🎯 called-it v1.0.0 · Every · 2026 · graded 2026-10-24 · jev-1.14.0

Every: 2026 predictions, graded

22 claims, each graded at three bars against evidence from 2026-01-01 to 2026-10-24. The strict bar is what a hostile critic would require, the lenient bar is the weakest reading a critic would still concede, and the gap between them is how much the grade depends on who is grading.

Strict

1 met · 5 not met · 16 unresolved

Standard

32%

7 met · 3 not met · 12 unresolved

Lenient

55%

12 met · 0 not met · 10 unresolved

Generosity gap

50 pts

lenient minus strict

Ladder

#	Claim	Author	Strict	Standard	Lenient	Best evidence
c30	And I think, shockingly, the most addictive technology of 2026—the narrative we might be talking about at the…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	✓ met provisional	T2Scientific American
c8	A mass-market robotic pet succeeds by the end of 2026.	Willie Williams	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T2BriefGlance
c14	agentic development will be maybe by the end of the year inside of the app that you're actually using.	Brandon Gell	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T1Microsoft 365 Blog
c6	Kate expects most companies to have at least one dedicated role focused on building and maintaining internal…	Kate Lee	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T1IBM Newsroom
c3	Natalia thinks the most advanced companies will have agentic workflows embedded in daily operations by year’s…	Natalia Quintero	○ unresolved provisional	✓ met provisional	✓ met provisional	T1Anthropic
c5	Dan predicts a third kind of engineer: agentic engineers who fully delegate implementation to AI and move up…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T1Lightcast (Stanford AI Index 2026 data)
c1	Anything a user can do in an app, an agent can also do. Then anything the app can do, the agent can do. Event…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	✓ met provisional	T1Microsoft 365 Blog
c28	I would guess that they will find it valuable this year. Um specifically because coding agents are good enoug…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	✓ met provisional	T1Microsoft Tech Community (Excel Blog)
c41	I don’t think any of the horses that are in the race will particularly stumble. I don’t think you’ll look up…	Reid Hoffman	✗ not met provisional	○ unresolved provisional	○ unresolved provisional	T2Decrypt
c12	What we're doing next year is um we're going to build this like unified memory and context layer where we hav…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T4Every
c32	But I think next year is going to be much more negative on AI than this year, in general popular discourse.	Reid Hoffman	○ unresolved provisional	✓ met provisional	✓ met provisional	T1Pew Research Center
c2	2026 is when agents go mainstream.	Kieran Klaassen	✗ not met provisional	✗ not met provisional	○ unresolved provisional	T1Microsoft 365 Blog
c7	There’s a whole class of highly creative, highly visual people who have been held back from making full exper…	Dan Shipper	✓ met provisional	✓ met provisional	✓ met provisional	T1AI in Design Report (Designer Fund / Foundation Capital)
c10	AGI, however you define that, is a lot more complicated and hard than we thought. And we will not have AGI in…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	○ unresolved provisional	T2Business Today
c29	So I think that's going to actually happen a lot more in in 2026 is is we're going to um use training approac…	Dan Shipper	○ unresolved provisional	✓ met provisional	✓ met provisional	T1Anthropic
c26	I think we will definitely see um layoffs that companies blame on AI um in 2026	Dan Shipper	○ unresolved provisional	✓ met provisional	✓ met provisional	T1Challenger, Gray & Christmas
c27	and I think mostly they will not be actually AI related they will be about the underlying business but they'l…	Dan Shipper	○ unresolved provisional	○ unresolved provisional	✓ met provisional	T1Stanford Digital Economy Lab
c13	I think we'll hire next year, but I don't think we're going to hire a ton. Like the whole thing is like keep…	Dan Shipper	✗ not met provisional	○ unresolved provisional	○ unresolved provisional	T4Every
c11	if you are an engineer who wants to incorporate AI into your process but doesn't want to reinvent the skill o…	Dan Shipper	○ unresolved provisional	✓ met provisional	✓ met provisional	T1FRED (Indeed Hiring Lab)
c18	I think midterms are going to be like kind of scary. Um because I think there's going to be deep fakes, which…	Brandon Gell	○ unresolved provisional	✓ met provisional	✓ met provisional	T2CNN (via KRDO)
c22	if you then zoom out though that kind that style of engineering is going to be on a pretty slow and steady de…	Dan Shipper	✗ not met provisional	✗ not met provisional	○ unresolved provisional	T1FRED (Indeed Hiring Lab)
c31	And I think that’s going to hamper their product direction and velocity in a way that—if we look back in a co…	Dan Shipper	✗ not met provisional	✗ not met provisional	✓ met provisional	T1Stack Overflow

Claims

c30. And I think, shockingly, the most addictive technology of 2026—the narrative we might be talking about at the end of the year—is how addict…

And I think, shockingly, the most addictive technology of 2026—the narrative we might be talking about at the end of the year—is how addictive it is to just make things.

Dan Shipper, 2026-01-05, Transcript: Reid Hoffman Makes Five Predictions About AI In 2026

Context: By end of 2026 the dominant tech-addiction narrative will be about compulsive making with AI coding tools (he adds that everyone will be prompting Claude Code), not social media.

Strict ○ unresolved
By the end of 2026 the dominant tech-addiction narrative is about how addictive it is to make things with AI, not about social media or AI companions.
no evidence gathered
Standard ○ unresolved
Multiple major outlets in 2026 described compulsive making with AI coding tools as an addiction or a defining behaviour of the year.
P(met) 0.35 · P(contradicted) 0.16 · evidence strength 1.45 of 2
Lenient ✓ met
At least one major outlet in 2026 described building with AI coding tools as addictive.
P(met) 0.74 · P(contradicted) 0.07 · evidence strength 1.44 of 2

T2neutral [lenient, standard] · 2026-01-31 · Scientific American: “People have cited "Claudeholism: The New Addiction of the Century" or asserted that "Claude Code is the opium of the permanent underclass" in reference to the software's ability to remove the pain of drudgery.”
T2neutral [lenient, standard] · 2026-02-26 · Bloomberg Law: “Claude Code Is Causing the Great Productivity Panic of 2026 (headline; body paywalled, framed as productivity pressure rather than addiction)”
T2supports [lenient] · 2026-04-21 · LeadDev: “The "just one more prompt" era is here ... "provides constant, unpredictable micro-rewards" ... "biologically impossible to wind down" (trade outlet, not a general-interest major outlet)”

c8. A mass-market robotic pet succeeds by the end of 2026.

