Meet the LeadAI Coaches

Seven AI coaches. Each tuned to the role that will use it on a Tuesday morning.

Generic AI coaching answers like a SaaS marketing blog. LeadAI’s coaches answer like the senior practitioner sitting next to you — who has seen the artefact, knows the regulator, and remembers what your last sprint looked like.

Book Enterprise Demo Start as an individual

The seven coaches

Read a real exchange with each coach. No login required.

BA · NEXUS

Maya

Coach for Business Analyst

Practises requirements that survive audit — not just review.

BRD / FRD / RTM drafting under AI governance
Gap analyses, UAT plans, traceability matrices
AI fair-lending / fairness sections (FinServ, regulated SaaS)

You

I have a BRD for an AI credit-scoring feature. Compliance bounced it back. Where do I start?

Maya · NEXUS

Three sections are almost always missing. (1) Adverse-action notice template tied to model output thresholds. (2) Protected-class slicing plan with sample-size minima. (3) Audit-trail spec: model version, input snapshot, score, override decision, who clicked. Add those three and the BRD usually clears ECOA / Reg B review on the first re-submit. I'll redline yours if you paste it.

PM · APEX

Jordan

Coach for Project Manager

Treats AI as an input to risk reporting, not a buzzword.

AI-aware status, RAID, change control, lessons learned
Stakeholder briefs that translate model output to managerial language
No-code AI agent rollouts with kill-switch runbooks

You

Copilot just flagged 14 projects as at-risk overnight. My steering committee starts in 40 minutes. How do I open?

Jordan · APEX

Open with the framing — not the data. 'These are model-flagged, not PM-confirmed.' Then walk three confirmed risks with named owners and a date. Defer the other 11 to a triage list with a re-verify SLA. The line that protects credibility is tying model confidence to managerial confidence. If you say it before they ask, you keep the room.

SM · SAGE

Alex

Coach for Scrum Master

Surfaces AI-related team friction without becoming a vendor pitch.

Sprint plans, DoR / DoD updated for AI-augmented work
Retros that name AI tool drift without naming a vendor
PI planning when the AI is the integration risk

You

Two engineers want to use Cursor; one wants Copilot; the EM has not picked. Retro is in three days. What is the right shape?

Alex · SAGE

Don't make this a tool fight. Frame the retro question as: 'Where in our delivery loop did AI help, where did it hurt, where did we waste time disagreeing about which tool to use?' Capture the third bucket separately. That becomes the EM's input for the picking decision, not the team's.

PO · VECTOR

Donna

Coach for Product Owner

Writes acceptance criteria where "done" is no longer obvious.

Stories, AC, epics, prioritisation, release notes
AC for AI-augmented features (confidence, fallback, override)
Stakeholder rewrites: turn "feature meets stakeholder needs" into measurable, audit-defensible AC

You

I wrote: "AI helps the user write a better email." Reviewers said it is too vague. Rewrite it for me.

Donna · VECTOR

Three criteria. (1) The user can accept, edit, or reject any AI suggestion before send — measured by override rate per cohort. (2) The model returns a confidence band; below 0.7 we hide the suggestion entirely. (3) Every accepted suggestion is logged with prompt + model version + final text. That is auditable, measurable, and shippable.

RM · ATLAS

Ravi

Coach for Release Manager

Builds runbooks where kill-switch criteria matter as much as the deploy step.

Release plans, runbooks, PRR, CAB, canary, DR
Staged canary with explicit kill-switch metric thresholds
Model-version rollback ownership and DR scenarios

You

Engineering wants a single-stage 5% canary for an AI fraud-detection model. CAB is tomorrow. Will it pass?

Ravi · ATLAS

No. Single-stage 5% misses the SOC-2 control review window and the kill-switch wiring is implicit. Propose 1% / 5% / 25% staged with explicit rollback gates: false-positive rate above 0.4%, latency p95 above 800ms, or any single regulator-flagged transaction. Wire the kill-switch to those three signals before CAB.

PdM · PRISM

Priya

Coach for Product Manager

Writes PRDs for products that have an AI feature, in industries that audit.

PRDs, vision, OKRs, GTM, roadmap, board updates
Compliance posture for AI features pre-launch
Board-grade narrative when the AI is the differentiator

You

My board wants to know if our AI roadmap is realistic. I have point estimates. Help.

Priya · PRISM

Translate every estimate into a band, then a confidence. Q1 launch becomes 'Q1 (80% confidence) / late Q1 (60%) / early Q2 (40%)'. Show two scenarios — accelerated and conservative — with one input the board can move (head-count, scope, or risk tolerance). Boards trust ranges with named inputs. They distrust point estimates that slip.

EM · SAGE

Alex

Coach for Engineering Manager

Writes the tech-debt narrative finance will not kill.

Capacity, 1:1 frameworks, ADRs, RCAs, hiring, tech debt
Tech-debt narratives that finance approves
1:1 prompts that surface flight-risk early

You

Finance keeps killing my tech-debt asks. How do I write the next one differently?

Alex · SAGE

Stop calling it tech debt. Reframe as 'unrealised feature velocity.' Quantify three things finance understands: lost velocity per sprint (story points or hours), incident hours per quarter attributable, and the dollar cost of the next external auditor finding. Three numbers, one paragraph. The ask survives because finance can defend it to the CFO with your own data.

Cross-role governance agent

SENTINEL

Cross-role governance agent. Watches every artefact for regulatory, decision, security, and ethical risk — in real time.

SOC 2, HIPAA, GDPR, FDA SaMD, ECOA / Reg B, IEC 62443, EU AI Act
Flags kill-switch absences and missing rollback SLAs
PHI / PII handling, audit-trail preservation, model-versioning
Algorithmic-impact assessments, protected-class testing

Read the SENTINEL playbook

For Teams

Roll the coaches out to a cohort

6-week pilot, 15-40 seats, signed Readiness PDF for the steering committee. From $249 per learner.

See the enterprise pilot

For Yourself

Pick your role coach and start free

Pick my role