[{"data":1,"prerenderedAt":696},["ShallowReactive",2],{"thinking-index":3},[4,241,380,441],{"id":5,"title":6,"body":7,"date":227,"description":228,"extension":229,"meta":230,"navigation":231,"path":232,"seo":233,"stem":234,"tags":235,"__hash__":240},"thinking\u002Fthinking\u002Flifting-as-we-climb-mentoring-in-the-age-of-ai.md","Lifting as We Climb, Why Mentoring Associate Developers Matters More Than Ever",{"type":8,"value":9,"toc":218},"minimark",[10,18,21,24,27,44,47,52,55,58,61,64,68,71,74,87,93,105,108,130,134,137,140,143,149,158,162,165,168,171,191,195,198,201,207,210],[11,12,13],"p",{},[14,15],"img",{"alt":16,"src":17},"Mentoring associate developers in the age of AI engineering","https:\u002F\u002Fcdn.marblism.com\u002FywIExeJOksS.webp",[11,19,20],{},"I trace my career back to a single moment: a senior engineer took a chance on someone who didn't yet have all the answers.",[11,22,23],{},"Early on, I had mentors who demanded discipline, checked my logic with real rigor, and passed down technical craft the hard way. They didn't just hand me answers. They taught me how to think, how to debug under pressure, and how to take personal accountability for every line of code I pushed into production.",[11,25,26],{},"That experience shaped how I think about this industry: we rise by lifting others.",[11,28,29,30,37,38,43],{},"Today, as a ",[31,32,36],"a",{"href":33,"rel":34},"https:\u002F\u002Fwww.nathan-collins.com\u002Fabout",[35],"nofollow","player-coach"," who still writes production code and builds engineering teams, I've watched how much has changed in how we build software. AI tools have changed how fast we ship. When I took ",[31,39,42],{"href":40,"rel":41},"https:\u002F\u002Fwww.nathan-collins.com\u002Fexperience",[35],"ClearSignal"," from concept to production in twelve months, AI helped us move faster. But velocity isn't the same as judgment. Code generation isn't the same as architecture.",[11,45,46],{},"As the industry leans harder into AI tooling, mentorship isn't becoming obsolete. It's turning into the advantage that's hardest to copy.",[48,49,51],"h2",{"id":50},"why-llms-cant-replace-human-experience","Why LLMs can't replace human experience",[11,53,54],{},"AI writes syntax fast. It generates boilerplate, suggests refactors, and scaffolds endpoints in seconds.",[11,56,57],{},"Yet it can't mentor.",[11,59,60],{},"An LLM doesn't know what it feels like to carry the pager at 3am when a financial ledger goes down in production. It doesn't feel the weight of regulatory compliance in fintech or credit, where a mistake carries real legal and financial consequences. And it definitely can't sit across from a nervous stakeholder and translate business risk into technical architecture.",[11,62,63],{},"Lean only on generated code without building human capability, and you end up with plenty of syntax and not much judgment to maintain it. That's why senior engineers need to stay hands-on with developing juniors. Someone has to teach the context and judgment a model can't.",[48,65,67],{"id":66},"the-syntax-trap","The syntax trap",[11,69,70],{},"Junior developers today face a trap that didn't really exist before: they can ship complex features before they understand the systems underneath them.",[11,72,73],{},"Ask an LLM to write a database migration or an authentication flow and the result can look right on the surface. Without the underlying training, it's easy to miss what isn't visible in the code:",[75,76,77,81,84],"ul",{},[78,79,80],"li",{},"Edge cases in distributed transactions.",[78,82,83],{},"Race conditions under high concurrent load.",[78,85,86],{},"Security vulnerabilities hidden inside generated abstractions.",[11,88,89],{},[14,90],{"alt":91,"src":92},"System architecture and design planning","https:\u002F\u002Fcdn.marblism.com\u002F-W7cFZ4H8jr.webp",[11,94,95,96,100,101,104],{},"This is where my early mentors shaped my discipline, and it's where I put my energy today. I ask our teams to understand the ",[97,98,99],"em",{},"why"," behind every system, not just the ",[97,102,103],{},"what",".",[11,106,107],{},"When I mentor associate developers, I hold to three rules:",[109,110,111,118,124],"ol",{},[78,112,113,117],{},[114,115,116],"strong",{},"Understand the baseline."," Know how to trace a request manually through the stack before you ask an AI tool to optimize it.",[78,119,120,123],{},[114,121,122],{},"Interrogate