Fc
Fortes Co-pilot
Mission Control · v0.1
Atyab Rehman
IT Lead
AR
›The operational substrate

About this platform

Fortes Co-pilot is built on six years of agent orchestration work across regulated industries. This page is for the people running it.

Platform version4.7.2·Hermes runtime · Buildf8a31c·Last review22 May 2026by Mani Kanasani

This page is admin-visible to: Vice Chairman, Director of Education, IT Lead, Principals. Other staff: hidden.

Section 1 · How the work is structured

The Three Roles framework

Architect · Manager · Worker

Architect

Defines the agent topology. Chooses which AI does what, where the handoffs happen, what data flows where, what gets human approval and what doesn't. At Fortes, the Architect role designed:

  • ◆The five-agent topology (Lexa, Watson, Sherlock, Nova, Ray)
  • ◆The escalation policy tree (14 active rules)
  • ◆The RBAC matrix across 12 roles
  • ◆The Hermes runtime configuration for UAE data residency

This work is done once, refined quarterly.

Manager

Sets the operating rhythm. Defines what "good" looks like, monitors skill performance, schedules retraining cycles, makes the call on whether a new opportunity from Intel becomes a build item.

  • ◆Runs the weekly skill review every Friday 14:00 GST
  • ◆Every agent's accuracy across every task is reviewed
  • ◆Any skill that dropped >2% triggers a root-cause investigation within 4 hours

This work is continuous, never done.

Worker

Writes the actual prompts, the skill modules, the integration code. Implements the Architect's decisions. Responds to the Manager's findings. Worker tooling at Fortes includes:

  • ◆Claude Code, Codex 5.5, Antigravity (the build runtimes)
  • ◆OpenClaw for prompt orchestration
  • ◆Hermes for the production runtime
  • ◆A 169-skill library that grows ~12 skills/month

This work is the visible tip. The Architect and Manager work is invisible from your end but consumes ~65% of total engineering time.

Most internal teams attempting this kind of build see the Worker layer and only the Worker layer. The Architect and Manager disciplines take 18–24 months to develop and are why most internal agent projects stall around month 6.

Section 2 · The build pipeline

The 8-Phase Autonomous Build Pipeline

PHASE 011–2 wk
Discovery

Workflow shadowing + xlsx + interview synthesis

Output: clustered automation requirements
PHASE 021 wk
Architecture

Topology design, RBAC, integration mapping, residency plan

Output: locked architecture document
PHASE 031 wk
Skill enumeration

Map automation requirements → atomic skills

Output: skill graph · Fortes: 169 atomic skills mapped
PHASE 043 d
Heartbeat config

3-minute monitoring cadence, alerting, kill switches

Output: runtime health envelope
PHASE 053 wk
Shadow build

Build skills against historical data; no production touch

Output: 7-day shadow audit pack per skill
PHASE 062 wk
Pilot deploy

Limited cohort, human-in-loop for every action

Output: pilot acceptance criteria met or rejected
PHASE 072 wk
Production rollout

Expand cohort, auto-actions enabled with approval policies

Output: live system
PHASE 08ongoing
Continuous improvement

Weekly skill review + Intel discovery + orchestration tuning

Output: compounding value

The Fortes Phase 1 build runs phases 1–7 over 12 weeks. Phase 8 starts at week 13 and never ends.

Section 3 · System-wide telemetry

Live operational metrics — across all Vertical Systems deployments

Not just Fortes. This is where the multi-client moat becomes visible.

Active deployments
0
Healthcare (US)2 · production
Dental (CA)1 · production
Legal (US)1 · production
Federal defense consulting1 · pilot
Education (Fortes Holdings, UAE)1 · prototype → Phase 1 starting
System-wide this month
AI actions executed
0
Human approvals routed
0
Skills shipped (new + upgrades)
0
Customer-impacting incidents
0
Median skill maturity time (days)
0
P99 inference latency (ms)
0
Cross-deployment learnings this quarter
Tone variants (added to library)18
Edge cases catalogued47
Compliance patterns surfaced14
Regulatory updates absorbed9

Vertical Systems' skill library compounds across clients. A tone variant developed for a US dental practice's parent-friendly billing reminder informed Fortes' fee reminder voice. A compliance pattern from federal defense work informed Fortes' KHDA audit format. Internal builds don't get this leverage.

Section 4 · Named deployments

Case work

Specific deployments with specific operational outcomes.

