Surviving the synthetic stampede: shielyule.uk goes live

Somewhere between the third quarterly slide deck promising an automated corporate Nirvana and the cold realisation that your department’s pilot bot has started advising payroll to liquidate capital assets into Solana memecoins, a quiet dread sets in.

A foggy landscape filled with old computer monitors displaying digital screens, surrounded by abandoned office chairs.

We were told the synthetic minds would quietly optimise supply chains while we sipped cold brew and contemplated higher-order strategy. Instead, we find ourselves locked in an administrative trench run with autonomous agents that hallucinate compliance certificates with unblinking, sociopathic confidence.

To prove that enterprise-grade sanity can still be enforced against the digital abyss, I have launched a new containment terminal: shielyule.uk.

Governing the Machine

For years, evidence of my running battles against runaway endpoints, rogue API keys, and proof-of-concept sprawl lived scattered across Jira backlogs, frantic late-night Slack threads, edge GitLab pipelines, and the existential ruminations of this blog.

The new terminal aggregates the entire survival manual into one hardened perimeter. It documents my actual work as an AI Delivery Lead and Developer—focusing strictly on what happens when bleeding-edge synthetic cognition collides head-on with regulated reality, evidentiary audit trails, and unforgiving enterprise architecture.

Here, governed AI is not a performative bullet point tacked onto the bottom of a vendor sales pitch to soothe risk committees. It is an active blast door.

The Arsenal Under the Hood

True enterprise delivery isn’t merely clicking “deploy” inside a sterile corporate tenant and praying the model behaves. It’s an asymmetric arms race between raw, untamed computational power and the regulatory machinery meant to cage it.

To keep the perimeter secure, my workbench doesn’t rely on corporate vendor lock-in alone:

  • The Nerve Center & Codeforge: Forged at speed inside Cursor, tracked through rigorous GitLab CI/CD gates, and pushed to the edge via Vercel and Cloudflare, proving that enterprise-grade rigor doesn’t have to move at the glacial pace of a steering committee.
  • The Intelligence Grid: Orchestrating multimodal engines across Gemini, stress-testing reasoning boundaries against Grok, running sovereign open-weights through Nemotron, and wiring autonomous negotiation protocols with WebMCP.
  • The Data Panopticon: Ingesting millions of telemetry rows through GCP and BigQuery, visualized into brutal clarity via Looker, and paired with deep cultural taste-graph heuristics powered by Qloo.

And when those wild prototypes are ready for the sterile corporate floor? That is where the enterprise straitjacket goes on: ring-fencing endpoints within Azure AI Foundry, enforcing orchestrations in Copilot Studio, and clamping Microsoft Purview surveillance tight around the data estate’s digital throat so an agent doesn’t inadvertently leak confidential board minutes to an unvetted third party.

The Dossier: What Lives on shielyule.uk

  • Architecture & Enforcement: Turning wild prompts and unruly LLMs into disciplined, deterministic, human-in-the-loop systems that survive pen-tests, ISO frameworks, and GDPR audits.
  • The Battle-Tested Arsenal: Hardened builds engineered to survive the friction of the modern panopticon:
    • Tribunal Ally — Leveling the playing field when institutional inertia and administrative machines grind citizens down.
    • GhostMerchant — Forensic spend control, delivery probes, and poison-pill merchant counter-measures for the automated commerce era.
    • DealTable — Synthetic, high-stakes negotiation simulation environments powered by WebMCP where autonomous agents test human mettle.
    • GridRecord — Real-time telemetry and intelligence dashboards mapping the creaking, high-voltage spine of national energy grids using NESO and DESNZ data pipelines.
  • Enterprise Battle Scars: Decades of delivery leadership across high-voltage energy networks, tier-one financial institutions, government apparatus, and global consulting conglomerates—environments where “failure” isn’t a retrospective talking point, but an existential regulatory catastrophe.
  • Extraction & Engagement: Clear coordinates for securing interim delivery leadership, AI architecture, and agile recovery before your shadow IT experiments breach containment.
A dark containment cell door marked with 'AUTHORIZED PERSONNEL ONLY' and warning signs. The door features blue lighting around its edges and is covered in yellow caution tape that reads 'DO NOT CROSS' and 'DIGITAL HAZARD.' Purple mist emanates from the base of the door, adding a mysterious effect.

Pragmatists Over Zealots

This dossier is engineered specifically for programme directors, risk executives, and enterprise hiring leads who know the chilling truth: an unmonitored agentic pilot built on vendor optimism will rot inside shadow IT, hallucinate a catastrophic compliance breach, and leave you holding the forensic bill.

If you need someone who can sling TypeScript and Python in Cursor by morning, orchestrate complex data pipelines across BigQuery and GCP by afternoon, and defend the architectural boundary in front of an enterprise audit board before close of play—that is what this terminal represents.

Why Break the Seal on a New Domain?

This domain remains my intellectual safehouse for dystopian dispatches, guerrilla technical commentaries, and chronicling the absurdity of corporate silicon worship.

shielyule.uk is the sterile tactical operations deck: a lightning-fast build deployed cleanly to edge nodes, built to satisfy web crawlers, risk officers, and enterprise recruiters alike without an ounce of bureaucratic bloat.

Operating from Scotland, I am available immediately for AI delivery leadership, AI rollouts, and enterprise cloud migrations that require someone who knows how to hold the perimeter.

