Know what to automate before you build.Score every workflow before you spend on one.
Our view: when an AI build disappoints, the model is rarely the problem. The problem is that nobody mapped which workflows were safe to automate before the building started.
“Your AI Reality Check — before you spend a quarter on the wrong build.”
Hybrid
High impact · High risk
AI drafts. Human approves.
e.g. Client quotes: AI drafts the number, a human checks and approves it.
Impact-Risk Matrix — your audit deliverable
Where AI Builds Go Wrong
Our argument: the trouble usually starts before the build, not inside it.
Skip the mapping step and you build twice. A team goes straight from “AI sounds useful” to building. Nobody checks which workflows are safe to run without a human watching, and which ones will cause problems the first time they do. Then a wrong quote reaches a client, and you pay to tear the automation out and rebuild it with a human checkpoint. The audit scores each workflow for Impact and Risk first, so that call gets made before anyone writes code.
A workflow that processes customer-visible output, touches pricing, or has significant edge-case variance is not a candidate for full automation — no matter how formulaic it feels on a Tuesday morning. But it might be a good Hybrid: AI drafts the work, a person checks it before it goes out. That distinction decides what you build. Get it wrong and you pay to build, then pay again to rebuild.
You already know which workflows feel broken. The expensive mistake is assuming you also know which ones are safe to automate.
The AI Workflow Audit exists to close that gap before you commit engineering time. It is the step before the build.
2 hrs
One Workshop Audit session
Founder plus 1-2 people who run the workflows. No prep or homework.
3-5 days
From session to your written memo
Business days, Workshop format. The Interview Audit takes 5-7 after the last interview.
3 tags
Automate, Hybrid or Manual for every workflow
Plus a written list of what not to build, so engineering hours go where they pay back.
Two Axes. One Decision.
Every workflow in your business gets scored on Impact and Risk. The intersection tells you exactly what to build.
How the scoring works
How much time, money, or quality does this workflow touch? High-volume, revenue-adjacent workflows score higher. Inbox management and low-throughput admin tasks score lower — even if they feel painful.
What is the cost of an automated error? Regulatory exposure, customer-visible mistakes, financial errors, and reputational damage all drive risk up. Internal-only, easily-reversible workflows score lower.
The three decision tags
Hybrid is not a compromise. AI drafts, a person checks and approves. In our view that is often the better build: it fails where someone can see it, and it does not need perfect edge-case training data. We treat full automation as the exception, not the default.
Hybrid
High impact · High risk
AI drafts. Human approves.
e.g. Client quotes: AI drafts the number, a human checks and approves it.
Select a quadrant to read its call
Two Formats. Same Deliverable.
The audit scales to your company size. Both formats produce the same decision framework.
Workshop Audit
Best for: 1-50 employees
For founders, small business owners, and early-stage ops teams who need clarity fast.
- Single 2-hour Zoom session
- Founder + 1-2 operators who run the workflows
- No prep or homework required
- Delivered in 3-5 business days
You walk away with:
Workflow map, scored Impact-Risk matrix, decision tags, and 3-5 prioritized build recommendations.
Interview Audit
Best for: 50-500 employees
For mid-size companies, multiple departments, and organizations with existing AI spend to evaluate.
- Structured 1:1 interviews with 5-8 department leads
- 30-45 min each — no prep required
- Ops, sales, finance, CS, and leadership covered
- Delivered in 5-7 business days after final interview
Plus everything from Workshop, and:
Cross-department workflow conflicts and dependencies identified. Vendor build-vs-buy recommendations included.
The 3 Patterns That Sink AI Projects
Three mistakes the Impact-Risk scoring is built to catch before anything gets built.
They put high-risk workflows in "Automate"
Quote generation, contract drafting, anything customer-facing — these feel like prime automation candidates because they are formulaic. But one pricing error in an auto-generated quote or one wrong clause in a contract creates a customer escalation, a legal issue, or a margin leak. The audit catches this and re-tags them Hybrid before a single line of code is written.
They overrate workflows they personally hate
The founder's inbox is the classic example. It is touched all day — it feels like the highest-impact thing to automate. But the actual throughput value is low; the time savings do not translate to revenue. The audit separates "painful" from "high-impact." They are not the same thing.
They assume full automation is the goal
A Hybrid build has AI do the drafting and a person approve the result. It fails where someone can see it, and does not need perfect edge-case training data. The scoring asks whether a workflow needs a human checkpoint before it asks whether AI can run it alone. Often it does.
How the Scoring Handles a Quote Workflow
A made-up example built to show the method. It is not a client engagement and reports no client results.
Example: B2B services firm — quote generation
Impact
4/5
Risk
4/5
Verdict
HybridThe Request
Say a firm wants quote generation fully automated end to end: populate pricing, write the scope narrative, apply standard terms, and send it to the client. No human in the loop after the trigger.
How It Scores
High impact, because quotes drive revenue and get written every week. High risk, because a margin error, a scope gap or the wrong positioning lands in front of the client. Full automation on a workflow like this is a do-not-build: one bad quote can cost more than the time the automation saves.
The Recommendation
Hybrid. AI pulls the CRM data, picks the pricing tier, drafts the scope narrative and fills in the quote template. A person checks the margin, adjusts the positioning and approves it. The human step is a review, not a rewrite.
What to Measure After
Time per quote before and after, how often the reviewer changes the draft, and win rate on reviewed quotes. A real audit sets these baselines up front so the build can be judged on numbers, not impressions.
What You Get at the End
A consulting-firm-grade decision framework — the kind you can walk into a board meeting with.
Delivered as a consulting-firm-grade presentation deck — the kind you can walk into a board meeting or investor conversation with.
On10 Solutions
Before You Book
The questions we hear most before a prospect books their first audit.
How long does the whole process take?
Who from our team needs to be involved?
What if we have already started building?
Do you implement after the audit?
What if the audit says we should not automate anything?
What industries do you work with?
Why a fixed fee instead of hourly billing?
Your competitors are already spending on AI.
The question is not whether to automate. It is which workflows. Skip that step and you can pay to build the wrong thing, then pay again to rebuild it.
Get the clarity you need before you spend a dollar on implementation.
Workshop memo in 3-5 business days/No commitment to implement/Fixed fee, fixed scope