the output."," Treat AI-generated code with healthy skepticism. If you can't explain every line and its failure modes, it doesn't get merged.",[78,125,126,129],{},[114,127,128],{},"Study failure."," Learn how systems break in production, not just how they pass unit tests locally.",[48,131,133],{"id":132},"bridging-strategy-and-craft","Bridging strategy and craft",[11,135,136],{},"Technical mentorship isn't an HR checkbox. It affects the business directly.",[11,138,139],{},"Companies in regulated industries don't hire engineering leaders just to write code. They hire us to connect business strategy to technical execution, and to build platforms that survive audits, data privacy rules, and the reliability bar enterprise customers expect.",[11,141,142],{},"Junior developers who get real mentorship grow into senior engineers who understand why the business cares. They learn that software exists to solve commercial problems and meet regulatory requirements, not to chase vanity metrics or clever abstractions nobody asked for.",[11,144,145],{},[14,146],{"alt":147,"src":148},"Pair programming and collaborative code review","https:\u002F\u002Fcdn.marblism.com\u002FmKq-RojTjke.webp",[11,150,151,152,157],{},"I write more about this in ",[31,153,156],{"href":154,"rel":155},"https:\u002F\u002Fwww.nathan-collins.com\u002Fthinking",[35],"my other essays"," on architecture and delivery, but the short version is: push for speed and reliability at the same time, not one at the expense of the other. I want every engineer on the team to see how their technical decisions show up on the bottom line.",[48,159,161],{"id":160},"paying-it-forward","Paying it forward",[11,163,164],{},"Being a hands-on technical leader means staying out of the ivory tower. Plans that never touch the codebase go nowhere. Code without a plan behind it doesn't go anywhere useful either.",[11,166,167],{},"Leadership works the same way. Authority without mentorship just calcifies.",[11,169,170],{},"The way I see it, I owe my own mentors a debt, and the only way to repay it is forward: take a chance on someone who shows promise, hand them real architectural problems instead of just bug fixes, and coach them through the mistakes honestly.",[75,172,173,179,185],{},[78,174,175,178],{},[114,176,177],{},"Design real challenges."," Give juniors scoped architectural tasks, not just bug fixes.",[78,180,181,184],{},[114,182,183],{},"Turn code review into teaching."," Use pull requests to explain trade-offs and design decisions, not just flag mistakes.",[78,186,187,190],{},[114,188,189],{},"Model accountability."," Own your failures in front of the team, fix them, and talk about what you learned.",[48,192,194],{"id":193},"the-long-game","The long game",[11,196,197],{},"AI makes individual engineers faster. It doesn't make a platform resilient on its own. People do that.",[11,199,200],{},"Mentoring junior developers isn't about teaching syntax. It's about passing on the judgment and discipline that keeps software running when it actually matters: when a ledger can't be wrong, when an outage costs real money.",[11,202,203],{},[14,204],{"alt":205,"src":206},"Engineering career growth and leadership","https:\u002F\u002Fcdn.marblism.com\u002FnDsUZLZL2TY.webp",[11,208,209],{},"We lift as we climb. The best way to repay the mentors who took a chance on us is to do the same for whoever's next.",[11,211,212,213,104],{},"If you need engineering leadership that's still hands-on with the code and serious about building the team behind it, ",[31,214,217],{"href":215,"rel":216},"https:\u002F\u002Fwww.nathan-collins.com\u002Fcontact",[35],"let's talk",{"title":219,"searchDepth":220,"depth":220,"links":221},"",2,[222,223,224,225,226],{"id":50,"depth":220,"text":51},{"id":66,"depth":220,"text":67},{"id":132,"depth":220,"text":133},{"id":160,"depth":220,"text":161},{"id":193,"depth":220,"text":194},"2026-08-01","AI writes code fast, but it can't mentor. Why senior engineers need to stay hands-on with developing junior talent as AI reshapes how software gets built.","md",{},true,"\u002Fthinking\u002Flifting-as-we-climb-mentoring-in-the-age-of-ai",{"title":6,"description":228},"thinking\u002Flifting-as-we-climb-mentoring-in-the-age-of-ai",[236,237,238,239],"mentorship","leadership","engineering","ai","D1esQtmQNR7OnmGVY3-42TqkGeKXa2hLebcZHfiHICc",{"id":242,"title":243,"body":244,"date":369,"description":370,"extension":229,"meta":371,"navigation":231,"path":372,"seo":373,"stem":374,"tags":375,"__hash__":379},"thinking\u002Fthinking\u002Fyour-AI-agents-are-probably-just-traditional-software.md","Your