Zeitgeist #9Saturday, October 3, 2026
Underlying Desire
The deeper driver is the desire to offload cognitive labor without losing control. People do not just want faster software, they want relief from repetitive coordination, follow-up, and decision assembly, while still trusting that important work will be done correctly. AI agents promise a rare combination: speed, competence, and the feeling that someone is finally handling the admin.
Key Evidence
OpenAI’s 2025 enterprise report surveyed 9,000 workers across nearly 100 enterprises, showing how far enterprise AI has moved from experimentation into operational use. Capgemini’s 2025 generative AI report says adoption rose from 6% in 2023 to 30% in 2025, a fivefold jump in two years. Stanford’s 2025 AI Index reports that U.S. job postings citing generative AI skills more than tripled year over year in 2024, while Microsoft’s AI Diffusion Report 2025 says roughly one in six people worldwide now uses generative AI tools.
Why Now
Three things changed at once: models got more capable at multi-step tasks, enterprise buyers got more comfortable putting AI near production systems, and workers started adopting AI on their own. The result is that the bottleneck moved from model quality to workflow integration, permissions, and reliability, which is exactly where startups can wedge in. The market is now ready for tools that make agents safe enough to touch real business processes.
Outcome: $100M ARR in seven quarters, plus customers including ADT, Cigna, SiriusXM, and Nordstrom; Sierra also said it works with 40% of the Fortune 50. ([sierra.ai](https://sierra.ai/blog/agents-as-a-service?trk=article-ssr-frontend-pulse_little-text-block&utm_source=openai))
Outcome: Over $100M ARR, more than 500 customers in 54 countries, 350 employees, and more than 25,000 custom agents running at customers as of March to August 2026. ([harvey.ai](https://www.harvey.ai/blog/harvey-raises-growth-round-at-dollar11-billion-valuation-co-led-by-gic-and-sequoia?utm_source=openai))
A control plane for AI agents inside mid-market and enterprise teams. It would let operations, finance, and support leaders define what agents can do, which systems they can touch, what requires human approval, and how every action gets logged. This works because the biggest blocker to agent adoption is not capability, it is trust, auditability, and permissions. Companies want the gains from automation without handing a model the keys to the kingdom.
A platform for building, monitoring, and improving business agents that run tasks across CRM, ticketing, email, and internal tools. The product would track success rates, failures, handoffs, and time saved, then recommend fixes when an agent gets stuck or makes too many manual escalations. It would sell to AI-native startups first, then expand into larger companies that need a practical layer between models and workflow systems. The winner here is not a prettier prompt builder, it is the observability stack for real work.
Underlying Desire
At the core, this trend is about control. Companies are handing decision-making to opaque systems, and humans still need a way to feel that the machine is understandable, defensible, and safe enough to ship. Underneath the legal language, buyers want something deeply practical: the confidence that they can move fast without being blindsided by regulators, customers, or their own models.
Key Evidence
The European Commission says key enforcement powers under the AI Act began applying on 2 August 2026, making compliance an immediate operational issue for many companies. The EU AI Act implementation timeline extends through 2028, according to the EU AI Act Service Desk, which creates a long runway for recurring compliance work. The AI Act covers general-purpose AI models, transparency rules, and prohibited practices, according to the European Commission, expanding the need for model inventories, audit logs, policy tracking, and risk scoring software.
Why Now
What changed is enforcement. The AI Act is no longer a draft, a framework, or a future concern, it is entering active application, according to the European Commission. At the same time, the long phased rollout through 2028 means companies cannot solve this once and forget it, they need systems that can update as obligations stack up and interpretations evolve.
Enterprise AI governance and compliance software for discovering, auditing, monitoring, and proving compliance across AI systems and agents.
