A charter, not a project plan
Delegated decision rights signed at executive level, so a lane lead can ask a business unit for something instead of negotiating for it.
The window is 48 hours. Your clock runs in months.
EVOCATI is a ninety-day cyber surge operation: a temporary command structure with real authority, delivered on the PRINCIPIA platform with everything needed to run it. It closes in less time than the average enterprise currently takes to remediate one complex application.
Drawn to scale, in days, from published 2026 data
2 days 88% of vulnerabilities with a public proof of concept were exploited within 48 hours CrowdStrike, 1H 2026
80 days median from CVE publication to confirmed exploitation, down from 120 days in 2025 VulnCheck, 1H 2026
160 days average enterprise mean time to remediation for complex applications Qualys, 2026
90 days: one EVOCATI campaign
0
30
60
90
120
160
01 / The counterweight
Almost everyone selling into this moment is selling you fear. We are not.
Of 1,061 vulnerabilities attributed to AI-assisted discovery, fourteen have been confirmed exploited in the wild. That is 1.3 percent. Anthropic's Project Glasswing produced 23,019 findings, which became 126 published CVEs, of which exactly one has been confirmed exploited.
We publish that figure in our own marketing because it is the honest reading, and because it changes what you should do about it. AI has not made every vulnerability lethal. It has doubled the volume of things you must triage, from 4,268 high and critical disclosures in Q2 2025 to 8,539 in Q2 2026, while collapsing the window on the small subset that matters.
The binding constraint is not fear. It is triage accuracy and remediation tempo. That is a throughput problem, and throughput problems are solved by structure, authority and cadence rather than by alarm.
Project Glasswing, in full
23,019
findings produced by AI-assisted discovery
126
became published CVEs
1
has been confirmed exploited in the wild
1.3%
of all AI-discovered vulnerabilities exploited in the wild
A task force you can actually stand up, and a date on which it ends.
In the Roman army, the evocati were veterans recalled to serve one campaign under a named commander, then released. They were not a standing garrison. They were experienced people assembled for a specific fight with a specific end.
That is the instrument. Not an assessment that produces a report, and not a tool that adds another queue to a team that is already the bottleneck. A working organization with delegated decision rights, a measured baseline, a daily cadence, and a handover that is written before the work starts.
Authority
Tempo
A measured baseline on day five
Frozen metrics before anything changes, so the difference at day ninety is a fact rather than an opinion.
Evidence
A record built as you go
A decision log that captures every delegation, override and exception while it happens, rather than being reconstructed afterwards.
The delivery stance
Trilogyworks advises. It never commands.
Every seat in the task force is filled by your people, the Chief included. Trilogyworks sits beside the campaign rather than inside it: shaping the structure, challenging the calls, and holding the evidence discipline, while your executives keep every decision right. That is deliberate. A capability that depends on an outsider staying is not a capability you own, and a third party inside the chain of command is a question you would rather not have to answer to an examiner.
03 / The shape of a campaign
Ninety days, four sprints, five lanes, one war room.
Each sprint closes on ratified exit criteria. Branch points at day thirty and day sixty are pre-authorized, so the campaign can be shortened, extended or stopped on evidence rather than on sentiment.
Sprint IDays 0 to 10Stand up
Charter signed, lanes staffed, cadence running, baseline frozen.
Sprint IIDays 10 to 30Compress the clock
Prove the model moves a measured number, not just activity.
Sprint IIIDays 30 to 60Harden and prove
Make the uplift stick under examiner-grade scrutiny.
Sprint IVDays 60 to 90Sustain and hand off
What survives becomes business as usual, with named owners.
Lane 1 Patching accelerationLane 2 Attack surface reductionLane 3 Application securityLane 4 Vendor riskLane 5 AI track
The lane architecture is tuned to each institution during stand up. Lane contents, scoring model and exit criteria are proprietary to Trilogyworks and are delivered under license.
An operating system for the campaign, not a slide deck about one.
Doctrine
Four-document operating set
Charter, ninety-day plan, operating manual and lane runbooks, generated for your institution rather than adapted from a sample.
Working kit
Fifty-plus artifacts
Trackers, registers, intake forms, decision records, status templates and tabletop material, all ready to use on day one.
Proof
Day-ninety evidence pack
Baseline, movement, decision log and sustain handover, written for directors and examiners rather than for engineers.
Principia
The platform the campaign runs on. Named for the headquarters tent at the center of every Roman camp, where command sat and the standards were kept. PRINCIPIA generates and versions the artifact set, carries the decision log, tracks lane exit criteria against the frozen baseline, and produces the evidence pack.
Delivered as an annual subscription hosted by Trilogyworks. Every tier includes it. The advisory tiers sit on top of it rather than replacing it.
On the method
We publish the shape of the method openly: the sprint structure, the lane architecture, the governance model and what the campaign produces. We do not publish its contents. The scoring model, the lane playbooks, the exit criteria and the artifact text are the intellectual property of Trilogyworks and are licensed.
05 / Who it is for
Three people have to say yes, and each of them is asking a different question.
The CISO
Authority and evidence
You already know what is broken. What you do not have is the authority to fix it across business units on a compressed clock, the capacity to do it without stopping everything else, and the record to prove afterwards that you acted reasonably.
The CEO and board
Fiduciary defensibility
The question a board has to answer is not whether the institution is secure, because nobody can promise that. It is whether the institution acted reasonably on what it knew, at the speed the evidence warranted.
The CIO and CTO
Capacity and containment
This is not another program running through your delivery teams. It is a temporary structure with its own cadence, its own decision rights, a named capacity plan with backfill, and a date on which it ceases to exist.
