Company · Roadmap
Gated, not dated: what lands in each wave
One3 is built in waves W0–W6. Each wave has exit criteria; if they aren't met, the next wave doesn't start. Year 1 = W0–W2.
How to read this roadmap
The roadmap is gated, not dated. Each wave has exit criteria. If they aren't met, the next wave's scope doesn't start — the gap is fixed first. The month and year ranges below are when a wave is earliest eligible, not a commitment. The strategy is re-cut every year from signposts and measured results.
Year 1 = W0–W2, fixed by the Year-1 Product Boundary: One3 Growth OS + Agency OS + rentable Revenue Agents, on a governed Action kernel, with a monthly Investment & ROI statement for every renting client.
Waves at a glance
| Wave | When (earliest) | Theme | Headline capabilities |
|---|---|---|---|
| W0 | Months 0–3 | Kernel + AI OS core | Tenants, Identity, Action Bus, Automations, Model Hub, Copilot, Knowledge, Fact Register |
| W1 | Months 4–6 | Growth OS MVP + first agents | CRM, Marketing, Forms, Desk, Content, SEO, Agents Studio, Dashboards, Migration Factory, Connector Health, Investment & ROI Ledger v0 |
| W2 | Months 7–12 | Growth OS GA + Agency OS + agent rental | Ads, Social lite, Billing, Agency OS, Digital Employee Store, Squad Builder, Revenue Squads, Avatar Studio + Rights Registry, AI Lab, FinOps |
| W3 | Year 2 | Business OS + builders + catalog scale | People & Finance Clouds, Office core, Apps GA, BI, Enterprise Brain v1, Training Gym, Certification, Outcome Contracting, Marketplace, first Industry OS |
| W4 | Year 3 | Operations + Decision Intelligence | ERP, Decision Engine GA, Simulation, multi-cell, ISO 42001, more Industry OS |
| W5 | Years 4–5 (+ Y6 consolidation) | Autonomous Enterprise | Business Twin, Factory, IoT/Vision, One3 Box, 1000+ agent catalog, One3 action models, business gyms |
| W6 | Years 6–10 | Agent Economy & Physical AI | Agent Network, agent payments, robotics bridge, global cells |
| H5 | Years 11–15 | Enterprise intelligence utility | Re-planned yearly from signposts |
W0 — Foundations (months 0–3)
A working kernel on which one Pack (CRM-lite) runs end to end through UI, API, MCP and an agent, with policy, audit and cost tracking. Exit criteria:
- 1
Tenant isolation proven
A synthetic tenant is created via API in under 60 seconds, with 0 leaks in automated cross-tenant tests.
- 2
Generated Packs
The CRM-lite Pack is at least 60% generated from its manifest *(measured)*.
- 3
Caller parity
The same Action from UI, REST, MCP and an agent produces identical audit and event records.
- 4
Durable approvals
An R3 Action from an agent pauses for approval and resumes after a service restart.
- 5
Honest undo
Every Action declares a reversibility class and the UI wording follows it.
- 6
Grounded Knowledge
Groundedness ≥ 0.85 and citations on 100% of answers on an internal eval set *(target)*.
- 7
Cost on every call
Every model call has a trace with cost; daily cost report per tenant.
W1 — Growth OS MVP + first agents (months 4–6)
5–10 design-partner agencies using One3 weekly on real client work, with the first agents in shadow mode. Exit criteria:
- 1
Weekly design partners
At least 5 design partners active weekly for 4 consecutive weeks *(target)*.
- 2
Flows run for real
Onboarding, Lead to Cash (to deal won), the campaign loop (email/WhatsApp) and Issue to Resolution run end to end for real partners.
- 3
Earned autonomy
At least 3 agent skills promoted from A1 to A2 on real data with the minimum Eval Coverage Score.
- 4
Every agent cost captured
Every agent-caused cost appears in the ledger with trace linkage; caps block spend in tests.
- 5
Isolation in CI
The cross-tenant isolation suite runs in CI with 0 leaks.
- 6
A migration path proven
CSV or HubSpot passes dry run and reconciliation on realistic data.
- 7
Pricing tested
Willingness to pay captured with design partners, including agent rent and usage credits.
W2 — Growth OS GA + Agency OS + agent rental (months 7–12)
Public launch; agencies resell One3 and rented agents. Year-1 gate:
- 1
Agencies active
10 weekly-active agencies managing at least 30 client sub-accounts *(targets)*.
- 2
Unit economics measured
Gross margin including AI and pass-through at the target set from W1 data; AI cost ≤ 15% of revenue *(target)*.
- 3
Agent catalog
20 agent templates Listed, 6 Verified; 3 Revenue Squads live with design partners.
- 4
Statements for everyone
Monthly Agent Investment Statements delivered to 100% of renting tenants, reconciling to provider invoices within tolerance.
- 5
Return evidence
Return evidence for at least 3 agents across at least 3 tenants each (assisted at minimum; experimental where holdouts ran).
- 6
Safe autonomy
At least 20% of routine Actions at A2+ with ≤ 2% reversal or compensation.
- 7
Zero incidents
0 avatar consent incidents and 0 cross-tenant isolation findings.
