Why VoidAi

Different shape.
Different problem.

Most "AI vendor" categories are about building agents. VoidAi is about the governance layer underneath — what makes the agents safe to run unattended in a regulated, audited environment. This page is the honest comparison every procurement team eventually asks for.

Positioning axes

AXIS

VOIDAI

"AGENT VENDORS" (TYPICAL)

Hosting model

Customer-hosted. Per-customer install. Zero data egress to vendor.

SaaS, multi-tenant. Customer data flows through vendor infrastructure.

Audit trail

Append-only JSONL ledger. 12-field canonical schema. Forensic-grade — bytes never overwritten on heal. Stress-tested 0/2000 torn reads at 200 writes vs 10 readers.

Application-level logs. Often retroactive. Frequently no public schema.

Capability scoping

Per-agent allowlist + immutable red-line list. Guardian denies before consulting any spec. Click the demo below.

Prompt engineering + role-based system messages. Soft guardrails, not architectural.

Promotion process

48h shadow window + eval-driven gate + spec-before-code. No "trust me" overrides. Audit-row evidence required.

Feature flags, A/B tests, manual sign-off. Often no shadow.

Compliance posture

SOC 2 Type II readiness via Vanta in progress; architecturally aligned today (TSC mapping in reference architecture). Single-tenant + audit log + capability gates pre-answer most questions.

Various — SaaS vendors carry their own attestations; data still flows through them.

Update channel

Customer-controlled GitHub mirror. Customer pulls when ready. VOIDAI_TELEMETRY_ENABLED=false default.

Auto-updates pushed by vendor. Visibility into what changed = zero unless documented.

Pricing model

Three packagings, all anchored: Solo $5–12k engagement, Mid $15–35k + MRR, Enterprise $75–250k + MRR.

Mostly per-seat or per-API-call. Hard to predict at scale.

Four neighboring categories — when each is the right pick

We don't think any of these are "wrong." Each solves a real shape of problem. The question is whether your shape matches ours.

The "agent platform" vendors

e.g., Voiceflow, Cresta, Sierra, similar

they solve: Building agents quickly. Often have great IDE / visual canvas / prompt-mgmt UX.

we solve: Governance, observability, audit, capability scope. Things that matter when an agent ships into a regulated environment.

overlap: Both build agents. We expect customers to use those tools to BUILD; VoidAi installs the governance layer UNDERNEATH so the agents can actually run unattended.

pick them if: If you need agent-building IDE + you do NOT need audit / capability gates / customer-hosted / single-tenant.

pick us if: If a procurement, security, or compliance team will see this. Or if "deploys unattended" is non-negotiable.

The "framework" players

e.g., LangChain Inc, LlamaIndex, similar

they solve: Open-source primitives for orchestrating LLM calls + tool use + memory. Highly flexible.

we solve: A productized governance + observability stack on top of those primitives.

overlap: None really — frameworks are libraries; VoidAi is a deployment + ops layer. We could install on top of LangChain agents.

pick them if: If you want library-shaped LEGO bricks and a team to assemble them yourselves.

pick us if: If you want a bounded, audited, capability-gated install that ships in 2-12 weeks and is safe to leave unattended.

The "AI consultancy" route

e.g., Boutique AI shops, AI arms of big consultancies

they solve: Custom AI builds for one customer. Often shipped as bespoke artifacts.

we solve: A reusable platform + reusable governance, installed once, owned by the customer.

overlap: Both deliver AI work to customers. Difference: theirs is mostly bespoke; ours is mostly productized.

pick them if: If your problem is genuinely one-of-a-kind and you can fund 6-12 months of bespoke build.

pick us if: If your problem fits a recurring shape (inbox triage, customer support, compliance evidence-gathering, etc.) and you want an asset that compounds.

The "DIY" path

e.g., In-house team building on Anthropic / OpenAI APIs directly

they solve: Maximum customization. Direct control over every line of code.

we solve: The 80% of governance + ops infrastructure you would have built anyway. Plus 24+ pre-built agents you can adapt.

overlap: Same fundamental tools (Anthropic API, file-system, audit logs).

pick them if: If you have 1+ FTE-years of engineering bandwidth available AND your needs are unique.

pick us if: If you would rather not rebuild WriteShield, Guardian, audit-write, integrity-sweep, the autonomy ladder, the eval runner, and 26 worker profiles from scratch.

Where we lose — honestly

A page that only lists strengths is marketing, not a procurement artifact. Here are the trade-offs we do NOT pretend to win:

// HONEST

Multi-tenant scale

Single-tenant by architecture means we cannot offer the per-seat economics of a SaaS. If your model is "$10/user/month for 10,000 users," we are not it.

// HONEST

Massive multi-region edge deployments

Today the platform installs on a single Windows or Linux host per customer. Multi-host orchestration is on the roadmap; not shipped.

// HONEST

Pre-existing IDE/visual builders

Agents are configured via spec.json + worker.md frontmatter. There is no drag-and-drop canvas. Engineering teams find this faster; non-engineers find it slower.

// HONEST

Hours-old startup cred

VoidAi is a 1-founder company with one in-progress SOC 2 attestation. If your procurement team requires proven track record at scale, we are honest about that and can introduce design-partner references when available.

The architectural differentiator — feel it

The single biggest thing competitors don't have: a per-agent capability allowlist + an immutable red-line list, enforced by the Guardian on every action. Click any action below to see the decision in real time.

Guardian — capability gate

deny-by-default · red-line list immutable

0 allowed 0 denied

Try an action

Cyan = within allowlist. Yellow = denied by deny-list. Magenta = red-line — immutable.

Guardian decision

Awaiting input.

No competitor we've benchmarked has this enforcement model.

Still comparing?

Show us the other contender.

Send us the deck or proposal you're evaluating. We'll respond with a side-by-side that names where they win + where we win, in your specific deployment shape. No sales theater.

Email us the comparison →