{
  "source_file": "0930",
  "source_platform": "Dev.to",
  "author": "Debashish Ghosal",
  "title_listed": "You Aren't Choosing an AI Tool. You're Choosing Who Gets Paged at 2 AM.",
  "keywords": "",
  "domain": "机器学习",
  "content_type": "深度技术解析",
  "is_paid_flag": "否",
  "url": "https://dev.to/debashish_ghosal/you-arent-choosing-an-ai-tool-youre-choosing-who-gets-paged-at-2-am-5g4a",
  "publish_date": "2026-09-06",
  "corpus_stage": "",
  "notes": "",
  "id": "d44a6cff1944",
  "collected_at": "2026-10-02T14:06:20",
  "title": "You Aren't Choosing an AI Tool. You're Choosing Who Gets Paged at 2 AM. - DEV Community",
  "body_markdown": "Part of\nAI Leadership in the Real World\n— how leaders turn scattered pilots into governed, adopted, measurable capability.\nTLDR: A support agent doing 50,000 chats a month needs ~3.5 FTE and $500k+/year just to stay accurate — while a typical 100-seat Copilot rollout sees only 20-30 seats used weekly. For SMBs, build-vs-buy isn't about features. It's about what you can afford to own for 24 months.\nWe thought we were choosing a tool. We were really choosing a future dependency, a support queue, a governance burden, and a second bill that arrives a year later.\nEvery vendor demo promised acceleration, control, and simplicity at once. Every internal proposal promised flexibility, ownership, and leverage. Nobody said both bills arrive late — one in engineering on-call, the other in consumption meters.\nGood platform decisions feel a little boring at first and very smart a year later.\nWhy AI is special (and why old build-vs-buy math breaks)\nTraditional software mostly stays still when you leave it alone. AI doesn't:\nIt drifts.\nKnowledge changes, customer language shifts, users ask harder questions once they trust it. Accuracy quietly drops from 90% to 70% with no error log.\nIt speaks for you — legally.\nA wrong Confluence page is embarrassing. A wrong chatbot answer is a commitment a tribunal can enforce.\nIt lives on someone else's deprecation clock.\nOpenAI gives at least 6 months before retiring a GA model. That's a hard deadline, not a backlog item. Prompts, evals, and output parsers all need rework.\nIt multiplies cost per request.\nOne human click = one action. One agent resolution = 6 lookups, drafts, updates, and logs — each potentially metered.\nIt turns connectors into permanent work.\nSalesforce, SharePoint, Jira, Zendesk all change auth, rate limits, and APIs. Your agent keeps running while its knowledge goes stale.\nGartner\npredicts 40%+ of agentic AI projects will be canceled by end of 2027\non cost, unclear value, and weak risk controls. McKinsey's\nState of AI 2025 (Nov 5, 2025)\nfound 88% report regular AI use in at least one function, yet only 39% report any enterprise EBIT impact (most under 5%; ~6% high performers). BCG's\nWidening AI Value Gap (2025, n=1,250)\nfound only 5% generating value at scale while 60% see minimal gains. The gap isn't building. It's owning.\nHere are 6 public cases every SMB CTO should know — scenario, what happened, what didn't work, and why.\n1. The chatbot that created legal liability: Air Canada (2024)\nScenario:\nJake Moffatt, booking emergency travel after his grandmother died, asks Air Canada's website chatbot about bereavement fares. The bot says: book full fare now, claim the discount within 90 days.\nWhat happened:\nThat was wrong. The real policy barred retroactive claims. Moffatt flew, applied with screenshots and a death certificate, was denied. He took it to the British Columbia Civil Resolution Tribunal —\nMoffatt v. Air Canada, 2024 BCCRT 149\n(Feb 14, 2024).