Conflicts escalate when people live in different stories. Tacitus ingests the emails, speeches, media, and memos around a dispute, runs multiple deep research agents, and helps humans see the full picture — then search for common ground with AI that explains itself.
Information is not the enemy of peace; it is the infrastructure. Tacitus unlocks that information — reconstructing narratives, stress-testing assumptions, and giving leaders a structured way to resolve conflict rather than live inside it.
Why Tacitus
From HR grievances to election narratives to ceasefire talks, the history of a conflict is usually scattered across inboxes, chat logs, press clips, and draft documents. Without a way to reconstruct that history, conflicts are governed by whoever shouts loudest or remembers best.
HR and legal teams are drowning in threads, not insights. Tacitus rebuilds an auditable record of “what happened, who said what, and under which policy” so leaders can act with speed and fairness.
Modern campaigns and parties operate inside competing storylines about identity, economy, and democracy itself. Tacitus maps these narratives, checks them against data, and clarifies where persuasion or de-escalation is still possible.
In conflict zones, narratives are contested across borders and media ecosystems. Tacitus helps mediators, analysts, and practitioners ingest reports, statements, and coverage to see fault lines — and where thin threads of common ground remain.
Solutions
Under the hood, Tacitus is the same conflict resolution engine. On the surface, it speaks the language of HR, campaigns, and peace processes — with tailored playbooks, data sources, and guardrails for each.
For people teams, in-house legal, compliance, and line leaders who need to handle complex cases without losing trust.
For campaign teams, parties, think tanks, and analysts who need to understand polarization without being consumed by it.
For mediators, multilateral missions, and NGOs who work in and around live conflicts and need a structured way to see the information landscape.
Product
Tacitus has three primary surfaces: the Email Teammate that lives where conflict appears; the Deep Research agents that interrogate context and assumptions; and the Resolution Workspace where humans endorse, revise, and record outcomes.
Forward or sync key threads. Tacitus ingests the conversation, builds a party map, extracts issues, and constructs an evidence graph. You can work entirely from your inbox.
Tacitus spins up multiple specialized agents: one for policy and law, one for precedent cases, one for media and polling, and one for domain-specific sources (e.g., human rights reports).
The final surface is a structured workspace: caucus notes, options, rationales, and the eventual decision or communique live together, with a clear record of which suggestions came from AI and which from humans.
Beyond the UI, Tacitus exposes a JSON-first API so you can integrate party inference, narrative mapping, evidence graphs, and assumption audits into HRIS, campaign tools, or peacebuilding dashboards.
curl https://api.tacitus.me/v1/cases/analyze \
-H "Authorization: Bearer <token>" \
-H "Content-Type: application/json" \
-d '{
"workspace_id": "people-ops-eu",
"domain": "business_hr",
"emails": [...],
"policies": ["HR-12", "HR-22"],
"run_deep_research": true,
"run_assumption_audit": true,
"run_common_ground": true
}'
In the ZIP, see assets/sample-case-hr.json and
assets/sample-case-politics.json for examples of Tacitus-style case graphs.
Engine
The core of Tacitus is a set of coordinated agents that treat conflicts as information problems: who knows what, who believes what, and where those maps can be brought closer together without erasing disagreement.
Tacitus does not pretend that every conflict is solvable. Instead, it runs systematic searches for potential convergence while keeping track of red lines and asymmetries.
In other words: Tacitus does not replace the politics or the diplomacy of conflict resolution. It simply makes the information dimension less opaque, so those human processes have a stronger foundation.
Architecture
Tacitus assumes large language models will keep improving and commoditizing. The durable value lies in how you orchestrate them around your data, your processes, and your institutional norms.
| Concern | Design choice |
|---|---|
| Hallucinations | Retrieval-first; unanswered questions allowed; low-confidence reasoning and missing data flagged explicitly. |
| Bias | Assumption auditor extracts value-laden claims and prompts humans to inspect them by party, issue, and source. |
| Privacy | No training on customer data; bring-your-own-model options; case- and domain-scoped indexes. |
| Consistency | Playbooks and analytics re-use patterns across similar conflicts while preserving local context. |
Domains
Below are example scenarios drawn from the three Tacitus lines. Each shows how the same engine behaves differently when paired with domain-specific data and norms.
An employee disputes a below-expectations rating citing ambiguous expectations and poor feedback. Threads span six months, multiple managers, and a re-org.
A strategic client claims breach of contract after a delayed rollout. Internal and external threads disagree on who shifted scope and when.
Two major campaigns frame the same set of economic indicators in opposite ways. Polling shows confusion and rising distrust among swing voters.
A party’s youth wing and parliamentary group clash over a coalition decision. Public messaging is incoherent; activists feel betrayed, leadership feels cornered.
A mediator is asked to support talks around a fragile local ceasefire where national forces, local militias, and regional sponsors all have different narratives.
Public statements harden; media on each side highlights different incidents; regional organizations are exploring de-escalation options.
Deep Research
The Deep Research agents are optimized for one thing: reducing avoidable ignorance in conflict. They do not tell you what to think; they ensure you can see what is already knowable, and where the gaps truly are.
Conflicts are full of unstated assumptions: about intent, capability, legitimacy, and history. The assumption auditor extracts these from language and evaluates them against the evidence graph.
FAQ
We are deliberately opinionated about the role AI should play in conflict resolution.
No. Tacitus is designed as an analyst and note-taker, not a judge or strategist. It helps you reconstruct history, interrogate assumptions, and draft options — but humans decide, endorse, and own final outcomes.
In production, Tacitus is deployed in your preferred cloud region with case- and domain-scoped indexes, encryption at rest and in transit, and optional single-tenant or VPC-peered setups. Foundation models are never fine-tuned on your data.
Yes. Tacitus is designed to plug into your existing LLM providers, logging, and observability stack. The orchestration and resolution logic sits above the model layer, so you can swap engines as requirements or vendors change.
We are working with a small set of organizations that handle complex conflicts regularly — people teams, campaign rooms, ombudsperson offices, and mediation practices.
We will not use or train on your content without explicit agreement.
Share your email and a short context note. We'll route it straight to [email protected] and follow up personally.