Skip to main content
Roadmap

Shield protects. Flow runs.
Here's what proves, watches, and learns next.

Each rung of the ladder answers the next question a team asks after the last one is settled. Everything ships the same way: local-first, source-delivered, no subscription.

ShippedIn buildOn the drawing board

PurfectShield

ShippedQ2 2026
“Is my data safe?”

The local gateway that swaps out secrets and PII its packs match before prompts reach your model provider — and proves it with a hash-chained, tamper-evident audit log.

Filter packs per domainSession-keyed placeholdersHash-chained audit + dashboardSource delivered at handoff

Shield Desktop Coverage

ShippedQ2 2026
“What about the apps my team already uses?”

Routes desktop AI apps — Claude Desktop, Cursor, VS Code Copilot, Zed, Continue — through the gateway, with backup, restore, and health checks per app.

Five apps covered todayOne-command install & verifyClean uninstall, configs preserved

PurfectFlow

In buildQ3 2026
“Are my workflows running?”

The workflow product: named agents in your Slack, Telegram, webhooks, and schedules — every model call through Shield, every action chronicled, spend counted in dollars. Three growth stages: 🌱 Root (RUN), 🌿 Shielded (SAFE), 🌸 Cambium (LEARN).

Engine, desktop app & CLI builtFive connectors + connector SDKFlow Trace journey replayIn review — shipping soon

Purfect Lens

On the drawing boardQ4 2026
“Can I trust what my agents are doing?”

The agent trust platform. Replayable behavior narratives ("show me why the agent did that"), cost intelligence that classifies outcomes instead of counting tokens, and one-click evidence packs that support your compliance program.

Behavior trace & anomaly baselinesPer-workflow ROI, in dollarsDual-chain verified evidence exportWorks on your existing history

Shield OS

On the drawing boardH1 2027
“What about the AI apps I don't know about?”

The AI firewall: an OS-level extension that sees every app talking to a model provider, routes the known ones through Shield, and makes the unknown ones visible — policy by MDM, evidence on the same audit chain.

System-wide coverage inventoryPer-app route/allow policyCoverage-drift alerts

Purfect Grove

On the drawing boardH1–H2 2027
“Can one team run this for many clients?”

The fleet tier for agencies and MSPs: one dashboard over every client install — integrity checks, signed pack distribution, white-label branding, per-client evidence. Client data stays on client machines; only integrity heads and counters federate.

Fleet dashboard & licensingSigned pack & policy pushPer-client evidence export

Essence & House Voice

On the drawing boardH2 2027
“Can my agents sound like us — without uploading us?”

A portable, encrypted context vault your agents load locally — with every learned fact chronicled, and forgetting recorded rather than hidden. Plus House Voice: a small private model fine-tuned on your content for tone and style, running on your hardware.

Local personal & brand contextProvenance for every learned factPrivate voice models per client

Drawing-board items are design specs, not promises — sequence and scope follow real customer engagements. When something ships, it ships the way everything here ships: on your machines, source included.

Want to shape what gets built next?

Roadmap sequence follows the teams we work with. Tell us the question you need answered.

Talk to us