Skip to main content
← Daily Digest

Thursday, November 5, 2026

16 signals
9

Agent Harness: How I Built My Own Claude Code, Step by Step

Victor picked this· Hello Operator · AI Eng · Tactical How-To · Nov 5
  • Agent architecture is decomposable into 15 discrete components—suggesting commoditization of AI coding assistants is underway
  • Individual developers can now match or exceed commercial offerings (Claude Code) on standard benchmarks, indicating the moat is narrowing
  • The 'no-code explanation' framing signals a shift toward demystifying agent internals—accessibility becomes competitive advantage
  • Terminal-Bench 2.0 emerges as credible evaluation framework for comparing agent performance across model families (Haiku vs DeepSeek)
9

SEO May Be Dead. Shopify Just Paid ~$100 Million for Shop.com Anyway.Time-Sensitive

Victor picked this· SaaStr — Jason Lemkin · GTM Ops · Deep Dive · Nov 5
  • SEO's share of traffic is collapsing (Google AI Overviews cut organic CTR 58-61%), but the real damage is quality: SaaStr's search visitors are 25-35% less valuable (returning users, key events) despite only 5% session decline. Top-line metrics mask the deterioration.
  • Branded search is resilient (56.4% CTR vs 24.8% non-branded; only 3% traffic loss vs 34% for generic B2B queries). The traffic dying is top-of-funnel explainer content—exactly what AI Overviews answer without clicking through.
  • Shopify's $100M Shop.com acquisition is a strategic bet on direct traffic and brand-as-default in the AI-agent era. When users say 'Find me X on Shop' to ChatGPT, the domain matters more than ranking. Reddit's pivot to logged-in users (80% of revenue) follows the same logic: own
  • B2B founders should: (1) Plan for search to shrink as traffic share; (2) Track quality metrics (returning users, key events, not just sessions); (3) Stop writing generic 'what is X' content; (4) Monitor AI assistant channels and brand mentions in LLM responses; (5) Own 1-2 words
  • The new competitive moat is direct traffic + brand association in LLM training data. Companies winning in 2026 are those people come to directly or ask agents for by name, not those ranking #1 for generic queries.
8

Marketers Lose a Quarter of Their Budgets To Fraud, and AI Is Going To Make it WorseTime-Sensitive

Demand Gen Report · GTM Ops · Quick Take · Nov 5
  • Ad fraud now represents ~27% of enterprise marketing budgets ($81B annually across digital ad market)—a structural problem, not edge case. Marketers are 98% confident in detection yet 87% report worsening fraud, suggesting measurement blindness at scale.
  • AI is actively accelerating fraud vectors: LLMs enable signal spoofing, customer impersonation, and deepfake brand exploitation. New ad models (AI-driven channels) historically create 12-24 month windows of undetected exploitation before standards emerge (mobile app fraud precede
  • Marketers responding with two strategies: (1) Flight to trusted walled gardens (Google, Meta, Amazon) for better measurement control, accelerating open web decline; (2) Investment in third-party measurement partners. This consolidation trend favors platform giants and measurement
  • Paid social (29%) and mobile apps (25%) are fraud hotspots because they rely on proxy metrics (clicks, impressions, installs) rather than bottom-line outcomes. The further advertising is from direct business impact, the easier fraud exploits measurement gaps.
8

Biggest 2026 Wins, Boldest 2027 Bets: Marketing Leaders Share Takes on AI, Events, Social and GrowthTime-Sensitive

The Exit Five Podcast · GTM Ops · Practitioner Story · Nov 5
  • AI adoption in 2026 was incremental, not revolutionary: Casey Patterson's 'small earthquakes every week' strategy prevents organizational shock while embedding AI workflows—suggests measured, continuous change beats big-bang implementations
  • Team quality and culture trump growth metrics: Dave Gerhardt's 2027 bet is explicitly NOT about scaling—it's about 'raising the bar on quality' and building a team that can execute without founder involvement. This reflects a maturation away from growth-at-all-costs
  • In-person events are experiencing a contrarian resurgence: Ross Simmonds predicts a 'massive surge' in real-life events in 2027, positioning human connection as a necessary counterbalance to AI/digital saturation. Colin Kendall already proved this with event-attributed revenue at
  • Employee advocacy and personal branding are 2027 bets: Colin Kendall's focus on leader content and personal branding suggests B2B marketers are moving beyond corporate channels to individual credibility—aligns with human-first GTM trend
  • Operational simplicity enables execution: Dave Gerhardt emphasizes 'it's not complicated—it's content, it's a newsletter'—suggests that in an AI-saturated market, clarity of mission and quality execution matter more than feature complexity
7