A mass-market robotic pet succeeds by the end of 2026.

Willie Williams, 2025-12-23, Every's 2026 Predictions

Context: The wildcard prediction, stated as the editor's rendering of Williams's view. He names Moflin as the one to watch. 'Succeeds' is undefined.

Strict ○ unresolved
A mass-market robotic pet was a clear commercial success in 2026: at least 100,000 units sold in the year.
P(met) 0.05 · P(contradicted) 0.29 · evidence strength 1.67 of 2
Standard ○ unresolved
A consumer robotic pet reached mass-market retail in 2026 and major outlets reported strong sales.
P(met) 0.08 · P(contradicted) 0.22 · evidence strength 1.61 of 2
Lenient ○ unresolved
A consumer robotic pet was widely covered as a commercial hit in 2026.
P(met) 0.02 · P(contradicted) 0.55 · evidence strength 1.11 of 2

T2neutral [strict, standard] · 2025-09-19 · Sherwood News: “Baseline (2025): The company is hoping to sell a total of 7,000 units in the UK and US by the end of March 2026 ... The G-Shock-maker sold some 10,000 of the pet robots in its home country as of the end of May”
T2contradicts [standard, lenient] · 2026-06-26 · New Straits Times (AFP): “Sony is halting sales of its robotic puppy "aibo" in Japan ... Sony declined to give a total sales figure for the aibo model released in 2018.”
T2contradicts [strict] · 2026-10-14 · BriefGlance: “Moflin has sold over 20,000 units in Japan since its November 2024 debut.”
T2neutral [standard, lenient] · 2026-10-14 · BriefGlance: “Pink Moflin retains the same emotional algorithm and personality capabilities as existing models, priced at $429. ... available in the U.S. starting October 1, 2026”
T4neutral [strict, standard] · 2025-12-23 · Every: “Baseline (2025, the prediction itself): About 85 percent of people viscerally hate it and want to throw it out ... 10 percent are neutral ... 5 percent absolutely love it. ... 'nailed the experience of threading the unc…”

c14. agentic development will be maybe by the end of the year inside of the app that you're actually using.

agentic development will be maybe by the end of the year inside of the app that you're actually using.

Brandon Gell, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: By the end of 2026 users will be able to request features or bug fixes from inside an app and have an agent write and ship the code; unclear whether this means any shipping product or broad availability.

Strict ○ unresolved
By the end of 2026 users of a widely used app can ask for a feature or a fix inside the app and have an agent write and ship the code into the app they are using.
P(met) 0.02 · P(contradicted) 0.51 · evidence strength 1.04 of 2
Standard ○ unresolved
By the end of 2026 at least one shipping product lets users request features or fixes in-app and has an agent implement them inside the app.
P(met) 0.05 · P(contradicted) 0.17 · evidence strength 1.23 of 2
Lenient ○ unresolved
By the end of 2026 at least one product shipped or publicly beta-tested in-app agentic development.
P(met) 0.28 · P(contradicted) 0.04 · evidence strength 1.02 of 2

T1neutral [strict, standard] · 2026-04-22 · Microsoft 365 Blog: “can take multi-step, app-native actions directly in your documents, worksheets, and presentations. (agents act in-app; no code shipped into the app)”
T2supports [lenient, standard] · 2026-06-24 · TechCrunch: “adding a feature to help users create custom plug-ins, like layout generators or vector path tracers, with prompts.”

c6. Kate expects most companies to have at least one dedicated role focused on building and maintaining internal AI systems by 2026—“someone wh…

Kate expects most companies to have at least one dedicated role focused on building and maintaining internal AI systems by 2026—“someone whose job is to continuously optimize how the organization uses AI.”

Kate Lee, 2025-12-23, Every's 2026 Predictions

Context: Prediction 2 of 3. Editor's paraphrase plus a direct quote. 'Most companies' is the checkable element; the bars will need a survey or job-posting measure.

Strict ○ unresolved
By 2026 most companies have at least one dedicated role for building and maintaining internal AI systems.
P(met) 0.17 · P(contradicted) 0.06 · evidence strength 1.07 of 2
Standard ○ unresolved
By 2026 most large companies have a dedicated internal-AI role, or the role is described as standard practice at large companies.
P(met) 0.62 · P(contradicted) 0.07 · evidence strength 1.27 of 2
Lenient ○ unresolved
Dedicated internal-AI roles grew substantially in 2026 and are common at large companies.
P(met) 0.48 · P(contradicted) 0.07 · evidence strength 1.89 of 2

T1supports [lenient] · 2026-04-13 · Lightcast (Stanford AI Index 2026 data): “That's a more than 280% increase in a single year, representing nearly 90,000 job postings in the US alone. (agentic AI skills in postings, 2024 to 2025)”
T1supports · 2026-05-04 · IBM Newsroom: “76% of surveyed organizations have a Chief AI Officer (CAIO) in 2026, up from just 26% in 2025.”
T1neutral [strict] · 2026-05-04 · IBM Newsroom: “a survey of 2,000 CEOs and equivalent senior leaders across 33 geographies and 21 industries (sample skews to large organizations; fieldwork February to April 2026)”
T1neutral [standard, lenient] · 2026-07-20 · Gallup: “Forty-seven percent of U.S. employees now say their organization has integrated AI tools ... up from 41%”

c3. Natalia thinks the most advanced companies will have agentic workflows embedded in daily operations by year’s end

Natalia thinks the most advanced companies will have agentic workflows embedded in daily operations by year’s end

Natalia Quintero, 2025-12-23, Every's 2026 Predictions

Context: Prediction 1 of 3. Editor's paraphrase of Quintero (Every's head of consulting). Leading companies run agentic workflows in day-to-day operations by end of 2026, with finance named as the example industry.