Amalgamind (ScribbleSoft case study)

IndustryHealthcare ops · enterprise
RoleMani Kanasani — CTO, ScribbleSoft
Engagement
  • ·Duration: 4 days
  • ·Team: 1 Architect, 1 Manager, 2 Workers
Output
  • ·201 documents generated
  • ·471 human-hours of equivalent work delivered
  • ·7-figure pipeline opened for ScribbleSoft
Public
  • ·Documented in OpenClaw masterclass (Mar 2026 · 111K views)

What Vertical Systems contributed: the Three Roles operating model, the OpenClaw orchestration patterns, the audit-compliance scaffolding.

US dental network (anonymised — under NDA)

IndustryDental services · 14 locations
ScopePatient communications, billing reminders, post-appointment follow-up
Engagement
  • ·Phase 1: 11 weeks
  • ·Currently: continuous (month 7)
Outcomes (month 6 snapshot)
  • ·Patient response rate (billing): +34%
  • ·Front-desk hours reclaimed: ~280/month
  • ·Bilingual coverage (EN/ES): added without staff hire
  • ·Compliance: HIPAA-aligned audit log, zero violations

Federal defense consulting (anonymised — clearance-gated)

IndustryDefense supply chain · 8(a)/HUBZone advisory
ScopeFPCC compliance prep, CMMC documentation, contract response
Credentials carried
  • ·Master's in Cybersecurity (Mani)
  • ·Lightship Security background — Common Criteria / FIPS 140 testing
  • ·Hari Mankani — 7+ years security operations, Master's in Cybersecurity
Why this matters to Fortes
  • ·Building agent systems for regulated environments requires understanding what regulators will ask for next year, not just today.
  • ·KHDA's data residency requirements share a common ancestor with FedRAMP and ITAR controls. We've already built that scaffolding.

Real estate brokerage (US)

IndustryReal estate · multi-MLS coverage
ScopeLead qualification, listing intake, agent-handoff orchestration
Engagement
  • ·Phase 1: 9 weeks
  • ·Currently: month 14
Relevance to Fortes
  • ·Same shape of problem: high-velocity inbound (parent/buyer leads), multilingual requirement, regulated documentation, human-in-loop approval on every parent-facing communication.
  • ·Pattern reusability to Fortes: 73% of skill library transferred.
Section 5 · What 3 AM looks like

Incident response

Recent incident: Apr 12 2026 · 03:47 GSTSeverity: P3Customer impact: noneDuration to resolution: 27 minutes
03:47:14
Heartbeat anomaly detected · client #4 · Watson skill v2.3
Auto-pause triggered on 1 specific tone variant
03:47:18
Slack alert routed to on-call Architect (Mani)
PagerDuty escalation: not triggered (P3, auto-pause held)
03:51:00
Architect acknowledged · investigation started
Initial read: rare combination of locale + character encoding breaking the bilingual handoff in Watson v2.3
04:02:00
Root cause confirmed · 0.4% of conversations affected
Of those, 100% gracefully degraded to human handoff (no broken conversations, just slower ones)
04:11:00
Fix patched in skill registry · v2.3.1 deployed
Re-enabled after 7-minute shadow test
04:14:00
Incident closed · post-mortem auto-drafted
NEXT BUSINESS DAY
Post-mortem reviewed with client IT lead
Learnings added to skill registry, 2 new test cases catalogued.

This is the 19th incident across 6 active deployments in 90 days. The other 18 followed the same pattern: detected before customer impact, resolved before business hours. The on-call rotation covers all six clients; the engineer who responded at 03:47 GST also responds to your incidents at 03:47 GST. There is no separate team.