A cartoon owl with glasses and a mustache, holding a smartphone and looking at the screen, set against a circular background with stars.

GridRecord.com – the Liturgy of the Laminated Core

An illustrated diagram titled 'The Ghost Interconnection Queue' depicting a bureaucratic system for renewable energy connection. It shows wind turbines and solar farms highlighting the interconnection process, average wait times of 7-14 years, and financial aspects like customer standing charges and shareholder profits. Key areas include impact assessments, environmental reviews, and technical studies, all emphasizing the challenges faced by clean energy initiatives.

Somewhere out in the rain-blackened moss beyond the A1(M), inside an unmapped compound defended by six strands of rusted barbed wire and a padlock that hasn’t yielded to a brass key since the Falklands War, sits a 400kV transmission transformer.

It does not merely buzz. It resonates with a faint, predatory nausea tuned precisely to 49.98 Hertz—the exact frequency at which human marrow curdles into cold lard.

It is forty-three years old. Its blistered casing is the shade of institutional grey porridge, and its mineral oil smells of dead generators dipped in rancid beef tallow. Inside its belly, sixty miles of oil-soaked cellulosic paper insulation are slowly decomposing into furanic sludge under thermal stress.

Illustration of a 400kV Supergrid Backbone showing various components including a transformer, power lines, and warning labels regarding dissolved hydrogen levels and dielectric oil concerns. Additionally, it highlights factors such as the British weather and an actuarial diagram showing yield information.

In a technologically coherent civilisation, this thirty-ton slab of groaning iron would be an object of state veneration. It would have a dedicated human custodian named Dave standing beside it with a calibrated Fluke acoustic camera, an infrared pyrometer, and an oil filter wrench—a man whose singular mandate is ensuring forty thousand induction hobs in Dunfermline don’t vaporise midway through an England penalty shootout.

Dave does not exist. Dave was culled in the 2007 “Lean Infrastructure Optimisation” drive. Dave’s institutional memory—specifically, knowing that if you tap the tap-changer housing with a copper mallet on a damp Tuesday, the phase angle doesn’t drift by four degrees—was written down as a £42,000 severance efficiency.

Instead, four hundred miles south in a glass terrarium cantilevered over the Thames, there is Julian. Julian is twenty-six. Julian possesses a first-class degree in Financial Econometrics, drinks oat cortados pulled at 9 bars of pressure, wears pristine vegan trainers, and has never inhaled atomised naphthenic hydrocarbons.

Julian is currently massaging cell H42 of RIIO_T3_Totex_Incentive_Optimisation_v14_FINAL_v2(1).xlsx. If Julian trims the assumed cost of debt by just 0.04%, an eighteen-billion-pound Regulated Asset Base uncoils, spewing forth an interim distribution of 4.2 pence per share.

Down in the sodden loam, the transformer issues a wet, seismic cough. Dissolved gas analysis would show acetylene spiking to 80 ppm. A soggy pigeon perches atop the 400kV bushing, blissfully unaware that its hollow bones are the sole thermodynamic break preventing a synthetic inflation-hedged swap trade in Zurich from collapsing into ash.

The Tabernacle of RIIO: The Alchemy of Totex

If you wander into the high offices of National Grid plc or the regional Distribution Network Operators (DNOs) on a grey Tuesday, do not expect to find grease-stained engineering manifestos detailing fault-ride-through capability, sub-synchronous resonance, or cold-load pickup decay curves.

You will find an altar dedicated to the liturgical rites of the institutional capital allocator: CPIH-deflated weighted-average cost of capital (WACC), return on regulatory equity (RORE), and the supreme artifact of modern British administrative occultism—the RIIO-ED2 / T2 Price Control Framework.

Infographic illustrating the RIIO-ED2 extraction engine, featuring components like Raw CapEx, OpEx, and a compounding base asset, along with various financial metrics and processes involved in the extraction model.

RIIO is an intellectual labyrinth engineered by administrative economists who genuinely believe that if you draft an equation long enough, high-voltage copper will decline to oxidise.

Under classic commercial logic, capital expenditure is an operational cost to be husbanded through thrifty maintenance. Under the RIIO regulatory architecture, it is the primary engine of sovereign wealth extraction.

The mechanics are pure accounting witchcraft:

  1. The Totex Sleight of Hand: Operating expenses (maintaining trees, hiring people with spanners) and Capital expenses (dropping £80m on GIS switchgear) are blended into Total Expenditure (Totex).
  2. The Slow-Money Capitalisation: A fixed percentage of Totex (often 70% or more) is categorised as “Slow Money” and funneled directly into the Regulated Asset Value (RAV).
  3. The Index-Linked Extraction: The regulator guarantees an inflation-hedged, statutory return on the entire accumulated RAV—harvested directly from the non-negotiable standing charge welded to the electricity bills of every single-glazed terrace in the Kingdom.

In this looking-glass reality, fixing an old 33kV fluid-filled cable with a patch kit is an operational nuisance that hits in-period margins. Demolishing the site and installing an over-specified, gas-insulated SF6 monster is an equity event.

The national transmission network is not being run as an engineering system to evacuate offshore wind from Dogger Bank; it is operated as a low-risk, asset-backed sovereign debt vehicle that happens to have an inconvenient, sparking byproduct called “electricity.”

The Actuarial Deadband of the Freezing Freezer

Meanwhile, out in the damp, physical universe, thermodynamics continues to misbehave.