Strict ○ unresolved
By the end of 2026 the most advanced companies run agentic workflows in daily operations, not pilots, with finance leading.
P(met) 0.22 · P(contradicted) 0.16 · evidence strength 1.41 of 2
Standard ✓ met
By the end of 2026 several large companies run agentic workflows in daily operations.
P(met) 0.73 · P(contradicted) 0.04 · evidence strength 1.96 of 2
Lenient ✓ met
In 2026 at least one large company publicly moved an agentic workflow from pilot into daily operations.
P(met) 0.80 · P(contradicted) 0.02 · evidence strength 1.92 of 2

T1supports [standard, lenient] · 2026-02-05 · Anthropic: “autonomously closed 13 issues and assigned 12 issues to the right team members in a single day (Rakuten customer quote)”
T1neutral [strict, standard] · 2026-05-27 · Stack Overflow: “63% of technologists still rarely or never let agents run entirely on autopilot.”
T2supports · 2026-02-09 · InvestmentNews: “At BNY, 134 software-based "digital employees" now work alongside the bank's roughly 48,100 human staff”
T2supports · 2026-02-09 · InvestmentNews: “Goldman has already rolled out an autonomous AI coder called Devin to its engineering teams.”
T2supports [standard, lenient] · 2026-09-01 · FM Magazine (reporting McKinsey State of AI 2026): “The share of large companies scaling AI agents in one or more functions increased from 27% to 40% ... while similar adoption among smaller organisations remained flat (22%).”

c5. Dan predicts a third kind of engineer: agentic engineers who fully delegate implementation to AI and move up a level to managing and orches…

Dan predicts a third kind of engineer: agentic engineers who fully delegate implementation to AI and move up a level to managing and orchestrating agents.

Dan Shipper, 2025-12-23, Every's 2026 Predictions

Context: Prediction 2 of 3, 'New roles are emerging'. Editor's paraphrase. A recognised job category of engineers who direct agents rather than write code emerges in 2026.

Strict ○ unresolved
In 2026 the agentic engineer, who delegates implementation to AI and manages agents instead of writing code, became a recognised third kind of engineering role.
P(met) 0.02 · P(contradicted) 0.07 · evidence strength 1.00 of 2
Standard ○ unresolved
In 2026 job postings or major-outlet coverage recognised engineers who direct agents rather than write code as a distinct role.
P(met) 0.56 · P(contradicted) 0.07 · evidence strength 1.70 of 2
Lenient ○ unresolved
In 2026 at least one named company hired for, or a widely read piece described, engineers whose job is directing agents rather than writing code.
P(met) 0.65 · P(contradicted) 0.04 · evidence strength 1.60 of 2

T1supports [standard, lenient] · 2026-04-13 · Lightcast (Stanford AI Index 2026 data): “grew from just 0.06% of postings in 2024 to 0.23% in 2025. ... That's a more than 280% increase in a single year, representing nearly 90,000 job postings in the US alone.”
T1neutral [standard, lenient] · 2026-05-27 · Stack Overflow: “This new survey shows 59% use agents at work at any frequency, compared to 31% in the 2025 Developer Survey.”
T1neutral [strict, standard] · 2026-10-06 · Stack Overflow: “73% of respondents who indicate using AI coding assistants/agents use them daily ... The agents of choice are Claude Code (66%) and GitHub Copilot (59%).”
T2supports [standard, lenient] · 2026-09-01 · Talentify (Booz Allen Hamilton posting): “Agentic AI Engineer (job title, Booz Allen Hamilton, McLean, Virginia; aggregator listing, not fetched)”
T4supports [lenient] · 2026-05-24 · Lenny's Podcast (Dan Shipper), via Podscripts: “the whole forward deployed engineer concept, I think is for real. (Shipper's later framing of the agent-directing role; not the 'agentic engineer' label)”

c1. Anything a user can do in an app, an agent can also do. Then anything the app can do, the agent can do. Eventually, anything a developer ca…

Anything a user can do in an app, an agent can also do. Then anything the app can do, the agent can do. Eventually, anything a developer can do—changing logic, fixing bugs, customizing behavior—will be something the agent handles on the user’s behalf.

Dan Shipper, 2025-12-23, Every's 2026 Predictions

Context: Prediction 1 of 3, 'Agents become infrastructure'. A staged claim about how software gets built; the first stage (agents can do anything a user can in an app) is the checkable 2026 part, the 'eventually' stage has no date.

Strict ○ unresolved
In 2026 agents can do anything a user can do in several major apps, and at least one shipping product already lets agents do what a developer does: change the app's own logic for the user.
P(met) 0.07 · P(contradicted) 0.35 · evidence strength 1.70 of 2
Standard ○ unresolved
In 2026 agents can do anything a user can do in several major apps.
P(met) 0.62 · P(contradicted) 0.06 · evidence strength 1.97 of 2
Lenient ✓ met
In 2026 at least one major app beyond coding tools shipped an agent that can do most of what a user can do in it.
P(met) 0.73 · P(contradicted) 0.02 · evidence strength 1.95 of 2

T1supports [standard, lenient] · 2026-04-22 · Microsoft 365 Blog: “agentic capabilities in Word, Excel, and PowerPoint are now generally available. ... can take multi-step, app-native actions directly in your documents, worksheets, and presentations.”
T2supports [standard, lenient] · 2026-02-25 · AlternativeTo: “Once configured, agents run without manual prompting, using triggers or schedules to keep work moving. ... Custom Agents are free through May 3, 2026, then will require Notion credits starting May 4, 2026.”
T2neutral [strict, standard] · 2026-02-26 · Phone Scoop: “you can ask Gemini to perform multi-step tasks on its own on your phone ... The feature is strictly limited to select apps and devices for now”
T2supports [strict] · 2026-06-24 · TechCrunch: “adding a feature to help users create custom plug-ins, like layout generators or vector path tracers, with prompts. (closest shipped example of an agent changing an app's behaviour for the user)”

c28. I would guess that they will find it valuable this year. Um specifically because coding agents are good enough that once they make their wa…

I would guess that they will find it valuable this year. Um specifically because coding agents are good enough that once they make their way into Excel in a widely distributed way

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Non-technical knowledge workers (the example is a deal-desk analyst living in Excel) will find AI genuinely valuable at work in 2026 once coding-agent capability is widely distributed inside Excel.