\nThe tribunal ordered Air Canada to pay C$812.02 (C$650.88 fare difference + interest + fees) for negligent misrepresentation. It explicitly rejected Air Canada's \"remarkable\" defense that the chatbot was a separate entity responsible for its own actions: \"it is still just a part of Air Canada's website.\"\nWhat didn't work and why:\nNo grounding to the actual policy page, no policy-layer guardrail, no correction path. The bot hallucinated a generous policy linked to the real bereavement page that contradicted it.\nSMB lesson:\nIf you ship a customer-facing agent — bought or built — you own what it says. For a 20-person company, one such incident is a support crisis and a trust crisis. Budget the eval + human escalation before launch, not after.\n2. The \"AI platform\" that was mostly services: Builder.ai (2025)\nScenario:\nBuilder.ai, founded 2016 as Engineer.ai, marketed \"Natasha\" as AI that builds software \"as easy as ordering pizza.\" Raised $450M+, valued at ~$1.5B in 2023 with Microsoft and Qatar Investment Authority backing.\nWhat happened:\nOn May 20, 2025 it entered insolvency. Audits revised claimed 2024 revenue of $220M down to ~$55M. Creditor Viola Credit seized ~$37-40M after a $50M debt facility. The US Attorney's Office for SDNY subpoenaed records. UK entity Engineer.ai Global Ltd went into compulsory liquidation.\nViral coverage said \"700 engineers faked the AI.\" That shorthand is contested — Gergely Orosz (Pragmatic Engineer), after talking to ex-employees, corrected it: there\nwas\na real AI team doing spec, estimation, and code workflows, alongside hundreds of outsourced developers doing delivery. WSJ had flagged heavy human reliance as early as 2019.\nWhat didn't work and why:\nThe hybrid model can be legitimate. The failure was marketing opacity + financial misrepresentation: buyers couldn't tell which part was automated, which was human, and what would happen if the vendor collapsed. Customers who outsourced their roadmap to it lost the platform overnight.\nSMB lesson:\nWhen buying an \"AI platform,\" diligence the labor model: what % is model vs. reusable components vs. humans? What happens to your code, data, and uptime if they go under? Get escrow and export in writing.\n3. The seat license nobody sat in: Microsoft 365 Copilot\nScenario:\nA 100-person SMB buys 100 Copilot for Microsoft 365 seats at $30/user/month ($36,000/year) — plus the required underlying M365 license upgrade many teams forget to price.\nList price:\nMicrosoft lists Microsoft 365 Copilot at $30/user/month (annual commitment) as an add-on requiring a qualifying M365 base plan — true all-in $42-$90/user/month depending on base tier. Copilot Studio is separate consumption: ~$200/mo per 25,000-credit pack.\nWhat breaks:\nIndependent utilization surveys consistently report only 20-30% of purchased seats see weekly active use at scale (directional, not audited). Change management (training, comms, support) is repeatedly estimated at 30-50% of license cost and rarely in the proposal.\nSeparate cautionary tale (consumer, not enterprise):\nAustralia's ACCC filed Federal Court proceedings Oct 27, 2025 alleging Microsoft misled ~2.7M Personal/Family subscribers by bundling Copilot rises ($109→$159 Personal, $139→$179 Family) while hiding the cheaper Classic no-AI option in the cancellation flow. The widely cited 100%+ ROI figures trace to a Microsoft-commissioned Forrester TEI study — real data, not independent data.\nSMB scale-down:\nAt 25 seats instead of 100, the waste is smaller in dollars but identical in ratio — 5-8 active seats carry the other 17. That's why seats must be earned, not allocated.\nWhat didn't work and why:\nPer-seat pricing assumes uniform adoption. In SMBs adoption is spiky: 15 power users love it, 60 try twice, 25 never enable it. You pay for 100, get value from 25. The prerequisite upgrade + idle seats kills ROI by month four.