Section 6 · What's actually included

What's in an engagement that isn't in the prototype

Included in Phase 1
Architecture
  • ◆Locked architecture document (signed off by your IT Lead and ours)
  • ◆Hermes runtime configuration tuned for UAE data residency
  • ◆Integration map across your 14 systems (SIMS, CRM, CPOMS, etc.)
  • ◆RBAC matrix definition signed by Ajay or designate
  • ◆Audit log retention policy mapped to KHDA's 7-year requirement
  • ◆Incident response runbook customised to Fortes (DSL escalation, BoD-impacting events)
Skill library
  • ◆Production skills for Phase 1 scope (count finalised at discovery)
  • ◆Skill registry write access (so you can audit any prompt)
  • ◆Skill version history retained for 7 years
  • ◆Shadow-mode testing access for any new skill before go-live
Operations
  • ◆99.9% uptime SLA with credits clause
  • ◆24/7 on-call rotation across all 6 Vertical Systems clients
  • ◆Weekly skill review every Friday 14:00 GST (Atyab + your nominated team)
  • ◆Monthly Intel review (the /intel screen) with your SLT
  • ◆Quarterly architecture review with Ajay or designate
Knowledge transfer
  • ◆Full skill module documentation
  • ◆Architect-to-Architect briefing for your IT team (optional)
  • ◆Read-only access to Vertical Systems' skill library (excluding other clients' specific tone tunings — only the generic patterns)
  • ◆Exit playbook: if engagement ends, you keep all skills, all data, all configurations. The Hermes runtime is open source.
Engagement structure
  • ◆Fixed-fee SOW for each phase, not hourly billing
  • ◆Milestone-based delivery — not paid against time spent
  • ◆Phase 1 and subsequent phases scoped and quoted separately
  • ◆Pricing and detailed scope finalised in a discovery conversation once your tech stack and integration requirements are confirmed
Outside Phase 1 scope · scoped and quoted separately
  • ◇Phase 2 and Phase 3 builds
  • ◇Custom skills beyond Phase 1 scope
  • ◇Direct staff training beyond the named team
  • ◇On-site presence beyond 2 days during build
Section 7 · Accountability

The team — small, signature view

Architect
Mani Kanasani · Founder & CEO, Vertical Systems
  • ·Master's, Cybersecurity · ex-Lightship Security (Common Criteria / FIPS 140)
  • ·Author of OpenClaw masterclass (111K views, Mar 2026)
  • ·Available: weekly skill review + escalations + quarterly architecture
Manager
[TBD — assigned at Phase 1 kickoff]
  • ·Will run weekly skill reviews, monthly Intel review, daily ops health monitoring
Architect-in-reserve
Hari Mankani · Co-founder, Vertical Systems
  • ·Master's, Cybersecurity · 7+ years security operations
  • ·Available for: compliance review, security architecture, CMMC/KHDA documentation guidance
Workers
Sadiq · Ganesh · Vyom — production build team
  • ·Currently on: Reverra Health 6-agent system (delivering Jun 14)
  • ·Rolling onto Fortes from Jun 17

Vertical Systems is a 10-person agency. The same Architect (Mani) signs off on every client's architecture decisions. The same on-call rotation covers every client's 3 AM. This concentration is deliberate — it's what keeps the skill library coherent and the incident response fast.

Section 8 · The disarming move

Exit playbook

If at any point Fortes wishes to end the engagement, the following applies.

01
You own everything. All skills, all configurations, all audit logs, all data flows are exported to your control within 7 business days.
02
The runtime is open source. Hermes is freely available. Your IT team can continue running the platform without Vertical Systems if you prefer.
03
Knowledge transfer is included. Up to 40 hours of Architect-to-Architect transfer for your IT team, included in the engagement at no additional cost.
04
There is no lock-in clause. Phase 2 and 3 are separate SOWs that you sign only if Phase 1 met your bar.
05
30-day pilot exit. Within the first 30 days of Phase 1, if the engagement isn't meeting your expectations, you pay only for work completed and we hand over what we've built. No claw-back, no penalty.

This page exists because the right way to choose a partner is to be told upfront how they handle the worst-case scenario.

Section 9 · Engineering policy on pricing

How pricing works

You won't find prices on this page or anywhere in the prototype. That's deliberate.

Pricing for an agent platform depends on variables that aren't visible from outside your organisation:

  • ◆Your existing tech stack — how much we integrate with vs replace
  • ◆Scope split — single school vs all three brands vs full Holdings rollout
  • ◆Integration depth — your SIMS, your CRM during the Zoho migration, CPOMS, PowerBI, the 11 other systems
  • ◆Operational ownership — whether your team takes the platform over after Phase 1 or we stay on
  • ◆Regulatory work — KHDA-specific compliance scaffolding vs generic

Any number quoted before discovery would be wrong, in one direction or the other, by 20–40%. We don't do that to our clients.

After a discovery conversation with you, your IT team, and a finance representative, we send a clear range the same day with each variable spelled out. You receive a fixed-fee SOW once scope is locked. Phases 2 and 3 are quoted separately at their respective kickoff points, never bundled into a single commitment.

Engagement structure: fixed-fee, milestone-based, scoped separately per phase, no hourly billing for build work.

We build agent systems for organisations that take their work seriously. Fortes Holdings has been doing that for 45 years. We hope this is the start of a long engagement, but it has to earn its way there one phase at a time.

— Mani Kanasani, Founder