Five contrasts worth being direct about.
Against a Big Four assessment
They produce a report and leave. EVOCATI produces a running organization with an end date. An assessment tells you what is wrong. A task force changes it and proves the change.
Against tooling and MSSPs
They add another queue to the team that is already the bottleneck. Tools find things faster. Nobody has yet built a tool that makes a business unit say yes.
Against staff augmentation
Bodies without an operating system produce activity. EVOCATI ships the operating system first and the people second.
Against an internal PMO
A PMO tracks. A task force decides. The difference between them is the charter, and the charter is the hardest part to manufacture from the inside.
Against the panic merchants
We put the 1.3 percent figure on our own front page. Buyers remember the firm that told them the truth about the size of the problem.
07 / Proof
Anonymized by default. Named only with written permission.
Reference profile one
A federally chartered wholesale financial institution. A five-lane surge organization governed by a four-document operating set and a template kit of more than fifty working artifacts.
54named seats
42FTE equivalent
8,500person hours
Reference profile two
A top-tier United States credit union. An enterprise-scale task force with board-level governance and a parallel AI enablement track running alongside the security lanes.
90day campaign
5lanes
4sprints
Trilogyworks does not publish breach reduction or incident avoidance figures, because no honest firm can attribute them. What we commit to are process outcomes: a baseline frozen by day five, exit criteria ratified or formally deferred at every sprint boundary, and an evidence pack at day ninety.
Start with the platform. Add authority when you need it.
Four tiers. The first is a product you can run yourself. The rest add Trilogyworks resources.
Start hereBase$60,000per year, from
PRINCIPIA subscription and the licensed artifact kit. You run the campaign yourself, with the full operating set generated for your institution.
Get startedStand-upOn requestplus subscription
Base, plus a fixed launch engagement. Trilogyworks advises through charter ratification, lane setup and the first sprint, then steps back as your war room takes over.
EnquireMost commonCampaignOn requestplus subscription
The full ninety days. Trilogyworks advises the Chief and Deputy, guides the cadence, informs the exit criteria and delivers the evidence pack.
EnquireCommandOn requestplus subscription
Campaign, plus continuity. Advisory coverage for the full engagement, plus board reporting.
EnquireFinal pricing reflects institution size, lane count and the capacity plan agreed during scoping. Every tier includes the PRINCIPIA subscription, which renews annually.
09 / Common questions
The objections we hear, answered plainly.
We already have a vulnerability management program.
You do, and it runs at business-as-usual tempo, which is correct for business as usual. This is a surge with a defined end date. The two are different instruments. If your existing program could already absorb a forty-eight-hour window, we would not be having this conversation.
Is this not just fear selling?
It is the opposite, and the front page of this site is the evidence. Only 1.3 percent of AI-discovered vulnerabilities have been exploited in the wild. We lead with that figure. The problem is throughput and triage rather than apocalypse, and the response we recommend is proportionate and time-boxed for exactly that reason.
We could run this ourselves.
Many institutions can, and the Base tier exists precisely for them. What most cannot manufacture internally is the charter, and the confidence to use it once it is signed. The Chief is always your employee. What an advisor who has done this before adds is the structure that makes delegated authority hold, and the standing to tell an executive that a lane is behind.
Why ninety days? It sounds arbitrary.
It is chosen rather than arbitrary. Short enough to hold executive attention through to the end, long enough to move a measured metric. The day thirty and day sixty branch points exist so the campaign can be shortened, extended or stopped on evidence.
What happens on day ninety-one?
The sustain block, which is a deliverable rather than an afterthought. Each lane transfers defined capabilities to named business-as-usual owners, and that handover is itself an exit criterion. A campaign that cannot hand over has not finished.
It is more than we budgeted.
Weigh it against remediating a single examination finding on a regulator's deadline, or against the cost of one incident. And if the budget genuinely is not there this year, the Base tier puts the method in your hands for roughly the cost of a mid-level contractor for two months.
Our regulator has not asked us for this.
That is the best possible time to run it. Doing this work under a matter requiring attention means doing it on someone else's schedule, with someone else watching, and with none of the credit for having moved first.
10 / Next step
Tell us where you are, and we will tell you which tier fits.
A first conversation is thirty minutes and covers three things: what triggered the question, what your current remediation clock actually looks like, and who would have to sign a charter. If EVOCATI is not the right instrument for your situation, we will say so in that call.
Institutions starting on the Base tier can be running inside two weeks.
Sources for every figure on this page
88 percent exploited within 48 hours: CrowdStrike 2026 Threat Hunting Report
Median 120 to 80 days, 23.43 percent same-day exploitation, 1.3 percent of AI-discovered vulnerabilities, Project Glasswing funnel: VulnCheck, State of Exploitation 1H-2026
Average mean time to remediation of five months and ten days: Qualys Enterprise Patch and Remediation Benchmark 2026
4,268 to 8,539 high and critical disclosures, Q2 2025 to Q2 2026: Rapid7, reported by SecurityWeek
88 percent exploited within 48 hours: CrowdStrike 2026 Threat Hunting Report
Median 120 to 80 days, 23.43 percent same-day exploitation, 1.3 percent of AI-discovered vulnerabilities, Project Glasswing funnel: VulnCheck, State of Exploitation 1H-2026
Average mean time to remediation of five months and ten days: Qualys Enterprise Patch and Remediation Benchmark 2026
4,268 to 8,539 high and critical disclosures, Q2 2025 to Q2 2026: Rapid7, reported by SecurityWeek
EVOCATI and PRINCIPIA are offerings of Trilogyworks.Method, scoring model and artifact contents licensed.