\nSMB lesson that worked:\nPilot with a 25-seat power cohort, instrument weekly active use from day one, and only expand when active-use >60% for 4 weeks. Disciplined teams treat seats as earned, not allocated.\n4. The second meter: Salesforce Agentforce ($2/conversation)\nScenario:\nA mid-size team handling 50,000 service interactions a month turns on Agentforce. List price: $2 per conversation — on top of existing Service Cloud seats.\nList price:\n$2 per conversation at list, before discounts — and before the Service Cloud seats underneath. Salesforce's current official model is Flex Credits ($500/100k; 1 standard action = 20 credits = $0.10) as the flexible alternative, with unused credits not rolling over and Flex vs. Conversations not mixable in one org.\nWhat breaks:\nConversation definitions (failed/escalated/abandoned handling) and per-workflow action counts are contract-specific — which is exactly why pre-sign modeling matters. Internal IT/HR agents with lots of short queries get the worst unit economics on per-conversation pricing — same $2 whether the answer saved $200 or $2.\nSMB scale-down:\nAt 5,000 interactions/mo instead of 50,000, the meter drops 10x (~$10k/mo at list) — but the modeling work doesn't. You still must define the billable unit in writing before signing.\nWhat didn't work and why — usage amplification:\nA human resolving a ticket = 1 action. An agent resolving it = query record + search KB + draft response + update case + send email + log interaction = 6 metered steps, depending on config. Teams estimated on\nuser requests\nbut were billed on\nagent actions\n.\nCopilot Studio has the same pattern: $200/mo for 25,000 credits, but generative actions burn credits faster than classic flows, plus separate Azure token + compute charges. Three bills, three consoles.\nSMB lesson that worked:\nDemand a cost-modeling pilot in writing: define \"conversation/credit\" for\nyour\nworkflow, exclude tests, cap overages, negotiate rollover + 24-month price lock. Design agents to cache lookups and confirm intent before chaining actions — teams report 20-40% consumption cuts from flow design alone.\n5. The build that became a hidden product: the 3.5-FTE support agent\nScenario (composite SMB, modeled on SearchUnify's 50k/mo reference — not a single named company):\nA 30-person SaaS company builds a RAG support agent on LangGraph + Confluence + Zendesk. Pilot hits 88% answer accuracy. CEO calls it \"done.\"\nWhat happened (per SearchUnify's 2026 vendor field analysis — directional, not audited):\nTo keep a 50k-interaction/mo agent production-grade you need roughly: 1.0 AI/ML engineer + 1.0 data engineer + 0.5 platform + 0.5 security/governance + 0.5 product analyst =\n3.5 FTE, ~$500k-700k/year\nbefore infra, tokens, observability, and audits. Major model migrations take 6-10 weeks each (re-benchmark, re-prompt, re-test, re-secure). Connectors drift: auth changes, endpoints retire, docs grow by thousands of pages.\nThree drifts compound: knowledge drift (docs change), data drift (tickets change), behavioral drift (users ask harder questions). Without evals, you find out from angry customers.\nSMB scale-down:\nAt 5k interactions/mo the token/infra meter drops ~10x — but the governance floor doesn't. You still need someone to own drift review, connector fixes, and the model-deprecation deadline. That's the firewall that pages\nyou\nat 2 AM when the model retires or the connector breaks.\nWhat didn't work and why:\nThe team budgeted 8 weeks to build, zero heads to operate. Two backend devs now spend Fridays fixing retrieval, tuning prompts, and patching connectors — the \"temporary\" bot became an unstaffed platform product. Software maintenance is 80-90% of lifecycle cost historically; AI maintenance runs 2-3x the initial build in the first few years because drift + vendor deprecation never pause.