Tacitum [email protected]

Timing is everything…

…in strategy, organization, and work, particularly temporary endeavors: the adaptive projects that deliver your strategic intent in an increasingly uncertain environment. Yet the rules that run the clocks are rarely written down. Tacitum makes them explicit, then lets AI agents run your project portfolio within them. People keep the authority to decide. Agents do the work of keeping everything in step. Tacitum transforms responsibility into resilience at machine speed.

Request early access
Benefit
$0
€0
¥0
Risk
$0
€0
¥0
Outputs flowing down as inputsOutputs flowing up as inputsEscalation, skipping tiers and handled off-cycleNow
Every time a tier fires, its outputs become inputs to the practices on the tiers above and below it rhythmically. Escalations skip tiers and are handled off-cycle. Where the clocks fail to line up, decisions wait, and portfolios lose time and money, and sometimes cost lives.

Most portfolio failures are failures of timing

A decision arrives after the window it was meant for. A program waits on a project that is waiting on a budget review. Every forum keeps its own calendar, and nobody owns how those calendars fit together. When conditions require everyone to pivot, the clocks are reset manually.

The rules that would fix this already exist in your organization. They live in habits and in the heads of experienced people, which is why they break when those people are busy or gone. As AI agents take on more of the work, the gap widens, because an agent that does not know the rules acts on its own clock.

Before an agent can be trusted with a portfolio

Three things have to be true.

  • It knows the organization's actual rules: which forums decide what, on which cycle, and what triggers an escalation.
  • It acts only within authority that a named person holds, on the record.
  • Every change to a rule is proposed, approved by a person, and logged, so the system can be audited after the fact.

Few organizations meet the first condition, because the rules were never written down. That is where Tacitum starts.

How Tacitum works

  1. Assess

    A guided self-assessment captures how your portfolio is actually governed: which forums meet, how often, what triggers an escalation, and who decides. Tacit rules that sit silently in the data of your projects, rules which have never been written and which may be unknown to those ruled by them, emerge from their hiding places to form a clear picture of how you really work.

  2. Register

    Each person knows a piece of those rules, but nobody sees the whole. Tacitum assembles the pieces from the artifacts your people already keep, and the result works like a Rosetta Stone, translating the strategy your leaders state into the timing your portfolio actually keeps. Tacitum turns the assessment into a machine-readable register of rules for cadence, thresholds, and synchronization. Every rule is visible, versioned, and owned by a named person.

  3. Operate

    Tacitum is not a consulting firm. Your self-assessment is the boot sequence for a suite of agents deployed seamlessly by you on-prem or in your own cloud. An AI agent for each portfolio practice, from prioritizing to controlling, proposes and carries out work. A synchronizing core checks every action against the register. That core is deterministic code, not a language model, so the rules hold even when a model errs.

One escalation, end to end

A worked example, drawn from the self-assessment of a multinational manufacturer with names and figures changed. Each step names who acts: an agent, the synchronizing core, a person, or the gateway. The rule codes are the register's own.

  1. Controlling agent

    A forecast crosses the line

    At the biweekly progress review, the Controlling agent reads the ERP through a read-only connector: line 3's forecast is 12 percent over the approved budget. It drafts the change-log entry. It cannot act on it.

  2. Synchronizing core

    The register disposes of it

    Rule THR-21, owned by the head of investment control, absorbs variance below 10 percent and interrupts above 25. A 12 percent reading, held for one project cycle, is disposed wait: the core opens an obligation on the investment committee, due within one of its cycles.

  3. Synchronizing core

    Routed by the calendar

    Wait means the next scheduled sitting. The committee sits monthly, 14 days away, so the item joins that agenda with its status report due five working days before. Above 25 percent, the core would have called an off-cycle sitting within one project cycle.

  4. Composing and Prioritizing agents

    Options, not decisions

    The Composing agent drafts two options and flags the coupled project. The Prioritizing agent may note a re-ranking but may not enact one: no rule grants mid-year authority, so the core withholds the action.

  5. Committee chair

    One person ratifies

    The chair ratifies option A at the November sitting. Who decided, when, and under which rule are recorded and content-addressed under UOR-ADDR, so the decision can be proven later.

  6. Gateway, then two agents

    Executed, then learned

    The gateway writes the ratified baseline to the ERP and records the tool's reply. The Measuring agent resets the benefit baseline; the Learning agent files the lesson against the trigger that caused it.

Six records, one chain. The agents never held authority over the baseline; the chair did, on the record.

What the customer sees

One product, three screens: the assessment that records how you govern, the register it compiles, and the queue where people decide.

The Assess screen: a project control forum's cadence, written tolerance, escalation path, and the timing gaps found
AssessIn use today
The Register screen: timing rules with their owners, versions, and one proposal awaiting ratification
Register2027 design
The Operate screen: the committee chair's ratification queue with two drafted options and the content-addressed ledger
Operate2027 design

Example data throughout. The Assess screen is redrawn from the live instrument; Register and Operate are designs for 2027.

Who this is for

Governance is a shared function, so Tacitum is licensed per organization or portfolio rather than per seat. Four roles have a stake in it.

  • PMO or portfolio governance leader

    Establishes the rules, prepares decisions, tracks obligations, and coordinates the forums. The register is this role's instrument, and the first conversation is usually here.