Outcome: Current status: Founded in 2020, named a Challenger in the 2026 Gartner Magic Quadrant for AI Governance Platforms, and publicly reports 1K+ AI use cases assessed, 5K+ mitigations deployed, and 50K+ LLMs and ML endpoints covered. It also disclosed an investment from Mozilla Ventures in March 2024. ([holisticai.com](https://www.holisticai.com/careers?utm_source=openai))
AI governance software that inventories AI systems, assesses risk, enforces policy, and tracks compliance across enterprise workflows.
Outcome: Current status: Credo AI raised $21 million in new capital in July 2024, after a $12.8 million Series A in 2022, and the company says customers use its platform to cut manual work by 60% and achieve 50% faster AI governance adoption. Public case studies include Mastercard and a global energy company with 500+ AI use cases. ([markets.financialcontent.com](https://markets.financialcontent.com/wral/article/bizwire-2024-7-30-credo-ai-announces-21-million-in-new-capital-grows-leadership-team-to-match-the-rapid-pace-of-ai-innovation-emerges-as-the-standard-of-responsible-ai-governance?utm_source=openai))
AI Act Ledger is a compliance workflow platform for enterprise AI teams, legal teams, and security teams that need to prove their systems are mapped, documented, and monitored against EU AI Act requirements. The product would maintain a live inventory of models, data sources, prompts, vendors, and use cases, then generate audit-ready evidence packs, policy change alerts, and risk scoring for each deployment. It would work because most teams already have scattered spreadsheets, model cards, and tickets, but no single system that turns AI governance into an executable process.
Policy Drift is a monitoring tool for companies shipping AI products across Europe that need to detect when models, prompts, or vendor behavior drift away from approved policy. It would plug into MLOps and vendor systems, flag changes in output patterns, record approvals, and create alerts when a system crosses a regulatory threshold. This would sell to mid-market and enterprise buyers because the hardest part of AI compliance is not the initial paperwork, it is proving that systems stayed compliant after launch.
Underlying Desire
At the deepest level, this trend is about control. People want to change their bodies without constant willpower battles, and once they start seeing real progress, they want a system that helps them keep the gains without friction, shame, or guesswork. For employers and payers, the desire is the same in different language: they want a way to pay for a high-demand therapy without losing money to dropoff, misuse, or opaque outcomes.
Key Evidence
Gallup reported in 2026 that 11% of U.S. adults currently use GLP-1 medications for weight loss, up from 3% in 2024, and 15% have used them at some point. Gallup also found U.S. obesity rates declined from 39.9% in 2022 to 36.8% in 2026, tracking the rise in GLP-1 usage. KFF found 12% of adults are current GLP-1 users and 18% have ever used them, confirming the market is already broad enough to support a services ecosystem.
Why Now
The market crossed from early adopter behavior into population scale, which makes the secondary effects measurable and monetizable. At the same time, employer coverage, Medicaid policy, and CMS model changes are shifting fast enough that companies need tools now to manage access, adherence, and ROI. The chaos is temporary, but the workflow needs are permanent.
An insurance-covered nutrition therapy platform that matches patients with registered dietitians, with a dedicated GLP-1 nutrition pathway for people taking obesity medications. ([berrystreet.co](https://www.berrystreet.co/about?utm_source=openai))
Outcome: Raised $50 million in 2025, and as of the August 2026 merger announcement with Healthify, the combined company said it serves 200,000+ patients with 2,000+ clinicians across all 50 states. ([techcrunch.com](https://techcrunch.com/2025/02/05/as-glp1s-boom-dietician-startups-berry-street-fay-each-nab-50m-rounds/?utm_source=openai))
A virtual nutrition care marketplace that connects people with insurance-covered registered dietitians. ([faynutrition.com](https://www.faynutrition.com/post/fay-series-b-announcement?utm_source=openai))
Outcome: Raised a $50 million Series B in 2025 for a total of $75 million raised, and its platform had over 1,000 providers at the time of the funding announcement, with projections to exceed 2,000 providers in 2025. ([faynutrition.com](https://www.faynutrition.com/post/fay-series-b-announcement?utm_source=openai))
A consumer nutrition and grocery planning app built specifically for GLP-1 users. It would recommend high-protein meals, portion sizes, hydration reminders, and grocery lists based on medication stage, symptoms, and weight-loss goals, then connect to delivery or retail partners. It works because GLP-1 users need a totally different consumption pattern, and most generic wellness apps ignore the real-world eating changes that determine whether people stay on treatment and feel good enough to continue.