- 8
Switching and trust
GoHighLevel and HubSpot migrations tested; launch-blocking facts verified; SOC 2 Type I or ISO 27001 Stage 1 completed.
Not now in Year 1
| Deferred | Year-1 substitute |
|---|---|
| Chats; Mail and Calendar clients | Notifications and Gmail/Microsoft 365 integrations |
| Full Projects & Timesheets | Client work board lite |
| Sites; Research; Survey; Documents & Reports | Landing pages in Marketing; fixed report templates; CSAT in Desk |
| Full Goals & Autopilot; Context Graph | Squad goals; relations and MDM-lite |
| Native mobile app | PWA plus WhatsApp approvals |
| Third-party Marketplace and agent listings | First-party Digital Employee Store only |
| Training Gym, Certification, Outcome Contracting, Replacement, Liability Desk as full apps | Internal scorecards, Verified badge process, rent + credits + capped bonus, manual replacement, incident workflow |
W3 — Year 2: Business OS + builders
What lands: People (Core HR, Recruit, Attendance & Leave, Payroll India, Expenses, Performance), full Accounts with GST returns data, TDS and bank reconciliation, Legal CLM and Sign; Apps Builder GA, Data Fabric and BI Studio; native mobile, full Projects and Timesheets and other Year-1 deferrals; Docs, Sheets and Meet; Studio full canvas and Video v2 with consented avatars; Contact Center and Customer Success; Context Graph, full Goals & Autopilot and a Decision Engine alpha; Marketplace GA with third-party agent listings; Training Gym, Certification and Replacement Engine; an Outcome Contracting pilot; first Industry Editions (Real Estate, Education, Professional Services), each only with three design partners signed.
Exit: at least 30% of active tenants use 3+ suites; tenants build custom Packs in the Apps builder; payroll accuracy 100% on parallel runs before GA; around 100 agent templates (target) with certification levels; SOC 2 Type II.
W4 — Year 3: Operations + Decision Intelligence
What lands: Inventory & Warehouse, offline-first POS, Commerce with ONDC, Procurement, Assets, Field Service, FP&A and GRC; Decision Engine GA, Forecast & Simulation and a Data Science Workbench; Slides, Learning, ITSM, Loyalty, Events and DevHub; around 300 agent templates (target) and the Certified Outcome level; Retail, Construction, Clinic and Automotive Editions; multi-cell (India plus one international region), silo tenancy, per-tenant adapters, ISO 42001 and BYOK.
Exit: Decision Engine recommendations accepted and executed with a measured outcome lift on at least 3 customer cases (documented, not claimed); ERP customers live on inventory and accounts.
W5 and W6 — Years 4–10
W5 (Years 4–5, plus Year-6 consolidation) — Autonomous Enterprise: Business Twin, Factory, Supply Chain, Manufacturing & Quality, IoT Hub, Vision AI, One3 Personal, the One3 Box sovereign appliance, RL business gyms, more Industry Editions, outcome-based pricing for Digital Employees only after the attribution gate passes, and 1000+ rentable agent templates through full Role Factory automation (target). Proof: goals run through Autopilot with at least 50% of routine Actions at A3+ in mature tenants.
W6 (Years 6–10) — Agent Economy & Physical AI: the Agent Network with cross-company negotiation and settlement under policy, agent payments through licensed partners, a Robotics Bridge, industry foundation adapters and global partner ecosystems.
H5 (Years 11–15): defined at the Year-10 strategy refresh.
Pivot and kill triggers
| Signal (measured) | Action |
|---|---|
| W0 exit not met by sprint 9 | Cut kernel scope; ship CRM and Desk with less abstraction |
| Fewer than 3 design partners weekly-active by sprint 12 | Stop feature work for 2 sprints; customer discovery; consider changing the ICP |
| AI cost above 30% of revenue for 2 months | Freeze new AI features; routing, caching and local-model sprint |
| Agent reversal or compensation rate above 5% on a skill | Auto-demote the skill; eval sprint before re-promotion |
| A template shows no positive assisted return for 3+ tenants | Pause new rentals; retrain or retire |
| Any avatar consent incident or isolation finding | Stop the affected feature for all tenants until fixed and re-audited |
| An app's weekly active tenants below 10% of eligible after 3 months | Stop investing; merge or deprecate |
Roadmap questions
Are these dates commitments?
No. Waves are earliest eligible. A wave starts only when the one before it meets its exit criteria.
What is in Year 1?
W0–W2: the kernel and AI core, Growth OS, Agency OS, the Digital Employee Store with 20 Listed and 6 Verified templates, 3 Revenue Squads, the Investment & ROI Ledger, synthetic avatars with consent controls, and migration from GoHighLevel and HubSpot.
When do HR and finance arrive?
People (HR) and full Accounts are W3, starting in Year 2.
When will there be 1000+ agents?
That is a W5 target for Years 4–5, reached through catalog growth gates.
How often is the roadmap re-planned?
The strategy is re-cut every year from signposts, drills and measured results, with an explicit build, deepen, integrate, partner, sunset and not-now list.
Help us hit the W1 gate
The W1 gate needs real design partners using One3 every week. That could be you.