Strict ○ unresolved
In 2026 coding agents reached Excel in a widely distributed way and non-technical knowledge workers found AI genuinely valuable at work.
P(met) 0.68 · P(contradicted) 0.03 · evidence strength 1.98 of 2
Standard ○ unresolved
In 2026 agent capability shipped broadly in Excel and reported value among non-technical workers rose.
P(met) 0.52 · P(contradicted) 0.04 · evidence strength 1.96 of 2
Lenient ✓ met
In 2026 Microsoft shipped agent or coding capability in Excel to general availability.
P(met) 0.98 · P(contradicted) 0.01 · evidence strength 1.98 of 2

T1supports · 2026-01-27 · Microsoft Tech Community (Excel Blog): “Agent Mode in Excel is now generally available in Excel on Windows and is rolling out to Excel on Mac”
T1supports · 2026-04-22 · Microsoft 365 Blog: “the default experience for customers with Microsoft 365 Copilot and Microsoft 365 Premium subscriptions. ... Over the last month, our data already shows increased engagement, retention, and satisfaction. (table: Excel e…”
T1supports [strict, standard] · 2026-05-20 · Gallup: “65% of employees working in organizations that have implemented AI say it has had a positive effect on their productivity and efficiency at work, including 17% who say the impact has been 'extremely positive.'”
T1supports [standard] · 2026-07-20 · Gallup: “More than half of U.S. workers (52%) now use AI in their role, with 30% using it frequently ... The highest productivity ratings come from employees using AI for coding assistance and automation”

c41. I don’t think any of the horses that are in the race will particularly stumble. I don’t think you’ll look up and go, Wow, I thought Cursor…

I don’t think any of the horses that are in the race will particularly stumble. I don’t think you’ll look up and go, Wow, I thought Cursor was fantastic—and it’s just gone.

Reid Hoffman, 2026-01-05, Transcript: Reid Hoffman Makes Five Predictions About AI In 2026 · guest, not the subject

Context: None of the current coding-agent leaders will collapse or exit the market during 2026. Hoffman: none of the coding-agent leaders will stumble. Pinned because it directly contradicts Shipper's own Cursor call (c31).

Strict ✗ not met
None of the coding-agent leaders stumbled in 2026: every one is still operating independently with no major shutdown, distressed sale or collapse in revenue.
P(met) 0.11 · P(contradicted) 0.96 · evidence strength 1.87 of 2
Standard ○ unresolved
None of the coding-agent leaders shut down, was acquired in distress or lost most of its users or revenue in 2026.
P(met) 0.38 · P(contradicted) 0.40 · evidence strength 1.92 of 2
Lenient ○ unresolved
None of the coding-agent leaders shut down or was sold in distress in 2026.
P(met) 0.35 · P(contradicted) 0.32 · evidence strength 1.92 of 2

T2supports · 2026-03-12 · CFOtech: “Replit has raised USD $400 million in a funding round that values the AI coding company at USD $9 billion,”
T2supports · 2026-06-09 · Dealroom: “has surpassed $4 billion in annualised revenue as of early June 2026 ... The run-rate is up from $3 billion in late April 2026 and $2 billion in February 2026.”
T2contradicts [strict] · 2026-06-16 · Decrypt: “The all-stock deal would make Cursor a wholly owned SpaceX subsidiary (Cursor no longer independent; not a distressed sale)”
T2supports [standard, lenient] · 2026-06-16 · Decrypt: “SpaceX signed a $60 billion merger agreement with Anysphere and its subsidiary X67 ... was reportedly raising a $2 billion funding round that would have valued it at $50 billion.”
T2supports · 2026-08-13 · TechNode: “has raised $400 million in Series C funding at a $13.3 billion valuation.”
T2supports · 2026-09-09 · Channel Insider: “raised more than $2 billion in a Series E funding round, sending its valuation to $48 billion. ... run-rate revenue increased from $492 million in May to nearly $900 million.”

c12. What we're doing next year is um we're going to build this like unified memory and context layer where we have um we have a bunch of uh bas…

What we're doing next year is um we're going to build this like unified memory and context layer where we have um we have a bunch of uh basically we can collect everything everything that you care about and who you are into one place and all of the apps and all the content can take that into account as they work with you.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Every will ship a unified memory and context layer across its apps in 2026; this is a company plan rather than an industry forecast.

Strict ○ unresolved
Every shipped a unified memory and context layer shared across its apps to users in 2026.
P(met) 0.02 · P(contradicted) 0.05 · evidence strength 0.96 of 2
Standard ○ unresolved
Every publicly launched a cross-app memory or context feature in 2026, as a release or a customer beta.
P(met) 0.04 · P(contradicted) 0.52 · evidence strength 1.01 of 2
Lenient ○ unresolved
Every publicly showed progress on a cross-app memory layer in 2026.
P(met) 0.43 · P(contradicted) 0.37 · evidence strength 1.13 of 2

T4neutral · 2025-09-26 · Every: “Baseline (2025, plan): working on a platform layer that unifies memories from each part of the ecosystem.”
T4neutral [standard, lenient] · 2026-03-29 · Every: “All notes are also accessible via Monologue's API, CLI, and model-context protocol (MCP), (cross-app access to notes, not a shared memory layer)”
T4supports [lenient] · 2026-04-24 · Every: “adding memory to its API agents ... Memory is live now through the API (it's not inside the app yet, but stay tuned). (Spiral only)”
T4supports [lenient] · 2026-05-07 · Every: “Each Spiral subscriber also gets their own personal memory store. ... lives in an Anthropic-hosted global memory store.”

c32. But I think next year is going to be much more negative on AI than this year, in general popular discourse.