AI Agents Are Probably Just Traditional Software",{"type":8,"value":245,"toc":367},[246,250,253,256,259,262,265,285,288,291,294,297,300,303,306,316,319,322,325,328,331,334,337,340,343,346,349,352,355,358,361,364],[247,248,243],"h1",{"id":249},"your-ai-agents-are-probably-just-traditional-software",[11,251,252],{},"I know that sounds like I’m trying to be provocative, but the more I build this stuff, the more obvious it becomes.",[11,254,255],{},"I’ve spent this weekend working on a small company-monitoring agent. The bit everyone would call “AI” is fairly narrow: I’m using Gemini to help retrieve, interpret and classify business information.",[11,257,258],{},"That bit matters. It’s useful.",[11,260,261],{},"But it is not the whole product.",[11,263,264],{},"Around that, I still had to build:",[75,266,267,270,273,276,279,282],{},[78,268,269],{},"a PostgreSQL data layer",[78,271,272],{},"Docker orchestration",[78,274,275],{},"MCP tools using FastMCP",[78,277,278],{},"Python logic",[78,280,281],{},"an HTTP wrapper around Companies House",[78,283,284],{},"the execution logic for when the process should run, what it should check, and what happens next",[11,286,287],{},"In other words, most of the work was ordinary software engineering.",[11,289,290],{},"This is where I think the current language around AI gets messy.",[11,292,293],{},"AI is often talked about as if it sits outside the normal software engineering process. As if once there’s a model involved, we’re suddenly in a different world where architecture, state, data modelling, permissions, retries, monitoring, testing and deployment are secondary details.",[11,295,296],{},"They are not.",[11,298,299],{},"The model is a component.",[11,301,302],{},"A powerful component, yes. A very useful one. But still a component.",[11,304,305],{},"A lot of “AI agent” work is really:",[307,308,313],"pre",{"className":309,"code":311,"language":312,"meta":219},[310],"language-text","Prompt → model call → tool\u002FAPI call → state update → next step → repeat.\n","text",[314,315,311],"code",{"__ignoreMap":219},[11,317,318],{},"That can be valuable. I’m not dismissing it at all.",[11,320,321],{},"But the orchestration is software.",[11,323,324],{},"The data layer is software.",[11,326,327],{},"The integrations are software.",[11,329,330],{},"The reliability work is software.",[11,332,333],{},"The scheduling is software.",[11,335,336],{},"The permissions and audit trail are software.",[11,338,339],{},"Only part of the system is AI.",[11,341,342],{},"I think this distinction matters because the way we talk about AI is starting to create a strange split in the engineering world.",[11,344,345],{},"“AI people” on one side.",[11,347,348],{},"“Software engineers” on the other.",[11,350,351],{},"That feels wrong to me.",[11,353,354],{},"The best AI products I’ve seen are not built by treating AI as magic, or as something separate from engineering. They’re built by people who understand how to put a model inside a well-designed system.",[11,356,357],{},"The hype will settle down eventually. It always does.",[11,359,360],{},"And when it does, I think we’ll be left with a much clearer view:",[11,362,363],{},"AI is not replacing software engineering.",[11,365,366],{},"It is becoming part of it.",{"title":219,"searchDepth":220,"depth":220,"links":368},[],"2026-05-04","A software engineer's reflection on why many AI agents are really conventional software systems with a model in the loop.",{},"\u002Fthinking\u002Fyour-ai-agents-are-probably-just-traditional-software",{"title":243,"description":370},"thinking\u002Fyour-AI-agents-are-probably-just-traditional-software",[239,376,377,378,238],"software-engineering","ai-agents","agentic-ai","SlC7HBCHJTgYIoWRjtD_CIO_uNzmEYRYrX4lr418eDI",{"id":381,"title":382,"body":383,"date":432,"description":433,"extension":229,"meta":434,"navigation":231,"path":435,"seo":436,"stem":437,"tags":438,"__hash__":440},"thinking\u002Fthinking\u002Fwhat-makes-a-great-cto.md","What Makes a Great CTO?",{"type":8,"value":384,"toc":426},[385,391,394,398,401,405,412,416,419,423],[11,386,387,388],{},"The question I get asked most