  • CFO, CIO, or investment committee sponsor

    Wants investment oversight, escalation, and accountability that hold between sittings, not only at them.

  • Committee chair and investment control staff

    Receive complete decision packages before the deadline and exercise recorded authority when they decide.

  • Security, architecture, and risk teams

    Validate the deployment boundary, the permissions each agent holds, and the evidence the system leaves behind.

What is different. Portfolio platforms record a decision after people make it, and approval workflows route a document without knowing the calendar or the thresholds. Tacitum holds the timing rules as an enforceable object that agents cannot bypass. That is a design distinction. Whether it produces faster and better-prepared decisions is a question of measurement, and the measures are stated below.

Agents can propose. Only people ratify.

An AI system cannot grant itself authority over when decisions get made. Someone has to hold that authority, on the record, and answer for it. Tacitum is built around that requirement, and it is what makes an agentic portfolio governable.

Every agent action is provable, not merely logged. Each request, decision, ratification, and result receives a content address under UOR-ADDR, an open standard authored by our co-founder Maura Clark, so a customer can show afterward what each agent did and under whose authority.

What we measure

What a customer should expect to change, and how both sides will know.

  1. Decision timeliness

    Elapsed time from a trigger to an authorized decision, and the number of decision windows missed.

  2. Decision readiness

    The share of committee items that arrive with the required evidence before the deadline.

  3. Accountability

    Overdue obligations, obligations without an owner, and unresolved conflicts of authority.

  4. Control effectiveness

    Unauthorized actions prevented, legitimate actions wrongly blocked, and whether what was executed is what was approved.

  5. Operating effort

    Staff time needed to prepare decisions and to maintain the register.

  6. Adoption

    Repeated use across committee cycles, and extension to further portfolios.

Speed and control are measured together. A system that recorded authority perfectly while slowing decisions would fail at its purpose.

Built to run where the data lives

Portfolio data is among the most sensitive information an organization holds: budgets, priorities, and decisions not yet announced. Many organizations will not send it to an outside model provider.

Tacitum's agents are being built on open models served on NVIDIA accelerated computing, so they can run inside a customer's own cloud or data center. The synchronizing core is deterministic code and runs on CPUs. Inference is reserved for the agents' work: reading evidence, drafting options, writing decision packages, and filing lessons. Each of those is a model call, and validating the system means hundreds of simulated runs, each requiring thousands of agent decisions, before it touches a real portfolio.

The agents reach the systems a portfolio already runs on through the Model Context Protocol, an open standard by which a system publishes what an agent may read and do. They never call those systems directly. Every request passes through the synchronizing core, which checks it against the register, and then through a gateway that holds the credentials and makes only the approved call, inside the customer's own environment.

Where Tacitum stands

  • The engine is already built upon institutional standards that govern a multi-trillion-dollar economy of temporary endeavors, standards our co-founder John Schlichter originated and that have been adopted globally for decades.
  • The self-assessment is live and in use with a multinational manufacturer.
  • The governance core is being tested in a simulated portfolio, under an evaluation design registered in advance.
  • The agent layer is in development for 2027.
  • The research behind it is being prepared for peer-reviewed journals in information systems.

Who is behind it

  • John Schlichter John SchlichterCo-founder

    Agents can only keep rules that someone defined. John led the international team of about 800 practitioners in 35 countries that wrote the global standard for organizational project management and PPM, published by ISO and adopted worldwide. His doctoral research at Trinity College Dublin examined how portfolio management becomes a dynamic capability that adapts as conditions change. For over two decades he has worked in the portfolios of global organizations, spanning Fortune 100 companies and equally esteemed government entities. John conceived Tacitum and writes its code.

  • Maura Clark Maura ClarkCo-founder

    Running agents inside a customer’s own data center, and proving afterward what each agent did, are infrastructure problems. An executive who knows her way around complex technology organizations, Maura has spent 15 years in telecom and data center networking, most recently helping enterprises, cloud providers, and critical infrastructure operators connect AI workloads. She authored UOR-ADDR, an open standard for addressing and verifying AI agent content adopted by the UOR Foundation, and contributed to its Rust reference implementation. She holds certifications in AI infrastructure from NVIDIA and IBM.

  • Scott Darden Scott DardenCo-founder

    Timing failures cost most where organizations must act in step across borders. For over two decades, Scott has led supply chain and logistics programs across Europe, Africa, and the Middle East, including defense procurement, enterprise logistics, and client program direction at Maersk, U.S. Department of State, and U.S. Department of Defense, among others. Over the years, he has founded several startups in Dubai. He knows what it takes to make explicit and tacit rules work when the work among contractors, suppliers, and customers crosses continents and cultures. He is fluent in both English and Arabic. يتحدث اللغتين الإنجليزية والعربية بطلاقة.

Strategic Advisor

  • Benn Konsynski Benn Konsynski

    Benn Konsynski is the George S. Craft Professor of Information Systems and Operations Management at Emory University’s Goizueta Business School, and he previously held faculty appointments at Harvard Business School and the University of Arizona. His research centers on decision support systems and on information systems that coordinate work across organizational boundaries, two foundations of Tacitum’s governed agents for portfolio decisions. He has served on the boards of several technology companies, and he holds a PhD in computer science from Purdue University.

Test it on a real portfolio

We are looking for a small number of PMO and portfolio leaders who want to see how their portfolio keeps time. Write to [email protected] to take part.