But I think next year is going to be much more negative on AI than this year, in general popular discourse.

Reid Hoffman, 2026-01-05, Transcript: Reid Hoffman Makes Five Predictions About AI In 2026 · guest, not the subject

Context: Popular discourse about AI in 2026 will be much more negative than in 2025. Hoffman: 2026 discourse much more negative on AI. Pinned as a guest claim for contrast with the hosts' optimism.

Strict ○ unresolved
Popular discourse on AI was much more negative in 2026 than in 2025.
P(met) 0.35 · P(contradicted) 0.46 · evidence strength 1.83 of 2
Standard ✓ met
Popular discourse on AI was more negative in 2026 than in 2025.
P(met) 0.95 · P(contradicted) 0.03 · evidence strength 1.94 of 2
Lenient ✓ met
Major outlets in 2026 described a backlash or a souring public mood toward AI.
P(met) 0.93 · P(contradicted) 0.02 · evidence strength 1.98 of 2

T1supports · 2026-07-28 · Gallup: “has increased from 31% in 2025 to 39% in 2026. (share saying AI does more harm than good; survey May 4-11, 2026)”
T1supports [standard, lenient] · 2026-07-28 · Gallup: “down slightly from 31% in 2025 (trust in businesses to use AI responsibly, now 27%) ... up from 73% in 2025 and 75% before that (79% expect fewer U.S. jobs)”
T1supports [standard, lenient] · 2026-08-18 · Pew Research Center: “52% of Americans say they are more concerned than excited about the increased use of AI in daily life (trend table: 2025 50%, 2026 52%; survey June 22-28, 2026)”
T1contradicts [strict] · 2026-08-18 · Pew Research Center: “52% of Americans say they are more concerned than excited about the increased use of AI in daily life (only 2 points above 2025's 50%; under-30s: For the first time, a majority of adults under 30 (55%) now say they're m…”
T2supports [lenient] · 2026-04-24 · The New Republic: “If there was any doubt over the brewing public backlash to this technology, the last few weeks have erased it. ... a rapidly growing populist backlash toward AI”
T2supports [lenient] · 2026-05-17 · Axios: “An AI hate wave is here (headline; page not fetchable)”

c2. 2026 is when agents go mainstream.

2026 is when agents go mainstream.

Kieran Klaassen, 2025-12-23, Every's 2026 Predictions

Context: Prediction 1 of 3. Editor's one-line rendering of Klaassen's view ('Kieran puts it simply'), not a direct quote. 'Mainstream' is undefined; the bars will need an adoption measure.

Strict ✗ not met
2026 is the year agents went mainstream: a majority of US adults or US workers used an AI agent during the year.
P(met) 0.09 · P(contradicted) 0.21 · evidence strength 1.47 of 2
Standard ✗ not met
In 2026 agent features reached hundreds of millions of people and the year was described as the one agents went mainstream.
P(met) 0.09 · P(contradicted) 0.76 · evidence strength 1.89 of 2
Lenient ○ unresolved
In 2026 agent features became generally available, not waitlisted, in at least two of the most-used consumer AI products.
P(met) 0.57 · P(contradicted) 0.25 · evidence strength 1.88 of 2

T1supports [standard, lenient] · 2026-04-22 · Microsoft 365 Blog: “agentic capabilities in Word, Excel, and PowerPoint are now generally available. ... They are also available to users with Microsoft 365 Personal and Family plans.”
T1neutral [strict] · 2026-05-27 · Stack Overflow: “This new survey shows 59% use agents at work at any frequency, compared to 31% in the 2025 Developer Survey. (technologists, n=1,100)”
T1neutral [strict] · 2026-07-20 · Gallup: “More than half of U.S. workers (52%) now use AI in their role, with 30% using it frequently ... Fifteen percent use it daily. (AI use generally, not agents)”
T2contradicts [standard, lenient] · 2026-02-26 · Phone Scoop: “The feature is strictly limited to select apps and devices for now ... The feature is launching first on the Google Pixel 10 series and Samsung Galaxy S26 series.”
T2contradicts [standard, lenient] · 2026-03-20 · 9to5Mac: “While that release turned out to be feature-packed, the new Siri features aren't present. ... Apple would instead target iOS 26.5 and iOS 27.”

c7. There’s a whole class of highly creative, highly visual people who have been held back from making full experiences because they can’t code…

There’s a whole class of highly creative, highly visual people who have been held back from making full experiences because they can’t code. Now they’re going to be able to code.

Dan Shipper, 2025-12-23, Every's 2026 Predictions

Context: Prediction 3 of 3, 'Designers become superheroes'. Designers and other non-coders ship working software with AI tools in 2026.