Creosoted spruce poles planted when Jim Callaghan was humming The Red Flag rot quietly below the soil line. Underground lead-sheathed cables laid during the Macmillan administration spontaneously decompose, quietly weeping dielectric fluid into the limestone aquifers.

When an 11kV spur trips during an unseasonal Atlantic blow, the rupture between physical reality and regulatory abstraction turns into pure, deadpan comedy.

A cartoon-style infographic illustrating the impact of a 70mph Atlantic gale, featuring elements like snapping utility poles, melting deep-freeze contents, and a financial ledger showing stable dividends despite weather disruptions.

To the family in rural Aberdeenshire watching their deep freezer liquefy into an ambient pool of venison and Viennetta while their breath clouds in the hall, a blackout is an intimate existential crisis. It is the immediate, alarming discovery that domestic continuity is merely a two-millimetre film of emulsion paint applied over howling, freezing mud.

To Julian’s spreadsheet, however, their hypothermia is an actuarial coordinate: Customer Interruptions (CI) and Customer-Minutes Lost (CML).

An outage is not a failure of stewardship; it is a parameter inside an Ofgem penalty-incentive collar. If an Atlantic depression brings down seven miles of line in the Cheviots, no one in the executive suite scrambles for a torch. They look at the cumulative CML balance. If the total interruption minutes sit safely within the negotiated regulatory deadband, the interim dividend remains immaculate and inviolable.

And if a winter storm is vicious enough to blow past the collar? It is classified as an “Exceptional Event”—a regulatory act of God that wipes the penalty clean and allows the restoration costs to be loaded directly onto the asset base for the subsequent price-control period.

You are sitting in the dark listening to your garden shed dismantle itself in ninety-mile-an-hour gusts, but take heart: the enterprise multiple on the regulated distribution license remains pristine.

The Paper Fortress: 2038 Interconnection Queues

Try to plug a 50MW battery storage installation or a community solar array into this network, and the curtain pulls back on the ultimate administrative bottleneck.

You submit your grid connection application. Fourteen months later, a formal document arrives from the network operator. In crisp, institutional prose, it informs you that your estimated energisation date is October 2037.

Illustration of a clean energy project depicting wind turbines and solar panels, along with an interactive substation labeled with waiting times for transmission study and energization estimates.

Why? Because the regional 132/33kV bulk supply point is thermally constrained, the upstream 400kV substation has zero fault-level headroom, and the queue ahead of you contains:

  • Three speculative 2-gigawatt data centres registered to post-office boxes in Tortola.
  • Six ghost battery storage applications submitted by private equity scouts who have no land rights, no planning permission, but a firm grip on queue positions they intend to flip for £80,000/MW.
  • A paper-shuffling transmission upgrade program governed by an ancient First-Come, First-Served queue rule that would make a Soviet bread line look like an agile sprint.

We are actively attempting to run an industrial clean-energy transition across a physical network whose telemetry still communicates using serial protocols and 1,200-baud modems devised during the Winter of Discontent.

The Receipts on the Slab: Enter GridRecord

The customary British response to this state of affairs is unfocused, sputtering anger: go onto social media, yell about corporate greed, and demand instant nationalisation.

It is totally useless. The institutional apparatus does not fear anger. It employs directors of corporate affairs whose sole vocation is to gaze into cameras with soft, sorrowful eyes and explain that “attracting low-cost institutional infrastructure capital from international credit markets is essential to lowering bills for hard-working families.”

You cannot shame an automated extraction framework. But you can cross-examine its telemetry.

The raw truth isn’t buried in encrypted Cayman servers. It is printed in plain text across their own legally mandated submissions: the Long-Term Development Statements (LTDS), the Network Development Plans (NDPs), the Regulatory Financial Statements, and the open SCADA logs.

  • Multi-billion-pound underlying operating margins running alongside connection queues quoting launch dates for domestic battery assets well into the late 2030s.
  • Predictable, ratchet-locked dividend streams extracted while 11kV rural circuits across the West Country and the Scottish borders still dangle from rotten timber stays waiting for the next moderate Atlantic low.
  • A regressive standing charge that extracts an identical daily tariff from an unheated pensioner’s flat in Paisley and a vacant penthouse in Mayfair, while the network’s copper assets are systematically sweated to the absolute margin of brittle fracture.

This is why we built gridrecord.com.

We are not running a protest group, and we are not issuing a polite consultation response. The civic task is far colder and far more surgical: drag the physical network’s autopsy and the financial ledger onto the exact same stainless-steel table, turn on the forensic lighting, and let the numbers speak.

On one side: the feeder faults, the furan oil measurements, the circuit-breaker trip counts, and the thirteen-year connection queues. On the other: the regulatory asset baselines, the cost-of-equity allocations, and the steady, silent extraction of capital from cold postcodes into offshore funds.

No slogans. No glossy public relations brochures. Just the cold, indisputable receipts of a system that bills you for the privilege of freezing in the dark while the spreadsheet prints money in the warm.

GridRecord launch

Somewhere out beyond the commuter belt, in a drizzle so persistent it feels institutional, an ancient substation transformer is quietly sizzling to itself. It has served through eight prime ministers, smells aggressively of rancid chip fat, and is vibrating at a pitch that suggests acute structural panic.

A detailed analysis dashboard featuring the National Grid (UK) balance sheet with financial data, including assets, liabilities, and dividend payout schedules, alongside various aged electrical components.

