Open Future Forum’s November 2026 research collection brings together 14 full reports and an Executive AI Statistics registry. The subjects range from finance and security to founder pricing, investment decisions and transaction diligence. Each publication addresses a different decision, using a shared body of first-party evidence where the questions overlap.
The collection includes updated security, marketing and investor samples, alongside buying, founder and production measures carried forward from earlier editions. That distinction matters. A new publication date does not mean every figure represents a new month of activity, and a larger sample does not establish that the market has changed.
This article introduces the collection and explains where to start. The reports contain the detailed analysis, examples, questions and methods. Their common purpose is to help readers judge which evidence is relevant to a decision and how much weight it can reasonably bear.
What the November edition adds
The evidence cutoff is 31 October 2026. The underlying package contains 69 exports, with 11 byte-identical duplicates excluded. The analysis retains 58 distinct exports and deduplicates eligible answers separately for each question. There is no single response base that describes every finding in the collection.
The security instrument expands from a published October-compatible base of 151 to 207 unique respondents. The marketing impact question increases from 218 to 314. Investor ownership expands from 245 to 373 answers, while investor impact grows from 237 to 364. Those two investor questions have different bases because completion differs by question.
These are cumulative sample updates. Some people appear in more than one export, and the retained answer can change when a later eligible response is available. The added cohorts therefore cannot simply be added to the previous headline bases. The registry records the reconciliation so readers can distinguish additional coverage from a change in the people or answers being counted.
The analytical work also extends beyond updating percentages. November examines benefit qualification, delegated decisions, security revocation, accepted marketing output, operational handoffs and the evidence needed for investment or transaction claims. Those subjects receive separate treatment because the evidence needed by one reader may be insufficient for another.
A few findings worth opening the reports for
In the updated security sample, 142 of 207 respondents, or 69 percent, select securing AI agents and their access as a security problem. Separately, 74 of 207, or 36 percent, report a dedicated AI security budget line. Both questions allow multiple selections. The figures describe a reported concern and a funding route; subtracting them would not measure a security funding shortfall.
Marketing’s updated impact question produces a different kind of distinction. Doing the work of more people and creating content faster each receive 141 mentions among 314 respondents, or 45 percent. Knowing the customer better receives 130 mentions, or 41 percent. These are respondents’ assessments of where AI makes a difference. They are not measurements of payroll reductions, incremental revenue or attributable marketing profit.
Among investor impact answers, building better products receives 190 mentions from a base of 364, or 52 percent. Cutting costs receives 141 mentions, or 39 percent, and helping customers receives 133, or 37 percent. Respondents could select multiple options. A reported area of impact does not establish how much value was created, how it was measured or whether it persists.
The reports develop the implications of these findings without turning them into company performance benchmarks. The complete distributions, including less frequently selected answers and contradictory selections, belong alongside the headline numbers.
Start with the decision on your desk
The Executive AI Leverage Report is the starting point for readers responsible for several functions. It examines how an AI commitment changes as it moves between the people requesting investment, approving expenditure, operating the system and assessing the result. Its worked example helps readers examine competing interpretations of the same expansion proposal.
The CFO AI Leverage Report focuses on the conditions under which a benefit can enter an investment case. Time released, expenditure avoided and revenue gained require different supporting records. The report considers timing, incremental costs and the circumstances in which an accepted benefit should be revised or withdrawn.
The CEO AI Leverage Report addresses the allocation of executive attention. A chief executive cannot personally review every deployment change. The report considers which decisions can be delegated, which dependencies need executive attention and what developments should bring a delegated decision back for review.
These reports are complementary, but they are not interchangeable. A finance team assessing the treatment of a benefit has a narrower question than an executive team deciding whether several functions can support expansion. Choosing the report by that immediate question is more useful than assigning everyone the same reading list.
For teams responsible for running AI
The CISO AI Leverage Report examines whether an organization can end an agent’s authority across the systems it can reach. Removing one credential may leave other access routes or downstream effects unresolved. The report develops questions about termination, residual authority and restoration, using the updated security instrument as context.
The AI Leaders AI Leverage Report looks at the evidence required after a production system changes. Its audience includes people responsible for platforms, applications and release decisions. It distinguishes those responsibilities and considers how review effort, exceptions and operating costs affect a release assessment.
The AI Transformation Report concentrates on the receiving operation. A team may demonstrate a useful capability while the process expected to use it remains unprepared. The report asks what the receiving team must accept before the organization makes that capability part of its normal work.
Read together, these publications help separate a security question, a technical release question and an operating-process question. A positive answer in one area does not settle the others. Their examples are proposed ways to examine a situation, rather than evidence that a particular company has passed those tests.