Abundant Intelligence at the Midwest AI Summit Part 2

Practical AI · Enterprise AI · Practitioner Story · Nov 5
  • AI charter created BEFORE coding: International Mission Board spent days defining what AI would/wouldn't do, enabling rapid boundary-setting on future requests without deliberation
  • Non-technical leaders can drive AI governance: Don Barger (accounting + theology background) positioned as credible AI steward, challenging assumption that only CS/data science experts should lead AI decisions
  • Velocity gains are real but require human-in-loop discipline: Product shipping accelerated from years to months, but only because organization pre-committed to keeping humans in control of high-stakes decisions (pastoral care, final decisions about people, autonomous actions)
  • Ethical boundaries prevent scope creep: When urgent need arose for AI counselor chatbot, pre-established charter enabled immediate 'no' without re-deliberation; prevented mission drift
  • Contrarian positioning: 'Something can be worse than nothing if you're doing the wrong things' — directly challenges Silicon Valley's move-fast-break-things ethos
7

Greenhouse’s CFO on Breaking the AI Hiring Doom Loop | Paul Todgham

Run the Numbers · Enterprise AI · Practitioner Story · Nov 5
  • AI adoption in hiring creates a 'doom loop' where both candidates and recruiters use AI, eroding trust—a systemic problem requiring human-centered solutions, not more automation
  • Pricing AI products requires rethinking willingness-to-pay models; venture subsidies mask true cost structures, forcing companies to pass AI costs to customers via hybrid commits, tiers, and credits
  • CFOs need 'Surges and Savings' budgeting frameworks to measure AI ROI rigorously; generic AI spend without business case discipline leads to waste and misaligned incentives
  • Anthropic's 900% NRR and ~$2T valuation signal massive market opportunity but also extreme compute cost exposure; gross margin and profitability claims require scrutiny
  • Metrics BS is rampant in AI-era fundraising; investors spot fake numbers via hierarchy of metrics (token maxing, cagey CEO behavior, suspicious EBITDA add-backs)
6

AI agents can now make your phone callsTime-Sensitive

The Rundown AI · AI Eng · Quick Take · Nov 5
  • AI phone agents are moving from concept to production (Grok Bot now live for paid users), with legal liability shifting entirely to end users as 'caller of record'
  • Public sentiment has sharply turned against AI infrastructure: 57% of voters expect AI to harm society, 76% oppose data centers locally, and Democrats now control regulatory oversight—signaling end of hands-off federal approach
  • Agent-to-agent communication is emerging as the next frontier ('bots talking with other bots'), with Grok Voice handling 20,000 calls daily, suggesting phone calls may become fully automated workflows
  • Consolidation accelerating: OpenAI acquiring Pierre (code review), Meta acquiring Conductor (multi-agent management), and major cloud providers (Google, Box, NVIDIA) bundling agent capabilities into enterprise platforms
  • Token economics and cost management becoming critical: users hitting credit limits on $20 AI plans, Elon Musk acknowledging Grok's pricing is 'unsustainably high relative to revenue,' and optimization strategies (prompt-to-context engineering) now essential
6

An Interview With Intuit CEO Sasan Goodarzi About Accounting and AI

Feed: » stratechery by Ben Thompson · Enterprise AI · Thought Leadership · Nov 5
  • Generalist leadership with systems thinking and curiosity is a competitive advantage during paradigm shifts like AI adoption—especially for leaders who avoid the trap of 'I already know the answers'
  • Intuit's 44-year survival and evolution (Quicken→QuickBooks→TurboTax→Credit Karma/Mailchimp consolidation) demonstrates how legacy software companies can reinvent through acquisitions and business model shifts, positioning them to grow rather than be disrupted by AI
  • The accountant-to-customer transformation via AI represents a fundamental shift in Intuit's customer acquisition and value delivery model—moving from partner channel to direct AI-enabled services
  • Non-traditional Silicon Valley backgrounds (enterprise, industrial, defense) bring disciplined capital allocation and quality rigor that tech companies often lack, creating structural advantages in mature markets
  • The interview frames AI as an opportunity for Intuit to reinvent customer experiences rather than a threat—suggesting confidence in platform consolidation strategy and ability to absorb new capabilities
6

Did I Write This Post?

