SMB AI readiness
SMB AI Readiness — Discovery & Adoption Roadmap
A free, self-service intake that captures deep operational context from a small business and generates a personalized, phased AI adoption roadmap. No consultants, no guesswork, and full transparency about what the system knows and doesn’t.
intake link → six-step wizard → four agent passes → confidence label → admin gate → the SMB’s roadmap
01
The problem it exists to solve
Don’t know where to start
AI tools are everywhere. SMB owners hear the noise but have no framework for deciding which workflows to automate first, which tools fit, or what’s realistic at their size. Analysis paralysis.
Generic advice doesn’t fit
Blog posts say “just use AI for customer support,” but they don’t know a three-person team, its compliance load, its tool stack, or the unwritten knowledge living in one lead engineer’s head. One-size-fits-all fails every time.
Consulting is too expensive
Traditional engagements start at $10K–$50K and take weeks. For an SMB with a $500-a-month tech budget that’s not viable. They need a starting point credible enough to act on, not a six-figure deck.
The bet: capture enough operational context — not just “what industry are you in?” but actual workflows, pain points, constraints, decision authority, and unwritten knowledge — and an agent can synthesize a roadmap specific enough to be genuinely useful. The key is asking the right questions and being honest about what gaps remain.
02
Dual-purpose intake
One form feeds two paths.
Channel A — full pipeline
AESOP Transformation OS intake
For organizations entering the full six-stage pipeline — Discover → Design → Build → Evaluate → Repair → Track. The form populates the organization profile, department structure, project portfolio, and operational context.
Channel B — standalone
TBG lead generation
The Bradbury Group shares a link. An SMB fills the form in under ten minutes. Behind the scenes AESOP generates a complete adoption roadmap; TBG reviews and approves it, and the SMB receives a credible, personalized AI strategy. Low-friction lead gen with real value delivered.
03
The intake wizard
Context density over question volume. Each step asks for narrative, not just checkboxes. Steps 1–4 are required — the minimum context a meaningful roadmap needs. Steps 5–6 are optional; the confidence score quantifies how much the system is guessing versus grounding in real data.
Step 1 required
Business Snapshot
Name, industry, size, revenue, acquisition channels, delivery model, health self-rating, what they wish they had more time for.
Step 2 required
Goals & Ambition
Top 12-month goal, definition of success, vision for AI, 3-year vision, biggest obstacle, AI tools already tried.
Step 3 required
How Work Gets Done
Team roles, three detailed workflows (name + steps + pain points), tool stack, integration level, unwritten knowledge, gut-feel decisions.
Step 4 required
Constraints & Guardrails
Budget ceiling + approval speed, compliance, non-negotiables, automation comfort, decision authority, who can veto, dealbreakers.
Step 5 optional
Departments & Team
Department names, leaders, goals; team member names, roles, emails, departments.
Step 6 optional
Current Projects
Project names, descriptions, statuses, archetypes.
04
Generation pipeline
When the SMB clicks Generate My Roadmap, four agent passes run asynchronously — about 30–60 seconds behind a loading screen.
Capture
Create the organization row and write four org_context documents from the form. Create departments, team members, and backlog projects if they were provided.
Confidence
Compute the heuristic confidence score — each step weighted, each field scored for presence. Outputs 0–100, a level, and specific gaps with impact statements.
Readiness gate
Does an org readiness assessment already exist? No → run the Org Readiness Agent (identifies gatekeepers, key-person risk, prioritized action items). Yes → skip straight to prioritization.
Prioritize
Build the strategic prioritization — a prioritized project list scored by effort, ROI, and strategic alignment, with quick wins flagged.
Roadmap agent
Claude Sonnet 4.5 at temperature 0.2 synthesizes all context into a phased roadmap: projects, milestones, capability targets, risk mitigations, and exclusion lists. Output is validated; up to two retries on error.
validation loopValidation errors send the roadmap back to the agent; on pass it proceeds to post-processing.
Post-process
Attach the source manifest, persist the roadmap, and notify an admin. The deliverable sits in pending_review.
Human gate
A TBG admin approves before the SMB sees anything. Approval makes the results page visible to the SMB.
05
Confidence — every roadmap carries an honesty label
Most AI-generated strategy ships as if it were certain. This one does the opposite: a confidence score quantifies exactly how much the system is extrapolating versus grounding in what you provided.
High confidence
75–100
Well-grounded. Most fields across all steps are filled. Strategic choices are data-supported. Ready to act on.
Medium confidence
50–74
Directionally sound. Key context is present but gaps exist. Project sequencing may shift as missing information is filled in.
Low confidence
0–49
A conversation starter. Significant context is missing. A starting point for discussion, not a finalized plan.
Step weighting
How Work Gets Done
25% · 14 fields
Business Snapshot
20% · 11 fields
Goals & Ambition
20% · 7 fields
Constraints
15% · 8 fields
Two confidence layers. The heuristic score quantifies form completeness. Separately, the roadmap agent flags LLM-detected gaps — the places where it knows it is making an assumption that needs validating.
06
Admin review — quality gated before delivery
Roadmaps are held in pending_review until a TBG admin approves them. Every roadmap that reaches a client has been vetted by a human.
Executive summary
A narrative synthesis of where the business is, where it could go with AI, and the strategic themes that emerged from its context.
Confidence score card
The 0–100 score with level, plus specific gaps — both heuristic and LLM-detected.
Phased implementation plan
3–5 phases with projects, milestones, capability targets, and risk mitigations. Projects tagged existing or AI-suggested.
Strategic themes
Cross-cutting themes like “Knowledge Capture First” and “Compliance-by-Design.”
Account CTA
A prompt to create a free AESOP account for deeper customization and the full pipeline.
07
What the intake touches
| Database table | Role |
|---|
organizations | Org record per submission — org_type: smb, name, slug. |
org_context | Four structured docs from wizard data — company_profile, strategic_goals, smb_operations, smb_constraints. |
smb_intake_submissions | Tracks each submission — token, status, confidence, review_status. |
departments / team_members / projects | Optional structure from steps 5–6 — roster, project portfolio with archetypes. |
org_readiness_assessments | Auto-generated readiness eval — verdict, score, dimensions, gatekeepers, key-person risks. |
org_readiness_action_items | Prioritized gaps — title, severity, dimension, owner_role, becomes_true_when. |
adoption_roadmaps | The deliverable — phases (JSONB), strategic_themes, source_manifest, confidence. |
Full provenance. Every roadmap stores a source_manifest JSONB field recording every input used at generation time. If someone questions a recommendation, you can trace it to the exact input that informed it.