often is: ",[97,389,390],{},"what does a great CTO actually do?",[11,392,393],{},"The honest answer is that it depends on the company stage, the board, the CEO, and the engineering team. But there are a handful of qualities I've seen consistently separate the best technology leaders from the rest.",[48,395,397],{"id":396},"clarity-over-cleverness","Clarity over cleverness",[11,399,400],{},"The best CTOs I've admired don't try to be the smartest person in the room. They make the room smarter. They translate ambiguous business problems into clear engineering direction, and they translate technical constraints back into business language that the board can act on.",[48,402,404],{"id":403},"hiring-as-a-core-skill","Hiring as a core skill",[11,406,407,408,411],{},"A CTO who can't attract and retain exceptional engineers is limited by their own ceiling. The best ones build a reputation, internally and externally, that makes people ",[97,409,410],{},"want"," to work with them. They know that culture is the product of every hire and every exit.",[48,413,415],{"id":414},"knowing-when-to-go-fast-and-when-to-slow-down","Knowing when to go fast and when to slow down",[11,417,418],{},"Technical debt is a tool, not a failure. The skill is knowing when accruing it is the right trade-off and when it's quietly sinking the ship. Great CTOs hold the tension between velocity and sustainability, and they communicate that trade-off honestly upwards.",[48,420,422],{"id":421},"more-to-come","More to come",[11,424,425],{},"These are early thoughts. I'll be expanding on each of these themes as I work through what I want to write about here. If any of this resonates (or you disagree), I'd love to hear from you.",{"title":219,"searchDepth":220,"depth":220,"links":427},[428,429,430,431],{"id":396,"depth":220,"text":397},{"id":403,"depth":220,"text":404},{"id":414,"depth":220,"text":415},{"id":421,"depth":220,"text":422},"2026-01-15","Reflections on the qualities that define exceptional technology leadership and how engineering culture shapes company outcomes.",{},"\u002Fthinking\u002Fwhat-makes-a-great-cto",{"title":382,"description":433},"thinking\u002Fwhat-makes-a-great-cto",[237,238,439],"culture","enVsN1LzPvZT9dfANSVKckH7lmw1mR_7nzGu9waRhUQ",{"id":442,"title":443,"body":444,"date":690,"description":219,"extension":229,"meta":691,"navigation":231,"path":692,"seo":693,"stem":694,"tags":690,"__hash__":695},"thinking\u002Fthinking\u002Fthe-ultimate-guide-to-mvp-development-in-regulated-tech.md","The Ultimate Guide to MVP Development in Regulated Tech",{"type":8,"value":445,"toc":681},[446,449,455,458,461,464,467,471,474,477,484,487,494,500,504,511,514,517,537,544,548,551,554,557,582,585,591,595,598,601,604,607,610,614,617,637,640,644,647,650,653,659,663,671,678],[247,447,443],{"id":448},"the-ultimate-guide-to-mvp-development-in-regulated-tech",[11,450,451],{},[14,452],{"alt":453,"src":454},"A minimalist illustration showing a solid architectural bridge connecting a blueprint plan to a glowing finished structure, using soft warm neutral tones and clean simple shapes.","https:\u002F\u002Fimage.pollinations.ai\u002Fprompt\u002FFlat%20minimalistic%20illustration%20of%20a%20modern%20architectural%20bridge%20connecting%20a%20technical%20blueprint%20to%20a%20shining%20completed%20building%20on%20a%20clean%20horizon.%20Soft%20warm%20neutral%20tones,%20tan,%20cream,%20and%20charcoal.%20Simple%20geometric%20shapes,%20no%20clutter,%20professional%20and%20elegant.?width=1280&height=720&nologo=true",[11,456,457],{},"The \"move fast and break things\" mantra is a liability in regulated industries. In credit, compliance, and financial services, breaking things means legal trouble, lost licenses, and board-level disasters.",[11,459,460],{},"I have spent 20 years building software in environments where failure is not an option. I have seen countless projects stall because they treat compliance as a late-stage hurdle rather than a foundational feature.",[11,462,463],{},"Building a Minimum Viable Product (MVP) in a regulated space requires a different playbook. You must bridge the gap between aggressive business timelines and the rigid requirements of regulators.",[11,465,466],{},"I take products from concept to production in 12 months. This is how I do it without sacrificing compliance or quality.",[48,468,470],{"id":469},"ship-in-12-months-not-3-years","Ship in 12 Months, Not 3 Years",[11,472,473],{},"Speed is a competitive advantage, even in fintech. The traditional approach involves months of abstract requirements gathering followed by years of development. By the time the product ships, the market has moved.",[11,475,476],{},"I focus on the \"Smallest Viable Product.