For marketing leaders and founders
The CMO AI Leverage Report builds on the expanded marketing sample. Its focus is the relationship between production speed, work that is accepted and customer decisions. It also sets out an observation protocol for search and answer-engine visibility, where inconsistent queries or changing measurement conditions can make comparisons difficult to interpret.
The YC Founder AI Report approaches the commercial problem from the seller’s side. It examines what a customer is buying, what counts as an accepted result and how attempts, review and delivery costs affect the economics of that result. The founder pricing evidence is explicitly carried forward; the analysis does not present it as a new November pricing survey.
A marketing leader and a founder may both be interested in how much more work AI makes possible. Their next decisions can still differ considerably. The marketer may need to establish whether additional output improves a customer outcome. The founder may need to establish whether the contract and delivery economics support the promised service.
For investors, deal teams and legal leaders
The VC and CVC AI Investment Report asks what evidence could weaken an investment thesis. It distinguishes financial objectives from the strategic objectives a corporate venture investor may also pursue. The updated investor answers provide context; they do not establish venture returns, exit outcomes or the merits of a particular investment.
The Investor AI Report examines what a company should do when an AI claim lacks sufficient support. It follows the work of supporting, qualifying or withdrawing the claim and carrying that correction into the materials where it appears. This is useful when several audiences have received different versions of a company’s performance story.
Three further publications apply the evidence to specific professional responsibilities. The Private Equity AI Report considers whether a claimed benefit survives a change in ownership. The Investment Banking AI Report examines the reconciliation between usage, eligible billable activity, revenue, cash and delivery expense. The General Counsel AI Report considers the conditions that can invalidate an earlier legal approval.
These three reports are operating syntheses. They do not claim to be separate surveys of private equity professionals, investment bankers or general counsel. Keeping that distinction visible allows readers to assess the source evidence and the proposed professional application separately.
The board publication remains a preview
The Board Director AI Governance Report remains a preview. The direct board-response count is too small to support a board benchmark. The publication instead uses broader operator evidence to develop questions about materiality, assurance and the circumstances that should reopen a board-level commitment.
That label is substantive. The presence of a board report in the collection does not mean the research measures how boards generally behave. Readers should use it as a basis for examining their own oversight arrangements, with appropriate consideration of their organization’s circumstances.
Use the registry before quoting a number
Executive AI Statistics is the fifteenth publication and the reference point for the collection’s figures. It contains 20 complete question distributions, derived measures, historical comparisons and source provenance. It lets a reader check the wording, denominator and treatment of a measure before reusing it.
That check can change the meaning of a sentence. An expectation of measurable return is not necessarily an expectation of recovering the full investment. Visibility into AI running costs does not establish complete accounting for a workflow. A reported improvement does not establish independently verified performance. These distinctions should remain intact when a finding moves into a presentation, article or investment memorandum.
For a citation, retain the question’s population, response base, edition and relevant qualification. Link to the dated report or registry page. Where a result is carried forward, say so. Where several answers could be selected, do not describe the percentages as mutually exclusive shares of a market.
The collection includes PDF versions, aggregate data downloads and editable worksheets for the 14 narrative reports. Those worksheets help readers apply the proposed questions. They are blank working tools, not validated scoring systems or completed assessments of any company.
The release does not introduce eight newly researched vendor market maps. Existing maps remain in the archive under their own dates. Readers considering a product should check its current capabilities and terms directly; inclusion in a historical map is not a current recommendation or evidence of suitability for a particular deployment.
A practical reading order for a working meeting
Before circulating a report, write down the decision the meeting is meant to address. “Discuss our AI strategy” leaves too much room for participants to bring unrelated examples. A narrower question, such as whether to extend a particular workflow to another team, makes it easier to choose relevant reading and identify the missing company evidence.
Ask participants to bring the records that bear on that decision. Use the relevant report to identify questions those records should answer. If a survey finding helps explain why a question deserves attention, keep that purpose explicit. It should not substitute for the organization’s own cost, operating or control records.
Close the meeting with the decision taken, the evidence still needed and a named person responsible for obtaining it. A disagreement may remain because the evidence is incomplete, because participants interpret it differently or because they have different objectives. Recording which problem remains makes the next review more useful.
What readers should keep in view
Open Future Forum’s evidence comes from a selective executive and investor network. It is self-reported and cross-sectional. Respondents are not a random sample of companies, and the collection is not a matched monthly panel. The reports do not establish population-wide adoption rates, causal effects or verified company returns.
External research appears in the reports with attention to its own sample, setting and design. Original recommendations and illustrative calculations are identified separately from survey observations. Earlier editions remain available so readers can examine the record behind a comparison rather than relying on a rewritten historical summary.
The November collection gives readers a choice of depth and entry point. Start with the decision that matters to you, consult the relevant report, and use the registry to check any figures you intend to repeat. The Open Future Forum research index brings the full collection and earlier editions together.