In a sane world, a grizzled technician named Dave would be watching over it with an infrared sensor and an adjustable spanner. In our world, Dave was made redundant during the mid-noughties efficiency drive. Instead, Dave has been replaced by Julian—a 26-year-old financial engineer sitting in a sleek glass box overlooking Canary Wharf, absent-mindedly adjusting cell H42 of a discounted cash flow spreadsheet.

If Julian tweaks his cost-of-capital assumption by 0.08%, an eighteen-billion-pound regulated asset base miraculously unlocks a dividend payout. Down in the mud, the transformer groans, drops frequency, and plunges forty thousand smart meters into the dark ages.

A split image comparing two characters: on the left, a worker named Dave in a yellow raincoat holding a wrench next to a control panel, emphasizing the importance of manual labor in keeping the lights on; on the right, a businessman named Julian using a computer mouse with financial documents and a city skyline in the background, highlighting corporate responsibilities and shareholder satisfaction.

Welcome to the Great British Energy Transition, where the electrons are entirely optional, and the real product is institutional yield.

Over the years, we didn’t just privatise our essential infrastructure—we transubstantiated it into a bespoke derivative product. Out in the sticks, a winter gale snaps sixty-year-old wooden poles like dried pasta, leaving your freezer thawing into a sad, lukewarm puddle of fish fingers. To you, it’s the sudden, bone-chilling revelation that modern civilisation is merely two hours of battery life away from the 17th century.

To the utility company’s financial model, however, your ruined evening is merely a manageable blip on an Ofgem penalty deadband. If the blackout minutes stay within the actuarial limit, the interim dividend remains completely untouched. You sit in freezing gloom trying to remember where you put the tea lights; the enterprise value of the distribution network license purrs along in pristine health.

The most chilling horror stories aren’t hidden behind state secrets or shadowy boardroom cabals. The real nightmare is published openly, cheerfully, and in glossy 4K across audited annual reports and investor consensus portals. Billions siphoned into executive long-term incentive schemes while standing charges climb remorselessly. Actual shovel-ready wind and solar projects are marooned in connection queues stretching into the late 2030s—their capacity locked up by speculative, spreadsheet-only battery arbitrage schemes that will never lay an inch of cable—all while obsolete 1960s switchgear is held together by prayers and high-grade gaffer tape.

A large, towering structure made of stacked books resembling a power line pylon, set against a cloudy sky with a marshy landscape in the foreground.

Ranting at the sky does nothing. What is required is a forensic, merciless audit trail.

Which brings me to what I’ve been quietly building.

This week marks the launch of my new project: GridRecord.co.uk.

GridRecord is designed to drag the entire pantomime out of opaque city boardrooms and lock it onto a verifiable balance sheet. It cross-examines the damp reality of Britain’s physical network—the blackouts, the frequency wobbles, the outage heatmaps, the decrepit backlogs—directly against the pristine numbers: corporate yields, regulated asset base inflations, and dividend extractions.

No spin. No regulatory fog. An unvarnished ledger of cause and consequence.

The full, unabridged dispatch on how we traded our national grid for an investment bank with pylons is live on the site right now. Run through the audit trail, examine the machinery for yourself, and see where your standing charge is really travelling:

Read the full exposé at GridRecord.co.uk — before the next transformer throws in the towel.

ComplaintPath – The Bureaucratic Panopticon Will See You Now

There is a distinct, rhythmic cadence to the collapse of late-stage civilization, and it rarely sounds like a meteor strike. More often, it sounds like the muffled, rhythmic thud of a high-speed printer devouring its three-hundredth successive “Final Response” letter, while somewhere in a windowless server farm in Slough, an autonomous corporate chatbot develops an existential crisis and begins filing complaints against its own motherboard.

A hand pressing a stamp marked 'WRONG DOOR' on a legal document labeled 'Petition for Jurisdictional Review', with digital displays indicating a rejection due to jurisdiction issues.

Welcome, dear reader, to the gleaming, algorithmically-lubricated labyrinth of modern UK consumer redress. If you have ever tried to dispute a rogue energy bill, extract a refund from a high-street bank that appears to view your capital as a personal venture fund, or inform a data controller that their cookie banner is roughly as legally compliant as a pirate radio station broadcasting from a North Sea oil rig, you will already know the drill.

You enter the Maze. You choose between three dozen distinct Ombudsmen, each possessing the administrative flexibility of a medieval drawbridge operated by a melancholic monk. You feed your grievances into a chatbot trained exclusively on corporate deflection poetry. And slowly, agonizingly, the statute of limitations bleeds out in a puddle of automated sympathy links.

It is beautiful, really. A masterclass in friction-as-a-service.

Enter ComplaintPath: Because Someone Had To Teach the Machines How to Say “No” With Precision

Naturally, when the good people at GenAI.works rolled out Track 01 of the #OpenAgentHackathon—Solving Fragmented Intelligence—we took one look at this bureaucratic abattoir and thought: What this desperately needs is more artificial intelligence.

Specifically, we built ComplaintPath.

It is not legal advice. It is not a ambulance-chasing claims company operating out of a caravan park in Lanarkshire. It is a cold, hyper-rational multi-agent pipeline designed to stare unflinchingly into the abyss of corporate rejection letters, check the statutory clock, and either open the exact right statutory door or politely, unblinkingly inform you that you are entirely out of luck.