A SaaS workflow layer for employers, benefits brokers, and telehealth clinics that manages GLP-1 access, adherence, side effects, refills, and outcomes in one dashboard. It would ingest claims, medication events, and patient check-ins, then flag churn risk, identify coverage gaps, and generate ROI reports for HR and payers. It works because the category is now big enough to justify specialized ops software, but fragmented enough that every stakeholder is still duct-taping the process together.
Underlying Desire
At the bottom of this trend is the desire to be remembered, expected, and safely held inside a web of other people. Humans are not just looking for entertainment or even companionship, they want proof that they matter to someone, that someone notices when they disappear, and that they have a place where they belong without having to perform for it.
Key Evidence
According to the WHO Commission on Social Connection’s 2025 report, loneliness and social isolation are widespread and under-recognized, and WHO says 1 in 6 people worldwide is affected by loneliness. The U.S. Census Bureau says 39.7 million one-person households existed in 2025, representing 29% of all households, up from 20% in 1975. The Federal Reserve says 13% of people lived alone in 2025, reinforcing the shift toward solo living arrangements.
Why Now
Three things changed. First, the WHO turned loneliness into a formal public health priority in 2025, which legitimizes intervention. Second, solo living has become structurally common, with a huge base of one-person households and people living alone. Third, software has finally gotten good enough to coordinate recurring human behavior, not just broadcast messages, which makes check-ins, cohorts, and care networks actually usable.
An AI relationship mentoring app that helps people improve friendships, romance, family, and work relationships.
Outcome: Raised nearly $5 million total, including a seed round in September 2023 led by Sequoia Capital, with AI Fund and NEA participating. ([blog.meeno.com](https://blog.meeno.com/blog/press-release-former-tinder-ceo-apple-and-headspace-executiv-renate-nyborg-announces-pre-seed-funding-round-led-by-andrew-ng-s-ai-fund?utm_source=openai))
A social platform that helps people find community events and turn them into real-life friendships.
Outcome: Raised $14 million in a Series A in September 2025, and TechCrunch reported 50,000 active users buying tickets plus more than 200,000 users browsing events. ([techcrunch.com](https://techcrunch.com/2025/09/10/this-gen-z-founded-app-just-raised-14m-to-take-on-the-loneliness-episdemic/?utm_source=openai))
Circle OS is a software platform for small communities, memberships, and cohort-based programs that need people to actually show up and stay engaged. It would handle recurring check-ins, nudges, accountability pods, attendance tracking, and lightweight care coordination, with templates for therapists, founders, alumni groups, parent circles, and neighborhood groups. It works because most community tools optimize for posting, not follow-through, and loneliness creates demand for systems that make connection habitual instead of accidental.
Reach Out is a care and check-in tool for families, friend groups, and eldercare networks that want a simple way to notice when someone drops off. It would automate rotating check-ins, escalation if messages go unanswered, shared task lists, and location-aware reminders for birthdays, appointments, and support needs. It works because people do not want a social network, they want reassurance, continuity, and a low-friction way to stay emotionally and practically present.
Underlying Desire
At the core, this trend is about autonomy. Creators do not just want more money, they want control over their relationship with the audience they worked to build. Underneath the business model shift is a very human desire to reduce dependence, preserve identity, and turn fragile attention into something lasting and portable.
Key Evidence
IAB projects U.S. creator economy ad spend will reach $37 billion in 2025, up 26% year over year and nearly 4 times faster than overall media industry growth, according to IAB's 2025 creator economy report. IAB also says more than half of brands now prioritize creator reputation and audience alignment, signaling demand for audience-quality and brand-safety tooling, according to IAB. MIDiA Research says creators are seeking greater control and audience ownership, pointing to growth in email, SMS, community, membership, and CRM products, according to MIDiA's 2025 report.
