ODAI · Optimized Decision-making AI
Numbers are numbers. People are a surprise.
Your firm has modeled every number it can. ODAI brings the same rigor to the people: your own, the other side’s, and everyone your decision touches.
Watch the film 4 minQuality of judgment, at machine speed.
People the artNumbers the science
The film
A country rose against a resort. The numbers never saw it coming.
The film can’t play in this browser. Ask us for a copy.
Information → Process → Judgment
A new category of AI. For judgment.
It surfaces what you hadn’t thought of, and works with your professionals until the decision is better founded than your traditional methods could make it.
What you already have
Generative AI writes. Predictive AI forecasts. Analytical AI classifies. All of it works on information, and your firm already owns plenty.
What you are buying
The AI entering the deal room makes the work faster: the data room read in minutes, the memo drafted in seconds. It measures itself in time saved.
What ODAI does
It tests the frame, the assumptions, the biases and the link from action to effect, with your experts as the filter. It measures itself by what the decision bought.
Your quant models
price the futures you imagined. ODAI finds the ones you didn’t.
Your advisers and institutes
send a report from outside the decision. ODAI works inside it, with your experts, at every turn.
Your AI tools
work on information and process. ODAI works on judgment.
Not a prediction machine. Not a content generator. Not a valuation model. Not a replacement for your investment committee.
Faster is not a direction
Every decision runs through one corner.
John Boyd, the fighter pilot who gave the world the decision loop, drew it as a triangle with orientation at the center. Observe, decide and act all pass through it. Orient to the wrong deal, and speed only gets you there sooner.
From loop to helix
A loop closes. A helix climbs.
Drawn as a loop, each cycle returns to where it began: the same advisers, the same memo, the same gut at the end.
The ODAI Helix™
Speed governed by judgment.
Each turn draws on the record of decisions, actions and measured effects, so the next deal starts one level up. And at every phase the triangle keeps turning: orientation runs through the whole deal, from diligence to integration, not once at the start.
Your bankers answer whether you can do the deal.
Counsel answers whether you may.
Your board answers whether you should, and what it will buy. That answer is the one most often given on instinct. ODAI gives it the evidence the first two already have.
Speed still wins deals: first to the asset, inside the other side’s timetable. But speed is a quantity, and the deal is won or lost on a quality: whether you understood the deal you were in. Be first to the deal. Be right about it.
Orientation produces the one output no deal team is paid to want: a reason to adapt.
Tame problems and wicked ones
The numbers have a mathematical answer. The people across the table don’t.
The regulator weighing your deal. The chief executive whose company you are buying. The founder in Palo Alto asking for a billion. Planners call these wicked problems, and the rest tame. Most of the work in a deal is tame. The decisions that lose deals are wicked.
Stated before it is solved
- Right or wrong, and testable
- The cash-flow model, the tax structure, the covenant test, the integration timeline
- Even a trillion-dollar portfolio’s value at risk has a mathematical answer
To the machine and its agents, at machine speed.
Not understood until a solution is tried
- Better or worse, never final
- Whether to do the deal at all, what it buys, and what it costs elsewhere
- The regulator, the seller, the founder, the board, the union, the public: everything that turns on people
To your professionals, with ODAI testing the frame.
Finance has modeled everything that can be counted. Monte Carlo itself came out of Los Alamos. But there is no equation for why a board will sell, why a regulator will stall, or why a country will rise against a resort. That is the surprise, and it is where deals are lost. ODAI puts the two side by side, and the checking runs in every direction: the machine checks your professionals for the bias and the stranded known; your professionals check the machine for the outlier and the false precision; and the record lets people check each other without the politics.
Your models are tame-problem machines, and superb ones. The losses on the record were wicked problems priced as tame.
The engine room
It remembers more than any deal team can, and prompts your professionals at every turn.
Any data source plugs in: the data room, your own research, the market feeds, the local press. ODAI thinks across all of it, and puts what matters in front of the person who needs it, when they need it.
What it holdsVast catalogs of data and curated knowledge
What it doesApplied analytical rigor, with your experts as the filter
- Surfaces what you hadn’t thought of: the unknown unknowns, the false knowns, the knowns stranded in another department or another language.
- Names the bias in the room while it is forming, before the investment committee votes.
- Applies the technique the moment calls for, from the key-assumptions check to the pre-mortem.
- Brings up the past case that resembles this one, with what is relevant in it marked, and what is not.
- Reads every party on its own terms, and war-games them on demand.