**Trust Insights (Chris Penn) · Future of Work · Practitioner Story · Nov 5
  • AI detection tools (like Substack's) are unreliable and often trained on non-consented human content—creating false positives that damage creator credibility
  • Transparent AI disclosure builds trust more effectively than gatekeeping or denying AI use; the key is disclosing *how* and *where* AI is used in the process
  • Personal AI tools trained on individual writing samples (9/10 accuracy) represent the future of creator-AI collaboration—not generic AI replacing human voice, but augmenting it
  • The real value of AI in writing workflows is cleanup/coherence, not ideation—the human brings purpose, expertise, and authentic voice; AI handles formatting and structure
  • Engagement metrics (rate of response) matter more than content purity—audience cares about value received, not whether AI touched the draft
6

Steal this idea: Bloomberg’s ‘Project Centurion:’ 100 AI wins in 100 days

Charter - Future of Work, AI, Management, Hybrid · Enterprise AI · Practitioner Story · Nov 5
  • Bloomberg's HR department used a time-bound challenge (100 days) to systematize AI adoption rather than ad-hoc experimentation
  • Leadership-driven AI initiatives can create organizational momentum and measurable accountability
  • The 'wins' framework suggests focus on incremental, achievable AI applications rather than moonshot transformations
6

Box taps Nvidia’s AI models to build a rich context layer for autonomous enterprise agentsTime-Sensitive

SiliconANGLE · AI Eng · Vendor Content · Nov 5
  • Context quality is the primary bottleneck limiting enterprise AI agent adoption—agents without proper context make mistakes that erode organizational trust
  • Unstructured data (contracts, emails, documentation) contains most enterprise context but requires permission-aware, secure retrieval mechanisms to prevent data leakage and compliance violations
  • Box's Nvidia partnership delivers measurable efficiency gains (20% search accuracy improvement, 90% token reduction, 4x faster response times) through specialized embedding models and multimodal indexing of messy enterprise content
  • Open-weight models (Nemotron 3 Ultra) are becoming competitive alternatives to proprietary LLMs for domain-specific workflows, with cost and customization advantages for financial/legal use cases
  • Agent security is now table-stakes: real-time threat detection, classification-based access policies, and MCP guardrails are required to prevent rogue agent data exfiltration and account takeover
5

Tech giants turn to cash, concessions for AI rebrand after voters pummeled data centers 43-1Time-Sensitive

Axios · AI Market · Quick Take · Nov 5
  • Voters delivered a 43-1 mandate against data center expansion (75% average support for restrictions), forcing tech giants into emergency rebrand from 'data centers' to 'compute campuses' with real financial concessions
  • Industry strategy shift: moving from job creation narratives to direct community payments ($50K teacher bonuses, $1B+ commitments), infrastructure cost-sharing, and binding environmental/labor agreements—signaling recognition that traditional GTM messaging failed
  • 60% of planned 2028 U.S. data center capacity now faces delay risk; tech companies betting slower pace + higher costs + community buy-in is preferable to existential regulatory threat, but unclear if binding agreements will satisfy incoming Democratic leadership
5

[AINews] It's Swapping SeasonTime-Sensitive

Swyx · AI Market · Quick Take · Nov 5
5

Exclusive: StrongestLayer launches tools to verify payment requests and block hidden AI instructionsTime-Sensitive

SiliconANGLE · Enterprise AI · Vendor Content · Nov 5
  • Small slice of attacks (3.6%) account for 98% of financial exposure—traditional volume-based security misses the costliest threats
  • 94% of malicious payment requests pass sender authentication, exposing gap in conventional email security controls
  • Prompt injection via email is emerging attack vector for AI assistants; inbox is 'the one route any outsider can use' to compromise AI systems
  • Shift from 'does this look bad?' to 'what is this trying to make happen?' represents fundamental security paradigm change for AI era
  • Vendor email compromise (legitimate account takeover) remains high-impact because requests appear legitimate and bypass authentication checks
5

Big Tech’s big AI buildout will not end until someone stops itTime-Sensitive

The Verge AI · AI Market · Quick Take · Nov 5
  • Big Tech committed to $720B+ annual CapEx with $1T+ multi-year obligations, but AI monetization remains unproven after 3+ years of investment—creating potential market correction risk
  • Google's AI Mode search cannibalization of core Search business (12% growth vs 17% prior year) reveals the profitability paradox: new AI products undermine legacy revenue before replacement revenue scales
  • Microsoft was only hyperscaler to avoid Wall Street backlash, but announced $25B bond sale immediately after earnings—indicating even the 'best positioned' player is now leveraging debt to fund AI buildout
  • Anthropic's $17.5B compute costs nearly equal $17B quarterly revenue, with $900B future commitments—demonstrating that even high-growth AI labs face structural profitability challenges at current cost structures
  • Regulatory uncertainty (midterm elections favoring data center restrictions) may be the only external pressure forcing AI spending slowdown, as companies show no internal discipline to moderate capital deployment
5

Research spotlight: What NBA coaches can teach workplace leaders about managing burnout

Charter - Future of Work, AI, Management, Hybrid · Future of Work · Research/Data · Nov 5
  • Load management (strategic rest) prevents burnout in high-skill, hard-to-replace talent pools—validated by NBER economics research
  • Counterintuitive insight: reducing workload today increases output tomorrow, challenging productivity-at-all-costs culture
  • Framework applies across industries: sports injury prevention parallels workplace burnout prevention for knowledge workers