\nSMB lesson:\nDon't ask \"can we build it?\" Ask \"can we staff 0.5-1.0 FTE for 24 months to own it — including who gets paged?\" If not, buy the boring platform and own only prompts + evals.\n6. What\ndid\nwork: narrow scope + outcome pricing (Intercom Fin + BCG 5%)\nScenario:\nSame SMB instead picks one high-volume, well-logged workflow — e.g., \"refund-status + plan-change\" tickets — and buys Intercom Fin at ~$0.99 per\nresolved\nconversation.\nWhat happened:\nIntercom's official Fin meter is $0.99 per\nresolution\n(standalone available with a 50-resolution minimum, no seats required) — you pay only when Fin resolves, not per query or token. Why this pattern works for SMBs: price maps to value (resolution, not login), the team can calculate cost-per-resolution vs. human cost, and scope stays narrow enough to govern. Industry studies widely report the same shape even where exact numbers vary by survey: BCG reported only ~5% generate substantial AI value at scale with \"deep and narrow\" pilots outperforming sprawl, McKinsey reported high AI usage but far lower EBIT impact, and mature portfolios skew heavily buy (commodity) with selective build (differentiating). Treat those survey percentages as directional, not audited — the mechanics (narrow scope + outcome meter) are the reliable part.\nWhy it worked:\nNarrow blast radius, verifiable outcome, and a vendor whose meter matches your P&L. The team kept ownership of KB quality + weekly refusal/deflection review — 2 hours/week, not 2 FTEs.\nWhat TCO actually means for AI (the 4 lines everyone underestimates)\nThe \"second bill that arrives a year later\" — that's TCO. Most teams price the build and forget the other three lines:\nBuild / purchase\n— the demo number everyone quotes (Copilot $30/seat, Agentforce $2/conv, or your 8-week sprint).\nOperate & maintain\n— 2-3x the build in the first few years (drift detection, connector fixes, model migrations at 6-10 weeks each, prompt rework).\nGovernance & change management\n— 30-50% of license cost (training, comms, support) or 0.25-0.5 FTE for evals, security review, incident response.\nMeter / exit reserve\n— the second meter (overages, usage amplification, repricing on renewal) + the lock-in reserve (what it costs to leave: data export, re-integration, M&A repricing).\nScore the rubric's \"24-mo TCO\" row on the\nsum\nof these four lines, not the first one. Here's a worked example for a 25-seat SMB handling 5,000 tickets/month:\nLine (24 months)\nBuy (Copilot + Agentforce/Fin)\nBuild (LangGraph + RAG)\nPurchase / build\n25 seats × $30 × 24 = $18k + Fin 5k × $0.99 × 24 = $119k →\n$137k\n8-week sprint ~$40k one-time →\n$40k\nOperate & maintain\nvendor-absorbed →\n$0\n2.5x build = $100k →\n$100k\nGovernance / change mgmt\n40% of license = $55k + 0.25 FTE evals $30k →\n$85k\n0.5 FTE × $150k × 2 =\n$150k\nMeter / exit reserve\noverages + repricing reserve →\n$20k\nlock-in: data export + re-integration →\n$15k\n24-mo total\n$242k\n$305k\nRead honestly: in this scenario the build runs ~26% more over 24 months and hands you the on-call burden. Swap your seat counts, ticket volumes, and rates — the shape stays. Build wins only when the workflow differentiates you\nand\nyou can staff the operate/govern lines.\nAn SMB CTO rubric (steal this for your next review)\nScore each use case 1 (buy) to 5 (build). Default to buy unless 2+ factors score 4-5.