- Delivers options with their confidence stated, never verdicts. Your professionals decide.
- Learns from every turn, in a record that is yours.
The question, the thesis, the draft memo, the number the seller gave you.
The false known in the seller’s memo. The bias in the frame. The regulator’s draft nobody read. The deal that failed like this one before.
They accept, reject or dig deeper. The machine argues back, and nobody defers to it.
What was decided, why, on what sources, with what confidence. The next turn starts there.
Not experts or the machine. The machine in the service of your experts. Any discipline the decision needs, law, quantitative finance, psychology, a region, a regulator: your specialists, Heuristka’s analysts, or both. One method.
The error catalog
The mistakes that lose deals have names.
Every deal team has met these. ODAI’s catalog names more than seventy, and grows, and it names them as they form, in the room, while there is still time to act.
Deal feverEscalation of commitment
Months in, three advisers paid, the announcement drafted. Walking away now feels like losing.
Holds the walk-away point you set before the room got warm, and shows when you have passed it.
The winner’s curseOverbidding
The highest bid wins, and the highest bid is usually the one that most overestimated the asset.
Prices what the other bidders know, and what none of you does.
The seller’s numberAnchoring
The memorandum sets the price, and every model in the building starts from it.
Rebuilds the case from sources the seller did not write.
Diligence that agreesConfirmation bias
The team finds what the thesis needs, and files the rest.
Runs the competing explanations side by side: what would we be seeing if we were wrong?
The chairman’s viewDeference to rank
The most senior opinion in the room becomes the room’s opinion.
Weighs every objection on its evidence, whoever raised it.
Synergies on paperThe planning fallacy
Integration costs half as much and takes half as long, in the model.
Tests the plan against the deals that promised the same thing.
Their shoes, our feetMirror-imaging
Assuming the other side wants what we would want.
Builds the counterpart on its own terms: control, legacy, a ministry’s timetable.
Convenient disbeliefThe inconvenient known
The risk everyone has seen and nobody wants to price, because the deal is wanted.
Flags it by name before the vote.
And the techniques that catch them
More than a thousand, built to decide under uncertainty with lives at stake. A few, by example.
- The key-assumptions checkEvery “we know” in the memo listed, sourced and tested.
- Competing hypothesesThe seller’s story weighed against the others the evidence also fits.
- The pre-mortemIt is two years on and the deal has failed. Why?
- The red teamThe deal argued against by people paid to break it, before the people paid to sell it.
- Scenarios and backcastingThe futures named, and the path back from the one you want.
- Indicators and warningThe signs that would change the thesis, named at signing and watched after it.
See around the corner
The windshield, not the rear-view mirror.
The deal is rarely lost on what you knew. It is lost on what you didn’t know you didn’t know, on what you knew that wasn’t so, and on what someone in the building knew and never said. ODAI goes looking for all three, before you sign. In most of the deals that failed, the fact that sank them was visible before signing to anyone who looked. The failure was in the decision, not the data.
| What you know | The deal killer | The opportunity |
|---|---|---|
| Known, not weighed | A $160 billion merger that died in 2016 on a Treasury rule change everyone had seen coming. | Already priced. The work is to make sure it was weighed. |
| Known unknowns | Deals abandoned in 2022 and 2023 under regulatory opposition that was known from the first day, each at a cost of about a billion dollars. | Scenarios, earn-outs and hedges priced better than the other side’s. |
| Unknown unknowns, imaginable | A deposit run that took about $42 billion out of a bank in a single day in 2023. | Positions ready for futures others called unthinkable. The pre-mortem finds these. |
| Unknown unknowns, unimaginable | The true black swan. | Speed of re-orientation. The team forms in hours, with the record behind it. |
| False knowns | €1.9 billion of cash that the audited accounts showed and that did not exist. Top ratings on mortgage bonds, market-wide. | The walk-away, the short, or the trade of the decade, for those who checked. |
| Stranded by seam | Breaches a buyer learned of after signing, and $350 million came off the price. | Knowledge in another department, or another adviser’s file, read across. |
| Stranded by rank | The junior who saw it first, and was not heard. | The warning weighed on its evidence. |
| Stranded by language | The local press, the regulator’s draft and the minister’s speech that nobody on the team could read. | Information the other bidders can’t reach. |
| Stranded by inconvenience | A litigation risk that was known before the deal was signed. Up to $10.9 billion to settle. | Convenient disbelief flagged before the investment committee votes. |
| Unknown knowns | The lessons from the last time in that country, held by someone who has left. | Institutional memory, surfaced. |
Every arbitrage is someone else’s stranded known.