You remain in total control. The machine prepares the ammunition; you pull the trigger. We never auto-file, primarily because giving an AI unmonitored access to the Financial Ombudsman portal feels like handing a toddler the launch codes to a Trident submarine.

Under the Hood: A Symphony of Corporate and Silicon Indifference

To build a machine capable of navigating the labyrinth without hallucinating a statutory deadline into existence, we had to enlist a rather formidable stack of sponsors who clearly share our dark appreciation for automated systems:

  • Zetaris (The Data Layer): Redress windows and portal URLs are not buried deep within some hallucinatory prompt hallucination. On every analysis run, our DataRetrieverAgent queries Zetaris live for the brutal, unvarnished truth of the scheme windows. In the product interface, it proudly declares Scheme data: Zetaris live. No fairy tales; just raw, relational architecture doing what bureaucrats pretend to do.
  • Meterless Nightshift & NVIDIA Nemotron (The Power): Letter comprehension runs through Meterless Nightshift on a dedicated Nemotron lane. It extracts facts with the icy detachment of a seasoned coroner. If a date is missing, the AI does not invent one—it simply stares at you with quiet disappointment.
  • Meterless World Model (The Cognition Engine): Every shared run state, entity, observation, and provenance trace lives here, ensuring that when the audit trail is requested, we can explain precisely which agent decided your complaint was six days too late.
  • Bright Data (The Web): When drafting pack section IV, Bright Data’s Web Unlocker refreshes context straight from the official scheme portals, ensuring our guidance doesn’t drift into folklore while you’re making a cup of tea.
  • Cursor (The Crucible): Scaffolding and shipping an entire multi-agent civic UI, adapters, gates, and PDF exporters in record time without throwing your MacBook out of a third-floor tenement window.
  • GenAI.works (The Arena): The wonderful madhouse that gave us the window to build Track 01.

The Four Horsemen of the Gate

When you feed a corporate refusal letter into ComplaintPath, the system reduces the boundless bureaucratic horror of modern Britain down to four immutable, code-enforced outcomes:

  1. Eligible: You may proceed. The system compiles a pristine markdown and PDF pack for you to take to the official portal.
  2. Premature: You jumped the gun. The corporate monolith is still legally permitted to ignore you for another eight weeks. Go back and wait in the dark.
  3. Out of Time: The clock has run out. The statute of limitations has turned your grievance into historical artifact. Pour one out for your lost consumer rights.
  4. Wrong Door: You are screaming at an energy supplier about a parking fine. The machine will gently redirect you before you embarrass yourself further.

The next step after download is wonderfully pedestrian: log on to the designated official portal, drop in your Gate Pack, and let the grinding wheels of statutory justice grind exceeding small.

Examine the Evidence

If you fancy auditing our descent into civic automation, or if you simply want to fork a repository that treats corporate obfuscation with the contempt it so richly deserves, the doors are wide open:

A tip of the hat to Zetaris, Meterless, NVIDIA, Bright Data, Cursor, and GenAI.works for providing the raw materials required to construct our little digital panopticon.

ComplaintPath is independent guidance software. It does not provide legal advice, and it certainly won’t save you from the heat death of the universe. But it might just help you get your £48.50 back from your broadband provider before the ice caps finish melting.

Undo – The Ghost of the Past Package

Why Your CI Pipeline is a Corporate Séance

A vintage, steampunk-inspired keyboard featuring a prominent red 'APPROVE' button, alongside a steaming mug of Earl Grey tea. A handwritten note warns, 'The previous package is still breathing. Do not wake it.'

Let’s be honest about how software delivery actually works in the wild.

For decades, the enterprise software ritual has followed a comforting, highly choreographed liturgy. You write some code. You sacrifice a chicken to the Jenkins gods. The CI pipeline turns green—a soothing, celestial lime-green glow that whispers sweet nothings into your ear, assuring you that all is right with the universe.

And then you deploy to production, and the server spontaneously combusts into a heap of weeping stack traces, leaving you frantically rummaging through the digital equivalent of a burning attic trying to find last Thursday’s release artifact.

Because here is the dirty little secret that the agile certifiers don’t want you to know: Most pipelines only ask one question: Does the new build work?

It’s a polite, forward-looking question. It’s optimistic. It’s hopelessly naive. In a real-world incident, the question that actually makes your stomach drop into your boots is far more brutal: Can we still run what we had before?

A stylized logo featuring a geometric fox design within a shield, accompanied by a green checkmark, set against a dark digital background.

Enter Undo: Necromancy for Cloud Run

That specific brand of existential dread is what birthed Undo—my entry for the GitLab Life After Code / Transcend hackathon over on Devpost.

The premise is delightfully cynical. Instead of blindly trusting that tomorrow’s shiny container image won’t accidentally erase your user database and summon ancient Sumerian demons into your AWS/GCP VPC, Undo implements a strict regime of digital necromancy.

It runs on a simple, ruthless loop:

  1. Prove: On every push to main, GitLab CI doesn’t just build the new toy; it resurrects the previous package, forces it to look into the mirror, and hits /health (ROLLBACK_OK) to verify it hasn’t lost its mind.
  2. Ask: If the past is still breathing, an event-driven GitLab Duo flow (no chat-theatre chatbots, just cold, hard automation) triggers automatically on pipeline success, reads the rollback proof, and opens a Work Item.
  3. Approve: I reply with one word: approve. (Because even in an age of autonomous silicon agents, a human being should still be the one holding the smoking revolver when production goes sideways).
  4. Promote: A manual promote_production job ships the exact same verified image out to the world.