Strict ✓ met
In 2026 a large share of designers and other visual creatives shipped working software without writing code.
P(met) 0.71 · P(contradicted) 0.03 · evidence strength 1.06 of 2
Standard ✓ met
In 2026 designers shipped working apps with AI tools at scale.
P(met) 0.78 · P(contradicted) 0.02 · evidence strength 1.96 of 2
Lenient ✓ met
Major outlets in 2026 reported designers building and shipping software with AI coding tools as an established trend.
P(met) 0.62 · P(contradicted) 0.06 · evidence strength 1.95 of 2

T1supports [standard, lenient] · 2026-06-24 · Figma: “the number of designers participating in development doubled to 41% (from 21%; 8,403 survey responses)”
T1supports [strict, standard] · 2026-10-06 · AI in Design Report (Designer Fund / Foundation Capital): “Half of respondents have pushed AI-generated code to production. (over 900 designers surveyed)”
T2supports [standard, lenient] · 2026-03-12 · CFOtech: “Zillow has around 600 Replit seats, and employees created more than 7,000 apps over the past year”
T2supports [lenient] · 2026-06-24 · TechCrunch: “The company is now adding code layers directly to the collaborative canvas ... adding a feature to help users create custom plug-ins, like layout generators or vector path tracers, with prompts.”
T2supports [standard] · 2026-08-13 · TechNode: “users have created more than 60 million projects since its launch in November 2024 ... Lovable-built applications receive more than 900 million visits each month.”

c10. AGI, however you define that, is a lot more complicated and hard than we thought. And we will not have AGI in 2027.

AGI, however you define that, is a lot more complicated and hard than we thought. And we will not have AGI in 2027.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: No AGI by 2027; the horizon is a year past the target year and AGI is explicitly left undefined, so a 2026 check can only falsify it if AGI is widely agreed to have arrived.

Strict ○ unresolved
We will not have AGI in 2027.
P(met) 0.49 · P(contradicted) 0.15 · evidence strength 1.88 of 2
Standard ○ unresolved
We will not have AGI in 2027.
P(met) 0.10 · P(contradicted) 0.16 · evidence strength 1.85 of 2
Lenient ○ unresolved
We will not have AGI in 2027.
P(met) 0.13 · P(contradicted) 0.14 · evidence strength 1.82 of 2

T2supports · 2026-09-07 · Business Today: “Huang: "AGI has arrived" ... Gary Marcus: "Huang gave no evidence and no definitions." ... "By conventional definitions, Astra still falls short."”
T2neutral · 2026-09-07 · Information Age (ACS): “Greg Brockman: "I think it's not unreasonable to feel that we are now in the AGI era." ... "I think it might be about this model."”
T2supports · 2026-09-07 · Information Age (ACS): “ARC Prize Foundation: "while we believe Astra represents meaningful progress towards generalisation, we are not claiming that it is AGI." ... Altman called AGI "a very poorly defined" and "irrelevant marketing term"”

c29. So I think that's going to actually happen a lot more in in 2026 is is we're going to um use training approaches and architectures that all…

So I think that's going to actually happen a lot more in in 2026 is is we're going to um use training approaches and architectures that allow them to like go off and do stuff.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: In 2026 labs will adopt training approaches and architectures that give agents more independence (loosening obedience-style alignment); the speaker calls it 'a very vague prediction'.

Strict ○ unresolved
In 2026 labs adopted training approaches and architectures that let agents go off and do things on their own for much longer.
P(met) 0.03 · P(contradicted) 0.30 · evidence strength 1.09 of 2
Standard ✓ met
In 2026 agents' autonomous task horizons grew substantially, or major labs shipped models marketed on longer independent operation.
P(met) 0.79 · P(contradicted) 0.06 · evidence strength 1.96 of 2
Lenient ✓ met
In 2026 at least one major lab shipped a model marketed on longer autonomous agent operation.
P(met) 0.95 · P(contradicted) 0.01 · evidence strength 1.99 of 2

T1neutral [strict, standard] · 2026-01-29 · METR: “The post-2023 doubling-time is 131 days under TH1.1, compared to 165 days under TH1 (top model in table: Claude Opus 4.5 at 320 minutes, interval 170-729; measured on 2023-2025 models)”
T1supports [standard, lenient] · 2026-02-05 · Anthropic: “spin up multiple agents that work in parallel as a team and coordinate autonomously ... sustains agentic tasks for longer ... stays productive over longer sessions”
T1supports [lenient] · 2026-02-05 · Anthropic: “Claude can use compaction to summarize its own context and perform longer-running tasks without bumping up against limits”
T1supports [standard, lenient] · 2026-02-23 · OpenAI Developers blog: “Codex ran for about 25 hours uninterrupted, used about 13M tokens, and generated about 30k lines of code. (GPT-5.3-Codex experiment, second lab marketing long autonomous operation)”
T1neutral [strict, standard] · 2026-09-08 · METR: “Measurements above 16 hrs are unreliable with our current task suite. ... Clarified that this page is no longer actively updated. (no 2026 within-year doubling measurement published)”

c26. I think we will definitely see um layoffs that companies blame on AI um in 2026

I think we will definitely see um layoffs that companies blame on AI um in 2026

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: In 2026 companies will announce layoffs and publicly attribute them to AI.

Strict ○ unresolved
In 2026 several large companies announced layoffs and blamed them on AI.
P(met) 0.49 · P(contradicted) 0.05 · evidence strength 1.95 of 2
Standard ✓ met
In 2026 multiple large companies announced layoffs and blamed them on AI.
P(met) 0.91 · P(contradicted) 0.02 · evidence strength 1.98 of 2
Lenient ✓ met
In 2026 at least one large company explicitly blamed AI for layoffs.
P(met) 0.93 · P(contradicted) 0.01 · evidence strength 1.93 of 2

T1supports · 2026-07-01 · Challenger, Gray & Christmas: “In June, Artificial Intelligence led all reasons for job cuts, with 14,029 announced during the month, or 31%. ... So far this year, AI has been cited in 101,743 job cut announcements, approximately 23% of all cuts.”
T2supports · 2026-02-27 · Fortune: “laying off nearly half its workforce, cutting 4,000 employees, down to just under 6,000 workers from over 10,000. ... We're already seeing that the intelligence tools we're creating and using, paired with smaller and fl…”
T2neutral [strict] · 2026-03-10 · Cape Argus (IOL): “about 9,238 of the 45,363 tech layoffs recorded worldwide in 2026 so far (RationalFX tally; lists Pinterest 675 and eBay 800 without stating how each tied cuts to AI)”
T2supports · 2026-03-11 · Bloomberg: “Atlassian to Reduce 1,600 Jobs in the Latest AI-Linked Cuts (headline; body paywalled)”
T2supports · 2026-04-15 · TheWrap: “laying off 16% of its global workforce, or about 1,000 employees. ... rapid advancements in artificial intelligence enable our teams to reduce repetitive work”

c27. and I think mostly they will not be actually AI related they will be about the underlying business but they'll use AI as a good excuse.

and I think mostly they will not be actually AI related they will be about the underlying business but they'll use AI as a good excuse.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Most 2026 layoffs blamed on AI will really be driven by underlying business problems, with AI used as cover; judging the true cause is inherently contestable.

