are positioning and process pieces that make no factual claims requiring citation.
- “Your Local AI Solution” is solidly sourced. Six of its seven citations are U.S. government data or peer-reviewed economics. It will hold up to scrutiny.
- “Stop Paying the Manual Process Tax” is the exposure. Eight of its eleven citations are SEO content sites or vendor blogs — not the studies they’re paraphrasing.
- “AI For Skeptics” is mixed. Its two adoption figures are well sourced, but the bubble case — the Morgan Stanley capex comparison and the AI-training payoff claim — rests on secondary blogs rather than the research behind them.
- One headline stat has no citation at all. The McKinsey “42% of finance activities are automatable” claim appears in bold with no link and no report named.
- Fastest fix: replace the aggregator links with the primary reports they’re quoting, and either source or drop the McKinsey figure.
8
Government or peer-reviewed sources
4
Journalism / trade press
12
Aggregators & vendor blogs
1
Claim with no source given
Gov data Federal statistical agency, primary
Peer-reviewed Academic, published
Journalism Reports on a primary study
Aggregator SEO or vendor content, secondary
Uncited No source provided
Articles With Cited Sources
AI for Southwest Michigan Small Businesses: What Actually Works
Verdict: Strong. Six of seven are primary government or academic sources. The one secondary link is Forbes summarizing a study that is itself worth linking directly.
| Source | Tier | Claim it supports | |
|---|---|---|---|
| 1 | U.S. Census Bureau — Business Trends and Outlook Survey | Gov data | 19.8% of U.S. employer businesses used AI as of May 2026; 37% of firms with 250+ employees; under 20% of firms with four or fewer. |
| 2 | U.S. Census Bureau — “AI Use in Businesses” (May 2026) | Gov data | Dec 2025–May 2026: AI use rose among firms with 20+ employees but did not change significantly among firms under 20. Also the sector splits (Information 39.7%, Finance 33.9%, Retail ~14%). |
| 3 | Forbes — “MIT Finds 95% of GenAI Pilots Fail” | Journalism | MIT Project NANDA, The GenAI Divide: ~95% of generative-AI pilots produce no measurable return; failure mode is workflow integration, not model quality. Also the 90%+ shadow-AI usage finding. |
| 4 | U.S. Census Bureau — Household Trends and Outlook Pulse Survey, “AI Use at Work” | Gov data | ~55% of workers used AI for at least one of 11 job tasks; the full hours-saved distribution; and the task-frequency table (search 37%, writing 32%, ideas 32%, summarizing 31%, admin 27%). |
| 5 | NBER Working Paper 31161 — Brynjolfsson, Li & Raymond, “Generative AI at Work” | Peer-reviewed | Study of 5,179 customer support agents: 14% average lift in issues resolved per hour, 34% for novices, near zero for the most experienced. Later published in the Quarterly Journal of Economics. |
| 6 | FTC — Operation AI Comply enforcement sweep | Gov data | “No AI exemption from the laws on the books”; enforcement against unsupported AI claims, including the DoNotPay “robot lawyer” order. |
| 7 | FTC Tech Blog — quietly changing terms of service may be unfair or deceptive | Gov data | Amending terms of service to permit new data uses can itself be an unfair or deceptive practice — the basis for the “read the terms on data reuse” guardrail. |
Stop Paying the “Manual Process” Tax
Verdict: Needs work. Most statistics trace to SEO statistics roundups or vendor marketing blogs rather than the underlying research. The numbers may well be right — but as cited, they are not verifiable.
| Source | Tier | Claim it supports | |
|---|---|---|---|
| 8 | McKinsey & Company — no link, no report named | Uncited | “42% of all finance and accounting activities are fully automatable.” Presented in bold as the article’s opening statistic with nothing to click. |
| 9 | AdAI News — Accounting AI Statistics 2026 | Aggregator | Bank reconciliation is over 90% automatable; AI-powered accounting delivers a 30% faster month-end close. |
| 10 | Tommaso Maria Ricci — AI for Accounting: Complete Guide 2026 | Aggregator | 70–80% reduction in invoice processing time and 90%+ reduction in error rates for automated AP; rolling 30/60/90-day cash flow forecasts at 85–90% accuracy. |
| 11 | Articsledge — AI Accounting Tools: Complete 2026 Guide | Aggregator | Businesses using AI accounting tools report operational cost reductions of up to 30%. |
| 12 | Remarcable — Construction Material Management Statistics 2026 | Aggregator | A contractor’s procurement cycle dropped from 2–3 hours to 10 minutes per transaction after automating ordering. Source is a procurement software vendor citing its own category. |
| 13 | Software Oasis — 2026 Construction Automation Statistics | Aggregator | 47% of construction tasks can be automated with current technology. The article credits this to the Brookings Institution, but links to the roundup rather than to Brookings. |
| 14 | Arda — The Alarming Costs of Downtime | Aggregator | 60% of manufacturing companies report weekly delays in material deliveries. |
| 15 | U.S. Bureau of Labor Statistics — JOLTS release | Gov data | Warehouse worker turnover at 49%. Primary agency, but the link points to the generic JOLTS table of contents — a reader cannot locate the figure from it. |
| 16 | DC Velocity — MHI/Deloitte Annual Industry Report coverage | Journalism | 84% of supply chain leaders plan to adopt AI within the next five years. Trade press summarizing the MHI/Deloitte report; the report itself is publicly available. |
| 17 | Bridgit — AI Construction Statistics for 2026 | Aggregator | Only 16% of contractors use AI or automation for scheduling; 60% report no plans to adopt. Source is a construction workforce software vendor. |
| 18 | SellersCommerce — Warehouse Automation Statistics 2026 | Aggregator | Logistics robot sales rose 500% between 2019 and 2025, from 75,000 to over 450,000 units annually. |
Priority fix. Sources 9–14 and 17–18 are content-marketing pages, several published by vendors selling into the exact category the statistic flatters. Where the underlying study is named — Brookings on construction automation, MHI/Deloitte on supply chain — link the report itself. Where no underlying study is named, treat the number as unverified. The McKinsey 42% figure (source 8) should be linked to a specific McKinsey publication or removed.
