How it works · the ODAI Helix™
A loop closes. A helix climbs.
Boyd’s loop returns to where it started; nothing carries forward but the next observation. The ODAI Helix climbs: each turn draws on the record of decisions, actions and measured effects, so the next turn starts one level up. Speed governed by judgment.
People the artNumbers the science
Four phases, every deal
Orient. Deliberate. Act. Iterate.
The why before the how
The frame, the five knowns, the environment in every language it speaks. Diligence that asks what you must know before you sign.
The deal argued
Every party understood on its own terms. The assumptions checked, the biases named, the pre-mortem run. The investment committee paper written from the record.
The hand-off
Sign, close, integrate. Your banks, lawyers and traders execute; ODAI shapes and tests what they execute, and audits what happened.
One level up
The thesis on watch. Effects measured against intent. Lessons fed to the next turn, and to the next deal.
The dynamo
Everyone is buying the machine. The gain comes when the room is rebuilt around it.
Electric power reached American factories in the 1880s. It took decades to show up in the productivity figures, and when it did, the gain came not from the motors but from the factories redesigned around them.
The deal room is at the shaft-and-belt stage of AI. The machines are in; the room is unchanged: the same advisers in the same rooms, the same memo, the same gut at the end. ODAI is the redesign.
Paul David, “The Dynamo and the Computer,” American Economic Review, 1990.
A staff the size of the deal
As many agents as the decision needs. Your professionals spend their judgment where it matters.
Most of the work in a deal is tame: it can be stated before it is solved, and it is right or wrong. ODAI’s agents absorb it. Your professionals spend their judgment on the wicked part: the one not understood until a solution is tried.
A person on the team sets the job: a minister’s record, a port’s traffic, a lender’s covenants.
The agent gathers what bears on the job and works it step by step, at machine speed.
Back and forth with the person who set it, until the work is ready to join the rest.
Findings with sources attached. The person decides; no agent does.
They serve any function, or a team formed around one question. Heuristka’s analysts work alongside yours, so you get the most from the system, and the system learns your firm, not the other way round. When a deal calls for rehearsal, ODAI runs every party against each other on demand.
War room to deal room
The method, translated. Line by line.
Every method on the left was built to decide under uncertainty with lives at stake. Every one on the right is the same method, doing the same job, in a deal.
| In the war room | What it does | In the deal room |
|---|---|---|
| Commander’s intent | The purpose stated before the plan, so every part can adapt without losing the point | The why before the how. The mandate refined between the board and the deal team. |
| Preparation of the environment | Political, economic, social, legal, information and infrastructure factors read before the first move | Preparation of the deal environment. Regulators, publics, unions, ministries, in every geography. |
| Center of gravity | The source of the other side’s strength, and your own | What makes the target worth buying, and what must be protected through integration. |
| Critical information requirements | What the commander must know before deciding | What you must know before you sign. The diligence that asks, not the diligence that lists. |
| Branches and sequels | Plans for the ways the plan may go, and what follows each | Scenarios, earn-outs, options, the walk-away. |
| The red team | People tasked to think as the adversary and break the plan | The deal argued against by people paid to break it. |
| Leadership analysis at a distance | How the other leader thinks, from open sources | The counterpart on its own terms, and every party at the table. |
| Indicators and warning | The signs that would change the assessment, named in advance and watched | The thesis on watch after closing. |
| The culminating point | Where further advance costs more than it gains | The walk-away, known before the room gets warm. |
| Measures of effectiveness | Did the action change what it was meant to change | Value realized, not deals closed. |
| The after-action review | What was intended, what happened, why, candid by design | The decision record. |
| Estimative standards | Every judgment carries its confidence and its sources | Every claim in the investment committee paper with its confidence stated. |
| Echelon | The same problem seen at each level of command | Seniority and function. The analyst, the deal head and the board, each at their altitude, from one record. |
Nine functions
Judgment augmentation, not automation.
Context by seniority01
One record, read at each altitude.
The error catalog02
More than seventy named ways judgment fails, checked against the deal as it forms.
Structured challenge03
The key assumptions checked, the competing hypotheses weighed, the deception looked for.
The counterpart, on its own terms04
Every party understood as it understands itself.
Cross-functional integration05
Legal, financial, political and operational lines read across each other, where the biggest seams in a deal sit.
The mandate refined06
Up and down, between the board and the deal team.
Testing the actual deal07
The plan tried against every party before it is real.
Training in the system08
Judgment taught while the work is done, for the analysts whose apprenticeship AI is removing.
Education in the market09
The country, the sector, the counterpart, before the first meeting.
The techniques are drawn from the structured analytic families intelligence analysts use: diagnostic (key assumptions, competing hypotheses, deception detection), reframing (red-hat analysis, structured analogies, the pre-mortem), foresight (scenarios, indicators, high-impact low-probability analysis) and decision support. More than a thousand in the catalog, named here by example.
After closing
The deal is not the decision. The decision is what the deal buys.
The culture the plan will meet, the people it must keep, the clause that caused the last strike.
The indicators that would change the case, named at signing and watched after it.
Correlated exposure read across holdings, not asset by asset.
Run as one strategy, with the record carrying forward what each decision taught.
One engine, any enterprise
Built for decisions measured in billions, across borders and over years.
Entering an emerging market: the ministry, the local partner, the public, the succession.
Moving a supply chain: the workforce it leaves, the one it finds, the regulator in between.
Buying across a border: two regulators, two cultures, one integration.
Bidding for a concession: the government that signs, and the one that follows it.
Sovereign decision-making
The platform is ours. The decision is yours.
- Yours to keep. The outputs, the record, and what your instance learns. Heuristka keeps the engine, the method and the catalogs.
- Where your data lives. Alongside the systems you already use, or fully offline in a sealed room when the decision needs one.
- Your firm’s lessons stay in your firm. ODAI learns from every turn; the turns stay yours.
- It does not value, audit, opine, structure or finance. Your banks, lawyers and auditors do that, and ODAI reads across their work.
- It does not decide. Options with their confidence stated, never verdicts. You decide, and you own the outcome.
- Open sources and lawful methods, applied to the deal. Intelligence method, not espionage.
Uncover what is hidden.
Augment judgment.
Sovereign decision-making.
Briefings are by invitation.
We hold a limited number each quarter: forty minutes, in person in New York or Washington, or by video. The method, the engine room and the record, run on a public deal like the ones you make.
Request an invitation