Strict ○ unresolved
Most 2026 layoffs blamed on AI were really driven by the underlying business, with AI used as the excuse.
P(met) 0.02 · P(contradicted) 0.07 · evidence strength 0.97 of 2
Standard ○ unresolved
Reputable analysis in 2026 found that AI-attributed layoffs were largely driven by other factors.
P(met) 0.62 · P(contradicted) 0.10 · evidence strength 1.33 of 2
Lenient ✓ met
In 2026 economists or analysts publicly doubted that AI was the real cause of layoffs announced as AI-driven.
P(met) 0.90 · P(contradicted) 0.05 · evidence strength 1.84 of 2

T1neutral [strict, standard] · 2026-07-01 · Challenger, Gray & Christmas: “artificial intelligence continues to reshape how companies think about headcount (Andy Challenger; Challenger reports the stated reason and does not audit it)”
T1supports [standard, lenient] · 2026-08-12 · Stanford Digital Economy Lab: “The adjustment appears to operate primarily through reduced hiring of young workers rather than increased separations.”
T2supports [standard, lenient] · 2026-02-02 · Fortune: “it just doesn't seem like there's major macroeconomic effects here ... the technology to become a convenient scapegoat for CEOs. (Martha Gimbel, Yale Budget Lab; Challenger: 55,000 U.S. job cuts in the first 11 months o…”
T2supports [lenient] · 2026-05-03 · Fortune: “there's some AI washing where people are blaming AI for layoffs that they would otherwise do (Sam Altman)”

c13. I think we'll hire next year, but I don't think we're going to hire a ton. Like the whole thing is like keep it small.

I think we'll hire next year, but I don't think we're going to hire a ton. Like the whole thing is like keep it small.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Every will add staff in 2026 but not many, staying a small company; 'a ton' is undefined (they were at about 20 people).

Strict ✗ not met
Every hired in 2026 but stayed small, adding only a handful of people.
P(met) 0.07 · P(contradicted) 0.94 · evidence strength 1.06 of 2
Standard ○ unresolved
Every added staff in 2026 and remained a small company.
P(met) 0.67 · P(contradicted) 0.04 · evidence strength 1.08 of 2
Lenient ○ unresolved
Every remained a small company through 2026.
P(met) 0.38 · P(contradicted) 0.05 · evidence strength 1.05 of 2

T4neutral [strict, standard] · 2025-09-26 · Every: “Baseline (2025, consulting arm only): 15 full-time employees”
T4contradicts [strict] · 2026-05-24 · Lenny's Podcast (Dan Shipper), via Podscripts: “We're almost 30 people now. ... So we've doubled in size in the last year.”
T4supports [standard, lenient] · 2026-05-24 · Lenny's Podcast (Dan Shipper), via Podscripts: “We're almost 30 people now.”
T4contradicts [strict] · 2026-05-27 · Every: “In fact, since the GPT-3 days, we've grown from four people to around 30 people, and we're hiring more now. (about 10 above the roughly 20 at end-2025)”
T4supports [standard, lenient] · 2026-05-27 · Every: “In fact, since the GPT-3 days, we've grown from four people to around 30 people, and we're hiring more now.”

c11. if you are an engineer who wants to incorporate AI into your process but doesn't want to reinvent the skill of software engineering there's…

if you are an engineer who wants to incorporate AI into your process but doesn't want to reinvent the skill of software engineering there's going to be a home and a job for you for a very long time.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Traditional engineers who use AI as an accelerant will keep having jobs; 'a very long time' has no end date, so a 2026 check is only whether that job category visibly persisted.

Strict ○ unresolved
In 2026 engineers who use AI as an accelerant without reinventing their craft still had a home and a job: traditional software roles did not shrink meaningfully.
P(met) 0.31 · P(contradicted) 0.12 · evidence strength 1.65 of 2
Standard ✓ met
In 2026 traditional software-engineering roles remained widely available and were not in collapse.
P(met) 0.94 · P(contradicted) 0.07 · evidence strength 1.95 of 2
Lenient ✓ met
In 2026 traditional software engineers were still being hired at scale.
P(met) 0.95 · P(contradicted) 0.01 · evidence strength 1.98 of 2

T1neutral [strict, standard] · 2026-06-26 · FRED (Indeed Hiring Lab): “2026-06-26,73.4952491432 (about 26.5% below the Feb 2020 level; CIO reported 27.5% from an earlier vintage)”
T1supports · 2026-10-16 · FRED (Indeed Hiring Lab): “2026-10-16,78.8649883609 ... 2025-10-16,65.0270378284 (Software Development Job Postings on Indeed, index Feb 1 2020=100; up about 21% year over year)”
T2supports [lenient, standard] · 2026-07-13 · CIO: “Indeed has tracked a rise of 15% for software development job postings in the US ... software developer and software engineer job postings were the top roles for June (49k postings)”

c18. I think midterms are going to be like kind of scary. Um because I think there's going to be deep fakes, which used to be a really scary wor…

I think midterms are going to be like kind of scary. Um because I think there's going to be deep fakes, which used to be a really scary word and now nobody talks about nobody says deep fake anymore.

Brandon Gell, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: AI deepfakes will be a significant, alarming feature of the 2026 US midterm elections.