The bubble may pop. The tools are staying.
Verdict: Mixed. The two adoption figures the argument turns on — Census BTOS and the U.S. Chamber survey — are primary and hold up. The bubble case does not: the Morgan Stanley capex comparison, the dot-com history, and the AI-training payoff claim all link to blogs or an encyclopedia rather than the research behind them.
| Source | Tier | Claim it supports | |
|---|---|---|---|
| 19 | Yahoo Finance — MIT report on generative-AI pilot failure | Journalism | MIT Project NANDA, The GenAI Divide: 95% of corporate generative-AI pilots produced no measurable impact on the bottom line, and the failures were organizational rather than technical. Note: this is the same MIT study cited in “Your Local AI Solution,” reached through a second, different secondary link. |
| 20 | AequiFin — “AI Bubble 2026: Is the Tech Rally About to Burst?” | Aggregator | Projected 2026 capex-to-sales ratio of 34% for big tech, above the 32% dot-com peak in 2000. The article credits Morgan Stanley, but links to a financial-advisory blog rather than the Morgan Stanley note. |
| 21 | U.S. Census Bureau — Business Trends and Outlook Survey | Gov data | 8.8% of U.S. firms under 250 employees use AI in actually producing their goods or services. Primary agency, but the link is the BTOS landing page — a reader cannot locate the 8.8% figure from it. |
| 22 | U.S. Chamber of Commerce — Empowering Small Business: The Impact of Technology on U.S. Small Business | Journalism | 58% of small businesses use generative AI — the broad end of the adoption range the article contrasts against the Census figure. This is the Chamber’s own primary survey rather than third-party reporting, so it sits closer to the research than the tier label suggests. |
| 23 | Wikipedia — Dot-com bubble | Aggregator | The NASDAQ fell more than 75% between March 2000 and October 2002, and roughly $5 trillion in market value evaporated. Tertiary encyclopedia; the underlying index history is public and citable directly. |
| 24 | Goldman Sachs — “The Late 1990s Dot-Com Bubble Implodes in 2000” | Aggregator | Amazon used the bust to build the infrastructure it still runs on — the basis for the “the technology survived, the nonsense didn’t” argument, and for the roughly 90% drop in Amazon’s stock stated in the same sentence. A firm-history marketing page, not research. |
| 25 | CloudSecureTech — Small Business AI Statistics | Aggregator | Firms that train their people on AI tools report substantially better results than firms that buy licenses — “some analyses putting the gap at more than double.” An IT-vendor statistics roundup that names no underlying analysis. |
Second priority. Sources 20, 23, 24 and 25 are secondary. The Morgan Stanley capex comparison (20) is the strongest piece of evidence in the bubble argument and currently rests on a financial-advisory blog — link Morgan Stanley’s own research or established coverage of it. The training-payoff gap (25) traces to a vendor statistics roundup that names no study; find the study or soften the claim to a qualitative one. Census BTOS (21) is the right agency but the wrong URL — point to the specific data release containing the 8.8% figure. Wikipedia and the Goldman Sachs history page (23, 24) support uncontroversial historical framing and are the lowest risk of the four.
Articles With No External Sources
These three posts are positioning, philosophy, and process. They assert no third-party statistics, so nothing here needs citing — but any future claim added to them will.
AI Without the Anxiety
April 8, 2026 · Responsible AI & data security practices
Your AI Automation Launch: What to Expect, Step by Step
April 5, 2026 · Six-phase engagement process
The Pressure Is Real. The Solution Is Here.
April 5, 2026 · Market conditions for local business