You can prod the corpse yourself right here: Undo Live Health Check.

Why GitLab and Google Cloud Made the Séance Possible

This wasn’t about building another clever wrapper chatbot that hallucinates python scripts while you sip your flat white. This was about life after code: verification, packaging, release, configuration, and governance.

GitLab became the spine—handling everything from the single .gitlab-ci.yml file to the staging gates and agent session history where judges can peer into the machine’s little mechanical soul.

And Google Cloud? That was the missing proof. A green pipeline on a local laptop is just a comforting delusion. Wiring GitLab CI directly into GCP—service accounts, Cloud Build, Artifact Registry, and Cloud Run—meant the app had somewhere real to land. It stayed tiny and razor-focused on the lifecycle, because when you’re staring down an infrastructure failure, you don’t need a bloated UI; you need to know if the last thing that didn’t break can still boot up.

The Builder’s Post-Mortem

If you’re diving into the next hackathon or trying to drag a legacy enterprise kicking and screaming into supervised autonomy, keep a few hard-earned truths in mind:

  • Event-driven flows beat chat demos. Stop chatting with your agents; make them trigger on structural events like pipeline passes and leave an audit trail that won’t disappear when the Slack tab closes.
  • “Green” is a polite fiction. A green pipeline does not mean “safe to promote” until you have physically verified that the previous artefact can still answer the phone.
  • Keep a human on the tripwire. Agents and CI can handle the heavy lifting in the middle, but a human still owns the final, glorious privilege of pushing the button.

Undo’s rule is simple, brutal, and effective: Don’t treat the next package as ready until the last one still boots—and don’t promote until a human says so.

A dimly lit, dilapidated room with overgrown plants and moss. In the center, a server rack displays the message 'ROLLBACK_OK'. The room features broken windows allowing rain and mist to enter, with an old sofa and table in the corner.

What 2,500 Hackathon Entries Taught Me About Spines, Sky Maps, and Agentic Leases

The verdict for the WebMCP Challenge is out: 7,000 registered souls, roughly 2,500 submissions from over a hundred countries, all hammered together across ten caffeine-soaked days.

I don’t envy the judging panel. Staring at 2,500 competing agentic architectures is the sort of cognitive punishment usually reserved for low-level tech purgatory.

My entry, DealTable, did not place in the top ten. And genuinely? That is completely fine. It was feral, slightly unhinged to build, and I’m thoroughly glad it exists.

The Pitch for the Fast-Scrolling Class

DealTable is a negotiation clean-room for the coming agentic web.

You hand it a mandate: your target, your walkaway threshold, and precisely how far you’re prepared to bend before dignity gives up. A counterparty pushes back in the chat. ChatGPT sits in the corner, acting as your advisor by calling typed WebMCP tools directly on the live page—not by hallucinating UI elements or squinting at raw DOM trees like an exhausted temp at 3:00 AM.

The moment a deal crosses your walkaway line, Human-in-the-Loop trips the emergency brake. Everything worth dragging a counterparty into court over gets stamped into an audit trail.

I built it because the real dystopia lurking in our automated future isn’t a gleaming chrome skull crushing a ribcage; it’s an overly polite corporate agent cheerfully agreeing to contractual terms that quietly turn your mortgage into a multi-generational performance art piece.

I just wanted a room where the human still has a hand firmly on the shock collar.

A man with a beard and glasses sits at a wooden table holding a cup, next to an old computer displaying green text on the screen. The setting is dimly lit with technological equipment in the background.

The Gallery as a Free Masterclass

After the shutters came down, I went prowling through the project gallery. Some of the builds were legitimately intimidating: spatial editors, shared notebooks, medical viewers, and collaborative wedding seating charts where human and model stare down the same live artifact.

That is the high-water mark now.

Mandate taking the crown made absolute sense. They were playing in the exact same theoretical pool as DealTable—authority, boundaries, delegation, and who gets to touch what—but they took the mechanics a leap further. Instead of just slapping an apologetic “Click Approve to continue” prompt onto a static API list, the tool surface itself mutates dynamically as permissions are granted or stripped away. Authority expires, the tools evaporate.

That wasn’t just a win; it was an architectural lecture I sat in the front row for, taking furious notes.

My internal post-mortem was ruthlessly mundane:

  • The Pitch Video: Too many Scots-accented ums, not enough surgical ruthlessness. Lead in lockstep: Problem → Mandate → Tool Call → Human-in-the-Loop → Closure.
  • The Narrative: Lead with the visceral bleeding problem, not the plumbing. WebMCP is the pipeline; nobody falls in love with the pipe.
  • The Execution: V1 ran a rule-based counterparty so I could focus entirely on the advisor bridge. Honest engineering, but in a hackathon arena, the theatre needs to be agent-native from lobby to exit door.
A person leans against a stone pillar in a dimly lit, futuristic environment, surrounded by glowing holographic displays of celestial navigation, a scroll, a dungeon layout, and an adaptive floor plan. A sign on the wall includes a humorous quote about coffee.

Five Winners I’d Actually Boot Up (And One I Suspect Solved a Judge’s Mortgage Dilemma)

Two thousand five hundred entries generate an extraordinary amount of speculative noise. These are the five that sliced clean through it.

1. MASIL — The One That Actually Matters

Reconnecting Korean elders to creative life.