Strict ○ unresolved
AI deepfakes were a significant, widely covered part of the 2026 midterms.
P(met) 0.18 · P(contradicted) 0.08 · evidence strength 1.89 of 2
Standard ✓ met
Multiple documented deepfake incidents involving 2026 midterm candidates were covered by major outlets.
P(met) 0.95 · P(contradicted) 0.02 · evidence strength 1.98 of 2
Lenient ✓ met
At least one documented deepfake incident involving a 2026 midterm candidate was covered by a major outlet.
P(met) 0.97 · P(contradicted) 0.01 · evidence strength 1.94 of 2

T2supports [standard, lenient] · 2026-03-13 · CNN (via KRDO): “a real-looking but fake version of a Democratic candidate, fabricated with artificial intelligence. ... the disclosure text is small, mostly faint and confined to a bottom corner of the screen. (NRSC 85-second Talarico…”
T2supports · 2026-03-28 · Reuters (via Honolulu Star-Advertiser): “The Talarico ad, for instance, is one of three recent ads created by national Republicans that use deepfake technology”
T2supports [standard, lenient] · 2026-03-28 · Reuters (via Honolulu Star-Advertiser): “I just voted to keep the government shut down. They say it would hurt farmers, but I wouldn't know. (Collins campaign deepfake of Ossoff; Paxton campaign ad shows an AI-generated version of Cornyn dancing with Democrati…”
T2neutral [strict] · 2026-06-20 · Washington Examiner: “the Republican PAC Citizens for Sanity paid six figures for an ad campaign ... AI is almost exclusively used in attack ads against opponents (no view counts given)”
T2supports [standard, strict] · 2026-10-13 · WRDW: “The Senate Leadership Fund's website features dozens of AI-generated videos. ... South Carolina and Georgia, Ossoff's home state, have no laws on the books addressing AI-generated political ads.”

c22. if you then zoom out though that kind that style of engineering is going to be on a pretty slow and steady decline.

if you then zoom out though that kind that style of engineering is going to be on a pretty slow and steady decline.

Dan Shipper, 2025-12-23, AI & I: Four Predictions for How AI Will Change Software in 2026 (episode captions)

Context: Traditional software engineering with AI as an accelerant begins a slow, steady decline (newspaper-industry analogy); a slow decline is hard to confirm or deny within one year.

Strict ✗ not met
Demand for traditional, code-writing software engineering declined in 2026, with AI coding agents named as a cause.
P(met) 0.01 · P(contradicted) 0.98 · evidence strength 1.70 of 2
Standard ✗ not met
Demand for traditional software engineering declined in 2026.
P(met) 0.63 · P(contradicted) 0.97 · evidence strength 1.86 of 2
Lenient ○ unresolved
Reputable 2026 data or reporting described a decline in demand for traditional code-writing engineering roles.
P(met) 0.45 · P(contradicted) 0.42 · evidence strength 1.52 of 2

T1supports [lenient] · 2026-08-12 · Stanford Digital Economy Lab: “Employment among workers ages 22-25 in highly AI-exposed occupations now stands about 19% below where it would be. ... by this same measure, the shortfall was 15% at the July 2025 data vintage and is 19% as of June 2026.”
T1neutral [standard, lenient] · 2026-08-12 · Stanford Digital Economy Lab: “The divergence remains when we exclude technology firms and computer occupations. (no separate software-developer figure)”
T1contradicts [strict, standard] · 2026-10-16 · FRED (Indeed Hiring Lab): “2026-10-16,78.8649883609 ... 2025-10-16,65.0270378284 (postings higher than a year earlier, not lower)”
T2supports [standard, lenient] · 2026-07-13 · CIO: “down 22% from 2025, according to Dice (conflicts with Indeed's rise over the same period)”

c31. And I think that’s going to hamper their product direction and velocity in a way that—if we look back in a couple years—we might say: that…

And I think that’s going to hamper their product direction and velocity in a way that—if we look back in a couple years—we might say: that was an interesting era. It’s still a widely distributed piece of software, but it’s not the next-generation thing that we thought it was.

Dan Shipper, 2026-01-05, Transcript: Reid Hoffman Makes Five Predictions About AI In 2026

Context: Cursor is the coding-agent leader most likely to stumble, caught between the IDE-plus-AI paradigm and the Claude Code paradigm; the stated horizon is 'a couple years', so an end-2026 check is partial.

Strict ✗ not met
Cursor stumbled in 2026: its growth stalled or reversed, or it took a down round, a distressed sale or a strategy-driven leadership exit.
P(met) 0.11 · P(contradicted) 0.94 · evidence strength 1.88 of 2
Standard ✗ not met
In 2026 Cursor lost momentum to agent-first coding tools, with slowing growth or a strategic pivot away from the IDE.
P(met) 0.33 · P(contradicted) 0.90 · evidence strength 1.85 of 2
Lenient ✓ met
In 2026 Cursor was described as under competitive pressure from agent-first coding tools.
P(met) 0.78 · P(contradicted) 0.06 · evidence strength 1.77 of 2

T1neutral [standard, lenient] · 2026-10-06 · Stack Overflow: “The agents of choice are Claude Code (66%) and GitHub Copilot (59%). (Cursor not named among agents of choice)”
T2contradicts [strict, standard] · 2026-03-02 · Bloomberg: “AI Coding Startup Cursor Hits $2 Billion Annual Sales Rate (headline; body paywalled)”
T2contradicts [strict, standard] · 2026-06-09 · Dealroom: “The run-rate is up from $3 billion in late April 2026 and $2 billion in February 2026. ... the enterprise business reported to have tripled in Q1 2026 versus Q4 2025”
T2supports [lenient] · 2026-06-09 · Dealroom: “intensifying competition from GitHub Copilot, OpenAI and Google in the AI coding market. (Claude Code not named)”
T2contradicts [strict] · 2026-06-16 · Decrypt: “SpaceX signed a $60 billion merger agreement with Anysphere and its subsidiary X67 ... SpaceX secured an exclusive option in April to acquire Anysphere”