Why Now
Two things changed at once: brand budgets got bigger, and platform dependence got riskier. As creator ad spend scales, brands want better proof of audience quality and attribution, while creators are feeling the fragility of algorithm-driven distribution more acutely. That combination makes first-party audience software commercially useful now, not later.
A newsletter-first platform that helps creators own their audience with email, websites, podcasts, communities, subscriptions, and ads. ([beehiiv.com](https://www.beehiiv.com/about?utm_source=openai))
Outcome: As of Q2 2026, beehiiv reports 75B+ emails sent all time, 450M+ unique readers, 170K+ publications, and $60M+ earned by users. Its SEC filing also described the company as incorporated in June 2021 and said it had recently surpassed 1 billion emails sent in a single month. ([beehiiv.com](https://www.beehiiv.com/about?utm_source=openai))
A membership platform that lets creators get paid directly by fans instead of relying on ads or feeds. ([patreon.com](https://www.patreon.com/about?utm_source=openai))
Outcome: Patreon says fans have sent more than $10 billion to creators since 2013, with creators generating more than $8 billion in revenue according to its 2024 update. The company also disclosed a prior Series F raise of $155 million at a $4 billion valuation. ([patreon.com](https://www.patreon.com/about?utm_source=openai))
Audience OS is a creator CRM for independent media operators who need to turn followers into owned subscribers. It would unify email, SMS, community membership, and sponsorship attribution in one dashboard, so creators can segment fans, automate offers, and prove which channels actually drive revenue. It works because creators are increasingly tired of renting reach from platforms and brands are asking for better audience proof before they spend.
Proof Layer is a brand-safety and audience-verification tool for creators and the agencies that buy from them. It would analyze audience composition, engagement quality, fraud signals, and historical campaign performance, then package that into a simple report brands can trust. This works because more than half of brands are prioritizing creator reputation and audience alignment, so the market needs a faster way to underwrite deals and reduce wasted spend.
Underlying Desire
At the deepest level, this trend is about trust and control. Businesses want to move money and goods without fear that a small formatting error, jurisdiction mismatch, or delayed filing will trigger fines, blocked payments, or an audit. Governments want visibility and enforceability without adding more human bureaucracy. The human desire underneath is simple: certainty in systems that are otherwise chaotic, so commerce can keep flowing without constant manual intervention.
Key Evidence
The EU’s ViDA package took effect in April 2025 and lets Member States introduce mandatory e-invoicing under specific conditions, according to the European Commission. ViDA also adds real-time digital reporting for cross-border trade based on e-invoicing, creating demand for validation, reporting, and audit software, according to the Commission. The European Commission says eInvoicing reduces administrative burden and improves interoperability across the EU, while the OECD’s 2025 tax administration work highlights electronic invoicing and digital transformation as core themes, according to the OECD.
Why Now
What changed is that the regulatory direction became operational, not theoretical. ViDA moved from policy discussion to an active framework in April 2025, so vendors can now build against a real rollout path instead of guessing at future rules. At the same time, tax agencies are pushing toward real-time digital reporting, which breaks the old batch filing model and forces software teams to solve for continuous validation and compliance.