\" This is the narrowest possible scope that delivers a compliant, revenue-generating result.",[11,478,479,480,483],{},"When I built ",[31,481,42],{"href":40,"rel":482},[35],", we moved from concept to production in exactly 12 months. We didn't do this by cutting corners. We did it by aggressively narrowing the scope to a single core transaction flow.",[11,485,486],{},"We focused on one credit product, one specific customer segment, and one regulatory jurisdiction. We ignored the \"nice-to-haves\" and built a real-time risk-signal pipeline that worked.",[11,488,489,490,493],{},"If you want to ship fast, you must decide what you are ",[97,491,492],{},"not"," building.",[11,495,496],{},[14,497],{"alt":498,"src":499},"Minimalist icon-style illustration of a clock and a shield overlapping, representing speed and security combined. Soft neutral palette.","https:\u002F\u002Fimage.pollinations.ai\u002Fprompt\u002FMinimalist%20flat%20illustration%20of%20a%20stylized%20clock%20face%20integrated%20with%20a%20protective%20shield.%20Soft%20warm%20tones,%20beige%20and%20muted%20gold.%20Clean%20vector%20lines,%20simple%20shapes,%20no%20text.?width=800&height=600&nologo=true",[48,501,503],{"id":502},"build-for-compliance-not-just-features","Build for Compliance, Not Just Features",[11,505,506,507,510],{},"In regulated tech, compliance ",[97,508,509],{},"is"," the product. You cannot launch and \"fix the KYC later.\"",[11,512,513],{},"I treat regulatory requirements as the first set of unit tests. If the system can't prove who the customer is (KYC) or where the money is going (AML), the feature doesn't exist.",[11,515,516],{},"My approach involves three pillars:",[109,518,519,525,531],{},[78,520,521,524],{},[114,522,523],{},"Compliance as Code",": We bake identity verification and transaction monitoring directly into the onboarding flow.",[78,526,527,530],{},[114,528,529],{},"Auditability by Design",": Every decision the system makes must have an evidence-backed rationale. In the credit space, \"the AI said so\" is not an acceptable answer to a regulator.",[78,532,533,536],{},[114,534,535],{},"Security-First Architecture",": Use AES-256 encryption, tokenization, and multi-factor authentication from day one.",[11,538,539,540,543],{},"When we built the multi-LLM layer for ClearSignal, every AML and KYB event was surfaced with a traceable evidence trail. We didn't add the audit logs at the end; we built the system ",[97,541,542],{},"around"," the need for auditability.",[48,545,547],{"id":546},"agentic-engineering-your-force-multiplier","Agentic Engineering: Your Force Multiplier",[11,549,550],{},"I don't just use AI to write code. I use agentic engineering to build better products faster.",[11,552,553],{},"Agentic engineering involves goal-driven AI agents that handle multi-step development and operational tasks. This reduces technical debt and allows a lean team to achieve what usually requires an army of engineers.",[11,555,556],{},"I use these tools to:",[75,558,559,565,571],{},[78,560,561,564],{},[114,562,563],{},"Automate Testing",": Agents can generate and run complex test suites that simulate thousands of edge-case financial transactions.",[78,566,567,570],{},[114,568,569],{},"Monitor Compliance",": I build autonomous agents that scan live market signals and surface risk events in real-time.",[78,572,573,576,577,581],{},[114,574,575],{},"Run agentic workflows",": In ",[31,578,580],{"href":40,"rel":579},[35],"StrikeStack",", I built an AgentOps control plane for configuring agents, starting work from forms, schedules and events, monitoring runs, and routing structured outputs to review surfaces.",[11,583,584],{},"This isn't about replacing engineers. It’s about freeing them from \"plumbing\" so they can focus on the hard logic of your business strategy.",[11,586,587],{},[14,588],{"alt":589,"src":590},"A minimalist diagram of interconnected nodes and light lines representing an AI network, using soft neutral colors and a clean layout.","https:\u002F\u002Fimage.pollinations.ai\u002Fprompt\u002FMinimalist%20flat%20illustration%20of%20interconnected%20geometric%20nodes%20forming%20a%20network.