\nFactor\n1 = Buy\n5 = Build\nDifferentiation\nGeneric across companies (follow-ups, expense coding, support deflection)\nEncodes how\nyou\nwin (your underwriting, your triage, your data moat)\nData sensitivity\nPublic/internal-low-risk, vendor has certs you lack (SOC 2, EU residency)\nProprietary + regulated, must stay in your VPC with audit trail you control\nChange frequency\nStable workflow, vendor ships monthly improvements you want free\nChanges weekly with your product; vendor roadmap would block you\nOwnership capacity\nNo one to carry 0.5 FTE on-call, evals, connector fixes for 24 mo\nNamed owner + backfill funded for upgrades, drift review, incident response\n24-mo TCO\nBuild TCO >2x buy on honest math (build + 2-3x maintenance + infra + governance)\nBuild TCO within 30% of buy\nand\nexit/lock-in reserve priced (re-pricing, M&A, seat+usage dual meter)\nNon-negotiables before any demo: SSO, audit logs, data residency/boundaries, admin kill-switch + human escalation, observability (resolution, hallucination, cost/interaction). If a vendor can't meet these, they're off the shortlist — no matter how good the demo.\nRun pilots in real workflows, not sandboxes. Document re-evaluation triggers (e.g., \"if cost/resolution >$X or active-use <Y%, we revisit in 90 days\"). Revisit on schedule, not when pain forces it.\nKey takeaway\nBuild-vs-buy for AI is really\nown-vs-coordinate\n. Own only the layer that differentiates you\nand\nyou can operate well for 12-24 months. Coordinate (rent) everything else — but model the second meter, staff the 2-hour weekly governance loop, and keep an exit path.\nHonest limitations\nThis piece has no access to your contracts or usage logs — substitute your seat counts, volumes, and rates into the math above. It doesn't cover fine-tuning economics or on-prem GPU breakeven (competitive only at sustained 5-6+ hrs/day utilization over 3-5 years). Regulated workloads (health, finance, public sector) shift strongly toward buy-the-certified-platform or build-in-VPC — get counsel early. And vendor pricing from 2025-2026 will move; re-benchmark every 6-12 months.\nOpen questions for your team\nWhat AI layer do you actually want to own at 2 AM, and why?\nWhat are your non-negotiables before you compare vendors?\nHow do you stop tool sprawl before 15 pilots become permanent?\nWhat TCO line (support, governance, change management, overages) are you most likely underestimating?\nWhen does a tactical tool become a strategic dependency — and who decides?\nIf you had to standardize on one AI lane tomorrow, what would drive the choice: cost, control, or speed?\nThe smartest build-versus-buy decision is rarely about the best feature set.\nIt's about choosing what your organization is prepared to own well — and refusing to rent what you can't afford to lose.\nSources\nPrimary (official / court / regulator / analyst press release):\nSalesforce,\nAgentforce Pricing (official)\n— Conversations $2/conversation; Flex Credits $500/100k; 1 action = 20 credits ($0.10); unused Flex Credits do not roll over; Flex + Conversations cannot mix in one org\nSalesforce Help,\nAgentforce Pricing (May 19, 2025)\n+\nFlexible pricing press release (May 15, 2025)\nMicrosoft,\nMicrosoft 365 Copilot enterprise pricing (official)\n— $30/user/mo add-on, qualifying M365 plan required\nMicrosoft,\nCopilot Studio pricing (official)\n— 25,000-credit packs $200/mo; credits billed per agent action/response\nMicrosoft Learn,\nLicense options for Microsoft 365 Copilot\nACCC,\nMicrosoft in court for allegedly misleading millions (Oct 27, 2025)\n+\nConcise Statement PDF\n— consumer Personal/Family bundling case (~2.7M subscribers; $109→$159 Personal, $139→$179 Family; Classic option hidden in cancellation flow; alleged, not yet decided)\nGartner,\nPredicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (June 25, 2025)\nMoffatt v. Air Canada, 2024 BCCRT 149 —\nBCCRT decision\n/\nCanLII\n; coverage:\nBBC\n,\nABA