Tax infrastructure software that automates indirect tax calculation, e-invoicing, tax ID validation, and tax return filing. ([fonoa.com](https://www.fonoa.com/company/about?utm_source=openai))
Outcome: Raised a $110 million Series C in May 2026 and acquired PwC's Indirect Tax Edge; Fonoa says it has raised about €175 million total. ([fonoa.com](https://www.fonoa.com/company/news/fonoa-raises-110m-funding-and-acquires-edge?utm_source=openai))
AI-powered accounts payable automation with Peppol-compliant invoice handling by default. ([be.linkedin.com](https://be.linkedin.com/company/cashfeed?utm_source=openai))
Outcome: Raised €1.3 million in July 2025, says it is trusted by 1,000+ finance teams across 5+ countries, and supports over 350 clients. ([tech.eu](https://tech.eu/2025/07/10/cashfeed-raises-1-3m-to-simplify-e-invoicing?utm_source=openai))
Invoice Relay would be a compliance orchestration layer for finance teams that sell across multiple EU countries. It would connect to ERP and accounting systems, validate invoice data against country-specific rules, route documents to the right tax authority or network, and keep a full audit trail of every submission and rejection. This works because companies do not want to rebuild their finance stack every time a country changes e-invoicing rules, and they will pay for one system that absorbs regulatory complexity instead of pushing it back onto AP and AR teams.
TaxFlow API would expose a developer-first compliance API for SaaS platforms, marketplaces, and ERP integrators that need to generate e-invoices correctly in different EU markets. Instead of forcing product teams to encode every local tax rule themselves, it would return validated payloads, jurisdiction-specific formatting, and reporting status in one layer. This would be especially attractive to software companies selling into Europe, because e-invoicing becomes a product requirement, not just a finance back office issue.
Underlying Desire
At the deepest level, this trend is about control in a system that suddenly feels constrained. AI companies want certainty: enough power, at the right price, at the right time, without public delays or surprise bottlenecks. Grid flexibility software promises something psychologically powerful, not just efficiency, but agency over a scarce resource that can now make or break growth.
Key Evidence
The IEA says electricity demand from data centers surged 17% in 2025, and AI-focused data centers grew faster than overall global electricity demand. The U.S. Department of Energy says data centers used about 4.4% of total U.S. electricity in 2023 and could reach 6.7% to 12% by 2028. The EIA's 2026 outlook says electricity consumed by data center servers will keep increasing, especially in standalone data centers.
Why Now
Three things changed: AI workload growth accelerated, utility interconnection queues got slower, and power became a first-order constraint on expansion. At the same time, the IEA says U.S. electricity load is expected to keep rising through 2027, which makes flexibility and demand response more valuable right now. Software can help operators act faster than physical grid upgrades.
Software that turns AI data centers into flexible grid assets by dynamically adjusting power demand in response to grid conditions.
Outcome: Raised $150 million Series A at a $1.05 billion valuation in August 2026, after earlier funding that brought total capital to $68 million in 16 months. The company says it has completed five global demonstrations and is commercially deployed at multi-megawatt, full data center scale. ([emeraldai.co](https://www.emeraldai.co/?utm_source=openai))
An AI-driven energy flexibility platform that manages distributed energy resources, industrial load, storage, and demand response.
Outcome: GridBeyond says it has 5 GW under management, 550 plus clients, 1,400 plus sites, and 55 plus services and programs. It also closed a €52 million Series C round and says it has deployed its platform at over 500 sites worldwide. ([gridbeyond.com](https://gridbeyond.com/about-us-2/?utm_source=openai))
A real-time energy orchestration platform for data center operators that forecasts facility load, detects flexibility windows, and automatically shifts non-urgent demand to lower-cost or lower-congestion periods. It would target colocation providers, hyperscale operators, and enterprise data centers that are under pressure to secure power and reduce peak charges. This works because the value is immediate: even small improvements in peak management, capacity planning, and utility coordination can translate into major savings and faster expansion.
A procurement and market-optimization tool for large electricity buyers that helps data centers, manufacturers, and commercial campuses compare tariffs, hedge exposure, and bid demand response capacity into local grid programs. The product would serve energy managers and finance teams that need a clean, auditable way to buy power smarter instead of relying on consultants and static contracts. It should work because the market is getting more volatile, and buyers want software that turns energy spend into a managed portfolio rather than a line item.
Lev turns an idea into the things a company actually runs on: positioning, market and competitive research, a pitch deck, and a go-to-market plan.
Start building for free