Every kind of known cuts both ways: the deal killer on one side, the opportunity on the other. And the deals that worked share a habit worth copying: they knew what to protect, not only what to fear. For an index, the world’s politics is mostly noise. For a single asset, it decides everything. ODAI works at the level of the asset and the decision.
Cases: Pfizer–Allergan; Nvidia–Arm and Adobe–Figma; Silicon Valley Bank; Wirecard and the mortgage ratings of 2007; Verizon–Yahoo; Bayer–Monsanto. Figures as publicly reported.
What it looks like
One vault. Every party, every known, one record.
An invented port-concession bid, three weeks in. Everything on this screen is something a deal team usually finds out too late.
The five knowns
The parties
Deliberation
The record
Models and people: each fails where the other sees.
The models behind the ratings gave subprime mortgage bonds top grades. Michael Burry read the prospectuses behind them, saw the collapse built into the loans, and bet against them. His own investors demanded their money back.
At Credit Suisse, the limit system showed Archegos breaching its limits for months. The business granted temporary increases instead, one of them for $900 million. The bank lost more than $5 billion.
Zillow’s pricing algorithm kept paying up for homes as the market cooled, because Zillow had tuned it to bid harder while rivals pulled back. Writedowns of more than half a billion dollars; the business shut; a quarter of the staff gone.
Goldman Sachs’s mortgage desk lost money ten days running and its risk models said something was wrong. About fifteen of its senior risk and trading people went through every position in some three hours and cut the risk, a year before the crisis broke.
Neither alone is enough. It was the deliberation that joined them.
Your professionals · catch your errors
The person who sees it first is rarely the person in charge.
The junior analyst on the spreadsheet. The local hire who reads the press. The engineer on the integration team. They see it first, and the warning dies on its way up.
In 2014 a research engineer barely out of Stanford found that Theranos’s tests were failing and its results were being doctored. He told the chief executive. She did nothing; the president belittled him; his own grandfather, a director, took the company’s side. The board and the investors missed what the junior saw. He was right.
Intent travels down the chain. The objection never travels up. ODAI weighs the warning on its evidence, not on the rank of whoever raised it, because the machine has no career to protect.
- The gut check, with evidence. Every instinct in the room tested against the five knowns and the catalog of errors before it reaches the investment committee.
- Aligned on the why. The mandate refined between the board and the deal team before the how is argued, so the politics come out of the room.
- Judgment for the next generation. AI is removing the junior analyst’s apprenticeship. ODAI teaches judgment while the work is done.
The other side · price theirs
The counterpart, on its own terms.
What they value is rarely price alone: control, legacy, a home-government interest, a lender’s covenant. How they think, and what their networks, lenders, shareholders and state ties make them see.
ODAI builds the counterpart from open sources: the owners and the board, how they decide, what their careers have taught them, what they have walked away from before. Then it does the same for every other party at the table: the lenders, the rival bidders, the regulator, the workforce. When the deal calls for it, it runs them against each other, so your professionals walk into the real negotiation having already sat through the one that went wrong.
- Every party, understood. What each wants most, and what each cannot accept.
- The walk-away found early. The point past which the deal stops being worth its price, known before the room gets warm.
Everyone your decision touches · move everyone else
Governments, regulators, workforces, publics. In their own languages.
In the spring of 2026 a luxury resort planned for Sazan Island and the Vjosa–Narta wetland on Albania’s coast, backed by an American investor, brought tens of thousands into the streets of Tirana night after night for weeks. They called it the Flamingo Revolution, after the birds that breed in the lagoon.
Almost everything that produced it was visible before a dollar moved. The protected-areas law had been changed in 2024 to make the project possible, and the environmental groups had objected at the time. The same investor had withdrawn from a project in Belgrade only months earlier, after protests there. The local press, the ecologists and the opposition had all said what they would do.
Don’t be an ostrich. Be a flamingo.
ODAI reads the local press, the regulator’s draft and the minister’s speech in the language they were written in, a capability that once belonged to governments and the largest firms. It anticipates how each party will react, and then helps you make the case to each of them, in their own interest and their own words.
Facts as publicly reported.
- Regulation before it arrives. The draft, the consultation, the minister’s signal.
- Social license. The community, the workforce and the public, each with its own red lines.
- Then make the case. Each party shown its own interest in the outcome. People move further for what they can verify than for pressure or payment.