Business Law Today\n— C$812.02 award, \"separate entity\" defense rejected\nBuilder.ai insolvency —\nTechCrunch (May 20, 2025)\n($1B+ valuation, $450M+ raised, insolvency proceedings); nuance correction:\nPragmatic Engineer — Builder.ai did not \"fake AI with 700 engineers\"\n(hybrid AI + human delivery; WSJ flagged human reliance as early as 2019)\nIntercom,\nFin — $0.99 per resolution (official)\n+\nPricing\n+\nAI agent pricing comparison (Fin vs Zendesk vs Agentforce)\n— outcome meter (pay only on resolution), standalone Fin with 50-resolution minimum; resolution-rate and adoption-lift figures are vendor-reported\nSecondary (vendor field analysis — directional, not audited benchmarks):\nSearchUnify,\nThe Real Cost of AI Agents: Maintenance, Model Drift, and Hidden Ownership Costs\n— 3.5 FTE / $500-700k/yr reference model at 50k interactions/mo, 6-10 week model",
  "metadata": {
    "title": "You Aren't Choosing an AI Tool. You're Choosing Who Gets Paged at 2 AM. - DEV Community",
    "description": "Part of AI Leadership in the Real World — how leaders turn scattered pilots into governed, adopted,... Tagged with ai, buy, tco, leadership.",
    "keywords": "ai, buy, tco, leadership, software, coding, development, engineering, inclusive, community",
    "author": null,
    "og:type": "article",
    "og:url": "https://dev.to/debashish_ghosal/you-arent-choosing-an-ai-tool-youre-choosing-who-gets-paged-at-2-am-5g4a",
    "og:title": "You Aren't Choosing an AI Tool. You're Choosing Who Gets Paged at 2 AM.",
    "og:description": "Part of AI Leadership in the Real World — how leaders turn scattered pilots into governed, adopted,...",
    "og:site_name": "DEV Community",
    "og:image": "https://media2.dev.to/dynamic/image/width=1200,height=627,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fghqh4ek5gbowwxstvui4.jpeg",
    "twitter:site": "@thepracticaldev",
    "twitter:creator": "@",
    "twitter:title": "You Aren't Choosing an AI Tool. You're Choosing Who Gets Paged at 2 AM.",
    "twitter:description": "Part of AI Leadership in the Real World — how leaders turn scattered pilots into governed, adopted,...",
    "twitter:card": "summary_large_image",
    "twitter:widgets:new-embed-design": "on",
    "twitter:image:src": "https://media2.dev.to/dynamic/image/width=1200,height=627,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fghqh4ek5gbowwxstvui4.jpeg"
  },
  "paywall": {
    "listed_paid": false,
    "detected": false,
    "marker": "",
    "public_only": false,
    "policy": "付费/会员内容只保留页面公开可见部分，不绕过付费墙"
  },
  "crawl": {
    "success": true,
    "status_code": 200,
    "error": ""
  },
  "code_blocks": [],
  "images": [],
  "audio": [],
  "videos": [],
  "links": [
    {
      "href": "https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027",
      "label": "predicts 40%+ of agentic AI projects will be canceled by end of 2027",
      "type": "a"
    },
    {
      "href": "https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2025",
      "label": "State of AI 2025 (Nov 5, 2025)",
      "type": "a"
    },
    {
      "href": "https://www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap",
      "label": "Widening AI Value Gap (2025, n=1,250)",
      "type": "a"
    },
    {
      "href": "https://www.salesforce.com/agentforce/pricing",
      "label": "Agentforce Pricing (official)",
      "type": "a"
    },
    {
      "href": "https://help.salesforce.com/s/articleView?id=004811240&language=en_US&type=1",
      "label": "Agentforce Pricing (May 19, 2025)",
      "type": "a"
    },
    {
      "href": "https://www.salesforce.com/news/press-releases/2025/05/15/agentforce-flexible-pricing-news",
      "label": "Flexible pricing press release (May 15, 2025)",