An agent suggests classic calligraphy references and surfaces Janggi moves via WebMCP, but leaves the brushwork and the completion of the game entirely to the elder.

This is the project I’d back in a noisy pub debate. Not because it had the most complex stack, but because the human stakes are palpable. Elders isolated at home, their creative universe contracting to the size of a glowing piece of glass. The agent isn’t stepping in to replace human volition; it acts as an errand boy to fetch the pieces of a world they can no longer easily walk out to, handing them back as tools for their own craft.

That is WebMCP with a moral compass: typed tools, a shared page, and basic human dignity preserved intact. It makes you feel slightly guilty that your own entry was built around defending freelance day rates.

2. Roque Nights — Built by Someone Who Looks at the Sky

A stargazing planner from an engineer at the Gran Telescopio Canarias.

The agent crunches seeing conditions, isolates the darkest nights, figures out what’s actually worth observing, and slews the interactive sky map into alignment while you sit back and watch.

Unreservedly brilliant. Zero slide-deck jargon about “agentic paradigm synergy.” Just an engineer who genuinely cares about photons, an automated assistant doing the tedious atmospheric arithmetic, and a shared visual map both are looking at. If you’ve ever stood in a freezing field at midnight wondering if it was worth hauling 20 kilograms of glass out of the boot of your car, you immediately understand why this exists.

3. Observatory — Unfortunate Name, Impeccable Dungeon Energy

A procedural map-maker for tabletop designers and weary dungeon masters.

Type “a lonely crumbled watchtower in a choking swamp,” and the system doesn’t spit out a hallucinatory jpeg with six fingers. It generates structured, fully editable vector terrain, rooms, and labels through WebMCP.

If teenage me had been handed this, I’d have treated the monitor like a sacred religious relic. The beauty isn’t “the machine drew a landscape.” The triumph is that it created a manipulable environment that you and your gaming group can spend four hours arguing over. Rename it so it doesn’t sound like a venture-backed SaaS platform from 2013, and ship it.

4. Alza — The Antidote to Cynicism

A floor-plan editor that surrenders its tools directly to the model.

Feed it a single snapshot and one verified dimension; it traces walls, extracts spatial dimensions, and constructs a navigable 3D model you can stroll through in your browser.

This is the demo you pull out when a non-technical sceptic asks why everyone is screaming about agentic standards. It’s structure over pixels. The agent handles the tedious measurement math; the human remains the architect of the living space.

It has a grounded utility that makes DealTable feel like a late-night corporate philosophy seminar—which, let’s be honest, it essentially was.

5. ArchMorph — And the “Someone on the Jury is Renovating” Hypothesis

Fifty-seven distinct architectural tools, dynamic floor plans, and layout validation.

A staggering build. It’s also remarkably close in theme to Alza. When two distinct entries in a top-ten roster both revolve around floor plans, structural walkthroughs, and spatial clearances, one cannot help but suspect at least one member of the judging panel was mid-kitchen-extension and having an absolute nightmare with their contractor.

I’m not levelling accusations. I’m just saying if an investigative committee looked into it, I wouldn’t be surprised.

What Winning Actually Looked Like

The judges didn’t care about developers merely chanting “look, we registered seven API endpoints.”

They rewarded a single, shared artifact on screen—a celestial chart, a battle map, an architectural canvas, a digital calligraphy desk—and a division of labour where the machine acts as the grease and the human remains the spine.

DealTable v2 is already queued up on the local machine. The spine stays; the execution gets sharpened with the lessons pulled straight from the winners’ circle.

For the curious or the brave:

A massive tip of the hat to the WebMCP organisers (OpenAI and DevPost), the public builders, and my long-suffering collaborative stack that helped me figure out what needsHitl means in anger (including that brief, glorious window where DealTable accidentally attempted to negotiate commercial real estate at $undefined/sqft).

Huge congratulations to the winners—particularly Mandate. You set the bar.

The Day I Outsourced My Backbone to an MCP Bridge

There is a distinct, uniquely contemporary brand of nausea reserved for corporate negotiation.

It usually begins on a Tuesday afternoon. You are sitting in the grey light of a laptop screen, staring at an email from an enterprise client whose signature line is longer than the Magna Carta. They want a thirty percent discount on your day rate. They are also proposing a payment schedule calibrated to conclude somewhere near the heat death of the universe.

Traditionally, you do what any self-respecting contractor does: you stare into the middle distance, do feverish mental arithmetic to calculate if you can survive on dry pasta until November, and draft a reply dripping with performative corporate politeness. You write, “Thanks for reaching out! Happy to find a middle ground,” while your spleen violently retracts into your ribcage.

The horror isn’t the money. The horror is the slow, wet erosion of your dignity in an unmonitored thread with no audit trail.

So, when the WebMCP hackathon opened with Devpost, I didn’t see an emerging browser protocol. I saw a containment unit. I decided to build DealTable: a clean, sterile room where synthetic entities could barter for scraps of my mortal labour, strictly supervised by a digital shock collar.

The Mathematics of Keeping Your Spine

In modern software architecture, people are terrified of artificial intelligence turning rogue, seizing missile silos, and sterilising the biosphere.

Personally, I am far more terrified of an AI agent negotiating a vendor contract on my behalf and cheerfully agreeing to a 90-day Net payment term because its sentiment-analysis model mistook predatory procurement tactics for “a collaborative synergy opportunity.”