The decision record
The mirror makes the windshield.
The record of the last decision is what lets you see further on the next. ODAI keeps it as a first-class object, not an afterthought, and the investment committee paper comes out of it already red-teamed.
It also protects the person who made the call. Your professionals take priced market risk every day; what they avoid is the unpriced career risk of a bold call. A decision with its reasoning, sources and dissent on the record is a bold call that can be defended.
After closing, the thesis stays on watch. The portfolio is read as one book, so one shock is seen across a dozen holdings at once, and a thirty-year horizon is run as one strategy rather than thirty one-year ones.
Measured in value realized, not deals closed.
Sovereign decision-making
Your platform. Your decision. Everything it produces is yours.
ODAI is the platform; the judgment, the outputs, the record and what your instance learns all belong to you. Heuristka keeps the engine and the method.
An operating system for your firm’s judgment. It runs where your data lives, alongside the systems you already use, and it is built to run fully offline in a sealed room when the decision is worth that much. Nobody learns a new system from scratch: Heuristka’s analysts work alongside your team, and ODAI learns your firm, not the other way round.
Bespoke and agnostic. Built around the decision in front of you, with as many agents as it needs. Agnostic twice over: it works with any system and any model, and it has no stake in the answer.
The decisions worth the most are made where cloud AI can’t go. ODAI goes in the room.
One engine for any enterprise: a sovereign fund entering a new market, a manufacturer moving a supply chain, a bank buying across a border, an energy company bidding for a concession.
War room to deal room · military-grade method
Until now, the answer was to hire the intelligence world. ODAI puts it in the box.
The largest firms know what a cross-border acquisition or a billion-dollar investment really turns on. In a world this connected, a deal can be sunk by a ministry, a strike, a sanction, a currency or a storm that nobody priced. Their answer has been to hire the national-security world: former intelligence officers at every level, and former generals. What they produce is a report, which may or may not reach the people who decide.
ODAI gives you the staff instead of the report: geostrategists, financial analysts, quants, meteorologists, land economists, whatever the problem needs, in the box and at your team’s elbow. It draws on two kinds of knowledge. The established: facts, data and settled science, uncontested. And the contested: what people will do, what a government intends, what a market believes. The contested is where ODAI earns its keep, with the cognitive and strategic analytic techniques intelligence analysts use. Anything your firm already runs, from its models to its legal AI, plugs in.
Not every fund can hire a team of former intelligence officers. Every fund can have the method.
Finance has borrowed from that world before. Monte Carlo simulation, the method behind most modern risk models, runs thousands of random trials to estimate the odds of an outcome. Stanislaw Ulam and John von Neumann devised it at Los Alamos in the 1940s for the nuclear weapons program, and named it after the casino. Scenario planning began with Herman Kahn at RAND, the U.S. Air Force’s think tank, after the war. Pierre Wack took it to Shell around 1970, and Shell met the 1973 oil shock better prepared than its rivals.
ODAI brings the rest of that toolkit: the red team, the pre-mortem, leadership analysis at a distance, indicators and warning, the after-action review. Methods built over decades to decide under uncertainty with lives at stake, and almost none of them yet in a deal room.
Where ODAI began. In March 2026, in an independently adjudicated NATO–Russia crisis war game run by Black Horizon, one of Heuristka’s founders ran his team’s information strategy with prompts and structures he had built and tested across several of the leading AI models. The game was won without a shot fired. The instruments on that board were as often economic as military: markets, supply chains, regulation, an industrial lobby. Writing it up afterward, he saw that the method was a business.
The published assessment: No Shots Fired, Military Innovation Lab, 10 March 2026.
Who gains
The beneficiaries first.
A sovereign fund’s owners are a nation’s future generations. A pension fund’s are the people who will retire on it. Better-founded decisions are owed to them before anyone else.
Edge, and a call that can be defended
What others missed, found first, and a record that shows why the call was right to make.
A staff the size of the deal
Every function sees every other line of the deal before it reaches the investment committee, without new headcount.
Judgment, taught while the work is done
Every catalog ODAI holds, applied to their part of the deal, and their warning heard on its evidence.
Heuristka
We have planned, fought and lost wars, and spent the years since working out why.
Heuristka, from the Greek heuriskein: to discover. Craftsmen of discovery and judgment. The founders carry the two lineages the discipline joins: the epistemic tradition of Bletchley Park and the practitioner tradition of military operations.