      "type": "a"
    },
    {
      "href": "https://www.microsoft.com/en-us/microsoft-365-copilot/pricing/enterprise",
      "label": "Microsoft 365 Copilot enterprise pricing (official)",
      "type": "a"
    },
    {
      "href": "https://www.microsoft.com/en-us/microsoft-365-copilot/pricing/copilot-studio",
      "label": "Copilot Studio pricing (official)",
      "type": "a"
    },
    {
      "href": "https://learn.microsoft.com/en-us/microsoft-365/copilot/microsoft-365-copilot-licensing",
      "label": "License options for Microsoft 365 Copilot",
      "type": "a"
    },
    {
      "href": "https://www.accc.gov.au/media-release/microsoft-in-court-for-allegedly-misleading-millions-of-australians-over-microsoft-365-subscriptions",
      "label": "Microsoft in court for allegedly misleading millions (Oct 27, 2025)",
      "type": "a"
    },
    {
      "href": "https://www.accc.gov.au/system/files/accc-v-microsoft-concise-statement-27-october-2025.pdf",
      "label": "Concise Statement PDF",
      "type": "a"
    },
    {
      "href": "https://decisions.civilresolutionbc.ca/crt/sc/en/item/525448/index.do",
      "label": "BCCRT decision",
      "type": "a"
    },
    {
      "href": "https://www.canlii.org/en/bc/bccrt/doc/2024/2024bccrt149/2024bccrt149.html",
      "label": "CanLII",
      "type": "a"
    },
    {
      "href": "https://www.bbc.com/travel/article/20240222-air-canada-chatbot-misinformation-what-travellers-should-know",
      "label": "BBC",
      "type": "a"
    },
    {
      "href": "https://www.americanbar.org/groups/business_law/resources/business-law-today/2024-february/bc-tribunal-confirms-companies-remain-liable-information-provided-ai-chatbot/",
      "label": "ABA Business Law Today",
      "type": "a"
    },
    {
      "href": "https://techcrunch.com/2025/05/20/once-worth-over-1b-microsoft-backed-builder-ai-is-running-out-of-money/",
      "label": "TechCrunch (May 20, 2025)",
      "type": "a"
    },
    {
      "href": "https://blog.pragmaticengineer.com/builder-ai-did-not-fake-ai/",
      "label": "Pragmatic Engineer — Builder.ai did not \"fake AI with 700 engineers\"",
      "type": "a"
    },
    {
      "href": "http://intercom.com/fin",
      "label": "Fin — $0.99 per resolution (official)",
      "type": "a"
    },
    {
      "href": "https://www.intercom.com/pricing",
      "label": "Pricing",
      "type": "a"
    },
    {
      "href": "https://www.intercom.com/learning-center/ai-customer-service-agent-pricing-comparison",
      "label": "AI agent pricing comparison (Fin vs Zendesk vs Agentforce)",
      "type": "a"
    },
    {
      "href": "https://www.searchunify.com/resource-center/blog/the-real-cost-of-ai-agents-maintenance-model-drift-and-hidden-ownership-costs/",
      "label": "The Real Cost of AI Agents: Maintenance, Model Drift, and Hidden Ownership Costs",
      "type": "a"
    }
  ],
  "repair_version": 2,
  "repaired_at": "2026-10-14T10:50:20Z",
  "source_url": "https://dev.to/debashish_ghosal/you-arent-choosing-an-ai-tool-youre-choosing-who-gets-paged-at-2-am-5g4a",
  "papers": [
    {
      "status": "ok",
      "url": "https://www.accc.gov.au/system/files/accc-v-microsoft-concise-statement-27-october-2025.pdf",
      "bytes": 1679788,
      "reused": false,
      "local_path": "papers/repair/3096b790342aa370d327.pdf"
    }
  ],
  "actual_video_files": [],
  "body_chars": 18734,
  "expected_body_images": 0,
  "discovery_status": "ok",
  "delivery_status": "ok",
  "missing_reason": "",
  "raw_html": "attachments/d44a6cff1944/pages/source.html",
  "local_html": "attachments/d44a6cff1944/pages/article.html",
  "body_base_path": "attachments/d44a6cff1944/pages/",
  "body_markdown_path": "attachments/d44a6cff1944/pages/article.md"
}