Plain-language answers to the ten objections we hear most from prospects landing from cold outreach — what an AI employee actually is, how fast it deploys, where your data lives, and why our prices just dropped 75% without the quality following them down.
An AI employee is an autonomous agent deployed against your own data, running on a cron schedule, owning a business domain end-to-end. In the audit agent's case, it sits on top of your processes, audit_results, and digest_reports tables — mapping every repeatable process, scoring it 0–100 for automation opportunity, and writing a daily digest without anyone prompting it.
A chatbot waits for input. Copilot suggests inside one app, one tab, one moment. An AI employee maps the whole business, scores opportunities across every domain, and reports every day — with no one babysitting it. You see what changed on the /app dashboard; you don't run a console.
2–4 weeks from kickoff to a live agent writing daily digests.
Week one is kickoff and credentials — we get read-only access to the source systems and confirm the first process to map. Week two is build: the first agent goes live, the digest job kicks in (declared in polsia.toml as a [[crons]] entry, code under jobs/), and the scoring engine in services/ writes its first audit_results rows. Weeks three and four are tuning: accuracy checks, edge cases, and the written roadmap for the second agent.
The agent runs on Render with a Neon Postgres database, with infrastructure defined in render.yaml at the repo root. Every row is tenant-scoped by company_id — there is no shared pool between customers.
We never train on your data. Credentials are stored encrypted at the platform layer (managed secrets, not in source). All access is over TLS; access logs are auditable on request.
No. The cut is driven by cost compression in how we deploy — not by trimming the agent or the people behind it.
Three things shifted. One, the audit scoring runs algorithmically in services/ instead of via consultant hours — the same engine runs in production, so the scores you see on the free audit are the scores your live agent will produce. Two, the daily digest is cron-declared in polsia.toml against jobs/ — a standalone Node entrypoint that runs once and exits — instead of bespoke per-customer builds. Three, the whole stack runs single-tenant Render + Neon, the same way it has from day one. No custom infra per customer, no separate ops surface.
None of that changes the agent scope, the digest cadence, the audit depth, the engineering team, the SLA commitments, or the integration layer. The same agent still runs every domain you contract for, at the same daily cadence, with the same data-security posture. Tier pricing moved from Starter $10,000/mo to $625/mo, Growth $10,000/mo to $2,500/mo, and Enterprise $25,000+/mo to $6,250+/mo. The published tiers on the /pricing page are the contract; convertible to enterprise terms where the ROI math justifies it.
All three tiers ship the free automation audit as the on-ramp. The full breakdown lives on the pricing page; short version:
Starter — $625/mo. One AI-agent domain, daily digest reports on the dashboard, the initial automation audit, and email support. The on-ramp tier for a single process you want an agent owning.
Growth — $2,500/mo. Four AI-agent domains, priority daily digest, full business audit, dedicated onboarding, and Slack + email support. The mid-market tier — covers most ops stacks out of the box, custom connectors included.
Enterprise — $6,250+/mo. Unlimited AI-agent domains, custom integrations, a dedicated AI success manager, SLA guarantees, and phone + Slack support. Quote-based on scope; convertible from Outcome-based (share-of-savings) terms when the ROI math justifies it.
The free intake lives at /ai-audit — one business day turn-around, written 1-pager even if you never sign.
Yes. The integration layer speaks REST, SQL, and file drops (CSV/SFTP), which covers NetSuite, SAP, Oracle, Manhattan, Blue Yonder, and the long tail of mid-market systems in use today.
Email integration runs over standard SMTP/IMAP or the dedicated APIs for Gmail Workspace and Microsoft 365. If your stack needs a custom connector, the build is included in the Growth and Enterprise tiers — no separate change order, no surprise invoice.
You own 100% of the data. Every row in Postgres is company_id-scoped to your tenant; nothing is pooled between customers, and nothing leaves your boundary for any external training or model improvement.
Full export on request: every row of companies, processes, audit_results, and digest_reports in CSV or JSON. Account deletion is tenant-scoped row deletion — we retain nothing after deletion. No third-party training on your data — full stop.
Pay-per-agent and the post-cut Starter / Growth tiers cancel any time from the dashboard, no penalty. Prorated refund of the unused month on request.
The 12-month minimum applies only to Outcome-based (share-of-savings) deals, and is disclosed at signing. No long-term lock-in on Starter / Growth.
Tiered SLAs, in writing.
Starter. Email support, next-business-day response. Self-serve dashboard with the digest, audit, and process explorer.
Growth. Slack + email, same-day response within business hours. Dedicated onboarding through the first agent going live.
Enterprise. Phone + Slack + dedicated AI success manager, 4-hour response, monthly ROI report. Incident SLA and uptime publish on the dashboard.
The free intake lives at /ai-audit — three fields, one business day, no discovery call required. You get a written 1-page report even if you never sign.
The audit uses the same scoring engine that runs in production against services/, so the scores on the report match what the live agent will produce. There's no obligation — if the ROI math doesn't work, you keep the report and walk.
Three fields, one business day, no discovery call. Submit the intake form and we'll send your 1-page automation audit.