To prevent the machine from liquidating my mortgage in the name of algorithmic politeness, DealTable relies on an ancient, barbaric concept: The Mandate.

Before you let the silicon speak, you set your floor. The target price. The concession tolerance. And, crucially, the walkaway limit ($w$).

The governance model is brutal:

$$\operatorname{apply}(p) = \begin{cases} \text{execute immediately},
& \text{if } p \ge w \\ \text{require human approval}, & \text{if } p < w \end{cases}$$

If the counterparty proposes an offer above your floor, the machines trade pleasantries and execute. The moment an offer dips even a fraction of a penny below your survivable threshold, execution freezes. The machine stops dead. It turns its digital head, looks you dead in the eyes, and demands an explicit, verified click: Approve or Reject.

Agents may hallucinate poetry, optimize supply chains, or pretend they have souls. But they cannot cross the floor. The human still owns the boundary.

Inside the Terrarium: How DealTable Operates

I built DealTable on a lean, serverless spine: Vite, React, and Tailwind, hosted on Vercel with local session state. No heavyweight databases or enterprise middleware—just pure client-side orchestration so hackathon judges could witness algorithmic bartering without signing their lives away to an authentication provider.

The secret sauce isn’t prompt engineering. It’s the Model Context Protocol (MCP).

Instead of treating an LLM like an omniscient wizard squinting at screenshots through brittle DOM automation, DealTable turns the browser window into a strictly typed operating theater. The page registers seven dedicated WebMCP tools:

  • get_deal_state: Ingests the current mandate, active asking price, conversation history, and live metrics.
  • set_mandate: Reconfigures the operational boundaries when market conditions sour.
  • parse_opening_offer: Strips incoming corporate jargon down to cold, quantifiable digits.
  • propose_offer & concede: Executes calculated counter-punches within authorized bounds.
  • hold_firm: An programmatic, polite equivalent of a flat refusal.
  • accept_term: Executes the closing sequence behind a cryptographic gate.

When ChatGPT acts as the advisor, it isn’t guessing. It is pulling live, validated state from the page and calling specific, sandboxed functions. When an action threatens the walkaway limit, the tool returns needsHitl.

A high-contrast Human-in-the-Loop modal slams over the viewport. The underlying JavaScript promise refuses to resolve until an actual warm-blooded creature clicks a button. The advisor is held in suspended animation, trapped in execution limbo while the human decides if the insult is tolerable.

Blood on the Terminal

Building autonomous negotiation arenas sounds pristine until you actually wire the pipes and watch the plumbing back up.

First came the bureaucratic indignities. Vercel threw an existential fit because a project repository contained capital letters and whitespace. A rogue hash mark sitting inside an npm run build command quietly sabotaged the Vite bundle like a loose bolt dropped into an aircraft turbine on deadline day.

Then came the existential bugs. During early test runs, our HITL modal inadvertently ingested an unparsed response object instead of the clean pending price structure. The result? A triumphant, unhinged bot proudly locking in a commercial lease for precisely $undefined/sqft. A dystopian victory for zero-cost real estate, perhaps, but tricky to defend in an audit.

Worse, demoing the workflow in real time created a bizarre psychological standoff. Triggering HITL via an active WebMCP tool call paused ChatGPT indefinitely, waiting for a human click on the host page. On a three-minute video recording, an AI staring wordlessly into space looks less like cutting-edge governance and more like catastrophic system failure. We restructured the demo rail to demonstrate deterministic tool execution in parallel with manual override triggers—making the safety rails obvious without subjecting the judges to awkward digital silence.

What the Silence Taught Me

We emerged from the hackathon with a working URL, a live audit export, and a few stark truths about the coming synthetic economy:

  1. Typed tools beat automated vision every single time. Letting an LLM scrape a webpage to make financial decisions is digital negligence. Forcing it to call strictly typed, validated tools that mutate isolated state is the only way to retain sanity.
  2. Never give one bot two jobs. In our early drafts, a single LLM tried to play both the cutthroat vendor and the impartial advisor. It rapidly degenerated into a schizophrenic pantomime where the model essentially negotiated with its own hallucinations. You must separate the actors: the principal sets the mandate, the counterparty pushes their agenda, and the external advisor sits outside the transaction.
  3. The safety switch cannot hide in a sub-menu. If human oversight is buried three clicks deep or masked behind opaque JSON logs, it doesn’t exist. HITL belongs front and centre—a flashing, unavoidable perimeter wire.

The Road to the Silicon Souk

The prototype works, but the future is significantly weirder.

Next comes swapping out our deterministic, rule-based adversary for a fully conditioned LLM adversary—one capable of simulating specific, predatory procurement personas (the Passive-Aggressive Startup Founder, the Enterprise Bureaucrat with Infinite Runway, the Venture-Backed Lowballer). After that, automatic parsing of 80-page commercial PDF contracts, stripping away the legalese to find the hidden clauses that usually bite you six months later.

Ultimately, we are barrelling toward an internet where autonomous agents will spend their days aggressively bartering with other autonomous agents over micro-transactions, service level agreements, and server runtime fees.

If we don’t build deterministic floors into the code now, we will wake up in a decade to discover our synthetic representatives have cheerfully traded away our rights, our margins, and our weekends—all to achieve a 98% polite closure metric.

I’d rather keep the walkaway limit in React state, thanks. At least when the world ends, my console will log the exact price at which I refused to sell out.

#Devpost #DealTable #WebMCP #Vercel #netlify