Starting today, Dictiva is positioned as an AI Agentic Governance Platform — the policy and procedure layer for AI agents, and the humans who run them.
What changes
The hero. The metadata. The pricing narrative. The SEO cluster we go after. The content strategy. The roadmap order.
What doesn't change: the company name, the product, the tenant data, your account, your API keys, your MCP integrations, or the subscription you're on today. This is a positioning move, not a rebuild.
Why now
Three things converged:
One — demo disappointment. When prospects heard "governance platform" they arrived expecting full GRC. When they heard "statement-first" they arrived curious. When they actually got into Dictiva, the thing that blew them away was never the statement lifecycle (though it's polished) — it was the fact that their AI coding assistant, or their customer-support agent, or their data pipeline orchestrator could read the policy corpus directly via MCP, understand which rules applied to it (human, agent, or both), and be audited against the exact statement version it saw at decision time. That's not a GRC story. That's a different story, and it's one the product is uniquely good at.
Two — the category window is closing. OWASP published their Top 10 for Agentic Applications in December 2025. Microsoft shipped an Agent Governance Toolkit in April. EU AI Act high-risk obligations take effect August 2026; Colorado's AI Act is effective June. Gartner's AI Governance Platforms market is growing 36% CAGR. Every incumbent GRC vendor will retrofit an "agent governance" module by Q1 2027. If we want to be part of how this category gets named, we have to move now.
Three — the schema already believes this. Every Dictiva statement has been carrying enforcement_mode, actor_applicability, and an agent_guidance JSONB blob for months. The MCP server has been live since mid-April with six tools across stdio and Streamable HTTP. The ai_audit_logs table has twenty-five columns and four indexes. These aren't new product surfaces. They're surfaces we under-claimed because our marketing was pointed at a different buyer.
What ships today, what ships later
I'd rather be specific than vague, so here's the honest split.
Shipping today: statement-first machine-readable policy. Structured agent guidance per statement (allowed actions, prohibited actions, required approvals, escalation rules, evidence requirements, failure modes). Live MCP governance server with six tools and dual transport. Full agent audit trail with export on Scale+. AI Context Layer with six grounding modules. Library and adoption model with divergence tracking. Regulatory mapping to 57 frameworks.
Shipping Q3 2026: procedure execution runtime (procedures have the schema; the executor doesn't yet). Agent approval workflow UI (statements declare requiredApprovals; the in-flow UI is in flight).
Shipping Q4 2026: audit-ready evidence dashboard that aggregates the primitives into a compliance-facing pack. Risk register expansion beyond process-level.
That's on the homepage now. "Coming Q3" badges instead of silent absence. If we slip, you'll see it; if we ship early, you'll see that too.
About pricing
Seat-based tiers stay through Q2 2026 unchanged — Community, Professional ($299), Business ($799), Enterprise (volume). Your Stripe subscription and AI credits keep working exactly as they do now.
In Q3 2026 we're rolling out a hybrid model with a new top tier called Agent Cloud — subscription floor plus metered overage, denominated in "governance actions" (not tokens). Existing paying customers will be grandfathered at current pricing for at least six months. More on this in a pricing deep-dive soon; the short version is that token-metered billing is a trap for non-engineering buyers and we won't ship that.
What to do with this
Nothing, probably. If you're using Dictiva today, keep using it. If you're evaluating us, the new positioning page is the best single read on what we actually do well. If you're an agent-heavy team looking at Fiddler or Arthur for observability, you should look at us alongside — not instead, necessarily, but alongside.
If you think we're wrong to narrow, tell me: carlos@dictiva.com. I'd rather hear it now than in a churn exit survey.
— Carlos