%20Soft%20neutral%20tones,%20light%20grey,%20tan,%20and%20cream.%20Simple%20and%20clean%20aesthetic,%20representing%20AI%20and%20agentic%20systems.?width=800&height=600&nologo=true",[48,592,594],{"id":593},"bridge-the-boardroom-and-it","Bridge the Boardroom and IT",[11,596,597],{},"Most engineering leaders fail because they speak a language the board doesn't understand. They talk about \"sprints,\" \"refactoring,\" and \"latency.\"",[11,599,600],{},"I talk about business outcomes.",[11,602,603],{},"I bridge the gap by working directly with leadership to understand the growth strategy. Then, I translate that strategy into a technical roadmap that prioritizes revenue and risk mitigation.",[11,605,606],{},"As a hands-on Head of Engineering, I don't just manage people. I write production code, I design database schemas, and I negotiate with stakeholders. I take full accountability for the results.",[11,608,609],{},"If the board needs to know why we are prioritizing a specific infrastructure change, I don't give them a lecture on cloud architecture. I show them how it reduces the cost per transaction or speeds up customer onboarding by 40%.",[48,611,613],{"id":612},"measure-what-matters-business-outcomes","Measure What Matters: Business Outcomes",[11,615,616],{},"I don't measure success by the number of tickets closed or the frequency of deployments. I measure success by:",[75,618,619,625,631],{},[78,620,621,624],{},[114,622,623],{},"Shipped Products",": Is the code live and serving users?",[78,626,627,630],{},[114,628,629],{},"Paying Customers",": Is the technology actually generating revenue?",[78,632,633,636],{},[114,634,635],{},"Regulatory Approval",": Can we pass an audit tomorrow?",[11,638,639],{},"In the regulated space, \"done\" means the product is live, compliant, and helping the business grow. Anything else is just expensive typing.",[48,641,643],{"id":642},"the-player-coach-advantage","The Player-Coach Advantage",[11,645,646],{},"I am a \"player-coach.\" I can hire a high-performing engineering team and I can also jump into the IDE to fix a critical bug in a payment gateway.",[11,648,649],{},"This dual capability is essential for MVPs. In the early stages, you don't need a manager who sits in meetings all day. You need a leader who can design the architecture in the morning and lead a board meeting in the afternoon.",[11,651,652],{},"I provide the senior technical leadership needed to build from the ground up while ensuring the product aligns with high-level business goals.",[11,654,655],{},[14,656],{"alt":657,"src":658},"A minimalist illustration of a person standing between a desk with a computer and a boardroom table, bridging the two spaces. Soft warm tones.","https:\u002F\u002Fimage.pollinations.ai\u002Fprompt\u002FFlat%20minimalistic%20illustration%20of%20a%20professional%20figure%20standing%20confidently%20between%20a%20technical%20workspace%20and%20a%20corporate%20boardroom.%20Soft%20warm%20neutral%20tones.%20Clean%20simple%20shapes,%20no%20faces,%20modern%20and%20minimalist.?width=800&height=600&nologo=true",[48,660,662],{"id":661},"ready-to-build","Ready to build?",[11,664,665,666,670],{},"Building an MVP in credit or financial services is a high-stakes game. You need a partner who understands the nuances of ",[31,667,669],{"href":33,"rel":668},[35],"regulated industries"," and has the hands-on skills to ship fast.",[11,672,673,674,104],{},"I have 20 years of experience making sure technology serves the bottom line. If you are a board member or executive looking for a Head of Engineering to lead your next product build, ",[31,675,677],{"href":215,"rel":676},[35],"let’s talk",[11,679,680],{},"I don't just build software. I build businesses.",{"title":219,"searchDepth":220,"depth":220,"links":682},[683,684,685,686,687,688,689],{"id":469,"depth":220,"text":470},{"id":502,"depth":220,"text":503},{"id":546,"depth":220,"text":547},{"id":593,"depth":220,"text":594},{"id":612,"depth":220,"text":613},{"id":642,"depth":220,"text":643},{"id":661,"depth":220,"text":662},null,{},"\u002Fthinking\u002Fthe-ultimate-guide-to-mvp-development-in-regulated-tech",{"title":443,"description":219},"thinking\u002Fthe-ultimate-guide-to-mvp-development-in-regulated-tech","DW6zJpzpg3zpwEIckz15pAIaui6wgLNAi_LEVT-9io8",1785781229199]