Bletchley Park was Britain’s secret codebreaking center in the Second World War. Its people broke Germany’s Enigma cipher with machines Alan Turing helped design, and built Colossus, one of the first programmable electronic computers. Turing went on to ask whether machines could think. It is where machines were first put to work on intelligence, and where computing and AI have some of their deepest roots.
In June 1940 a young Bletchley analyst, without rank or standing, read the pattern in German naval signals and warned that the heavy ships were moving. He was not believed, because of who he was, and the aircraft carrier HMS Glorious was lost. He spent the rest of his life writing Bletchley’s official history: what intelligence could and could not do for a decision. One of our founders was his student.
U.S. Army veteran who commanded special operations aviation units and task forces through deployments to Iraq and Afghanistan, with extensive mission experience across the Middle East and East Africa. On the Joint Staff, advised the Chairman of the Joint Chiefs on special operations and sensitive activities, led the Joint Staff’s work on a large standing counterterrorism plan, and prepared orders for approval by the Secretary of Defense and the President. Nearly a decade teaching special operations and strategy at Marine Corps University to more than 200 officers and officials from across the U.S. government and over twenty countries. M.S., Conflict Analysis and Resolution, George Mason University; M.S., Computing Technology in Education, Nova Southeastern University.
In corporate termsHe ran operating units of up to 250 people, fifteen aircraft and more than half a billion dollars of equipment, and made life-or-death calls against weather, terrain, hostile forces and equipment at its limits: decisions under pressure, with no second chance. He knows the Middle East from the ground. Then the very top of the institution: advising its chairman on the matters too sensitive for the full board, preparing him for meetings of the National Security Council, and drafting orders that went to the Secretary of Defense and the President of the United States for approval. A master’s in conflict analysis and resolution, which is negotiation by another name, and another in computing. And nearly ten years teaching more than 200 senior leaders from over twenty countries, many of whom went on to lead their services and agencies.
Creator of ODAI’s cognitive core. Cambridge doctorate under Sir Harry Hinsley, Bletchley Park codebreaker and official historian of British wartime intelligence; Visiting Fellow, Wolfson College, Cambridge. Formed in Australian, British and American institutions: directed strategic policy for the Royal Australian Air Force and advised Australia’s Parliament on defense. A decade in the U.S. war colleges, creating and directing the Defense Department’s most comprehensive graduate program in strategic deterrence and counter-proliferation. Principal strategist for counter-WMD and red team lead at U.S. Special Operations Command. Designed and led the U.S. Air Force Chief of Staff’s war game with the Indian Air Force. Testified to the U.S.-China Economic and Security Review Commission. Author of The Third Offset Strategy: A Return to Ungentlemanly Warfare and founder of the National Security Desk. A record of anticipating strategic shifts before they entered mainstream discourse: the methods al-Qaeda would use (1997), Russia’s massing for a major escalation in Ukraine against prevailing estimates (2021), and the first-mover logic of unconventional nuclear use (2022).
In corporate termsTrained by the official historian of the most consequential intelligence operation of the twentieth century, and a Visiting Fellow at Cambridge. Ran his own strategic-intelligence firm, selling assessments to paying subscribers. Built a Defense Department graduate program from a standing start in four months, on weapons of mass destruction, the highest-stakes risk there is. He raised more than $1.5 million for it and persuaded three global commands and the White House National Security Council to fund it and lend it their experts. Then chief strategist and head of the red team at a global command: the person paid to find what the plan missed before an adversary did. Author of a book that took the Pentagon’s technology slogan and turned it on its head: the way to hem China in from the west, it argued, is human networks, with special operations working behind the lines. A public record of calling the turn against the consensus: the methods al-Qaeda would use, four years before September 11; that Iraq could become a long war when the forecast was six weeks; that Russia’s army was hollow, and that it was massing for a major escalation in Ukraine in 2021, against the prevailing estimates. Sees America from inside and out, which is how most of the world’s capital sees it.
By invitation
Uncover what is hidden.
Augment judgment.
Sovereign decision-making.
The platform is ours. The decision, and everything it produces, is yours.
How does your firm make its biggest judgments? How would it make better ones?
Briefings are by invitation, and we hold a limited number each quarter: forty minutes, in person in New York or Washington, or by video.
You leave with:
- The five knowns run on a public deal like the ones you make
- The errors most likely to be in your room, by name
- The questions your investment committee isn’t yet asking
Nothing confidential is needed. The button opens an email to us with your answers filled in.